Amazon’s latest film drama isn’t a movie that it’s producing or streaming, but rather the situation surrounding a project it has dropped.

Amazon MGM Studios has backed away from “Artificial,” a nearly finished film about OpenAI CEO Sam Altman. The studio said last week that the film would “be better served if it were released by a different studio,” according to reports in Puck, Variety, and elsewhere.

The film, directed by Luca Guadagnino, stars Andrew Garfield as Altman and focuses on the brief period when Altman was fired from his position at OpenAI in 2023 and then rehired, according to Variety. The film has been referred to as “‘The Social Network,’ but for the AI era,” in a nod to the 2010 film about Facebook.

The New York Times reported that Amazon MGM had spent around $40 million on the project and had tested it in four markets. The decision to drop the film, planned for a 2027 release, “shocked the filmmakers,” the Times said.

Amazon announced a $50 billion investment and strategic partnership with OpenAI in February. The ChatGPT maker and Amazon Web Services deepened their technical ties, expanding an existing $38 billion multi-year agreement by $100 billion over eight years, with OpenAI planning to run more of its AI workloads on AWS.

Amazon said it is “working closely with the filmmaking team to find the film a new home.” The Hollywood Reporter said Netflix and Focus Features have both passed on “Artificial.”

Puck, which first reported the news, said the finished film struck Mike Hopkins, the head of Prime Video and Amazon MGM Studios, as darker in tone than the script Amazon had originally bought, and that Hopkins made the decision to drop it after watching a cut. Amazon’s public statement gave no reason.

Guadagnino, who directed the 2017 Oscar winner “Call Me by Your Name,” previously worked with Amazon MGM on “After the Hunt” and “Challengers.”

“We have the utmost respect and admiration for Luca Guadagnino as an award-winning filmmaker — not to mention a longstanding relationship that we hope to continue,” an Amazon spokesperson said in a statement.

Other “Artificial” cast members include Monica Barbaro as former OpenAI CTO Mira Murati, Yura Borisov as former OpenAI chief scientist Ilya Sutskever and Ike Barinholtz as Elon Musk.

Carrying bullhorns and signs depicting Amazon executives as war criminals, about two dozen people protested outside the Amazon Spheres in Seattle on Thursday evening, calling on the company to stop providing technology to Israel for what they described as genocide in Gaza.

The protesters said they were trying to disrupt what they believed to be a gathering of Amazon executives, state and local leaders, U.S. State Department officials and Australian government representatives on an upper floor of the Spheres, on the eve of the World Cup match between the U.S. and Australia.

Contacted Friday, Amazon described the gathering differently. The company said the event underway during the protest was for members of Seattle’s business and sports communities, Australian parliamentarians, and Amazon employees celebrating the World Cup. A separate meeting concluded before the protests began, the company said, without specifying who attended that meeting.

“We respect individuals’ rights to engage in peaceful public demonstrations,” said Montana MacLachlan, Amazon spokesperson, in response to GeekWire’s inquiry. The company, she added, is “committed to being a responsible corporate citizen in the Puget Sound region, Washington state, and every community we serve.”

The protest group, which goes by the name Amazon Worker Intifada, described the protest as part of an effort to escalate pressure on the company’s leaders over the issues. An affiliated group, No Azure for Apartheid, has been protesting Microsoft for more than a year over its work for Israel.

The protesters object to Amazon’s work with Israel, including Project Nimbus, a $1.2 billion contract that Amazon and Google won in 2021 to provide cloud and AI services to the Israeli government, including the Israeli military and weapons suppliers, according to leaked contract and procurement documents.

The protesters marched to the Spheres shortly before 6 p.m. Thursday, walking in a circle outside the glass-domed buildings with signs, drums, balloons, noisemakers and Palestinian flags, engaging in call-and-repeat chants such as, “Say it loud and say it clear — Amazon’s a war profiteer.” 

Amazon workers and soccer fans walked by on the sidewalk, some stopping to take in the scene. Small groups of people in business attire walked through the protest to the Spheres entrance.

A banner at the edge of the space read “Amazon War Criminals Meeting Here.” Another depicted Amazon CEO Andy Jassy and AWS CEO Matt Garman, with blood on their hands, embracing what appeared to be a bomb. “We See Your Crimes,” it read.

Members of what appeared to be a wedding party, including a woman in a white bridal dress and a man in a suit, emerged at one point from one of the restaurants at the base of the Spheres and tried unsuccessfully to persuade the protesters to stop or move elsewhere.

In a press release after the protest, the group said its demonstration forced Amazon to reroute attendees, and that an arriving Australian delegation had to use a different entrance to get around the protesters. The group also said an event attendee grabbed and shoved a protester’s camera.

After protesting for an hour at entrances on both ends of the courtyard between the Spheres and Amazon’s Day One tower, the group moved to the Lenora Street side of the Spheres, where they released two helium balloons on strings with loud noisemakers attached, attempting to position the noisemakers outside the windows where an event could be seen taking place inside. 

One of the leaders of the protest Thursday was Ahmed Shahrour, a Palestinian software engineer in Amazon’s Whole Foods division in Seattle who was fired in October over internal Slack posts criticizing the company’s ties to Israel.

Amazon said at the time that he violated multiple company policies, alleging that he “misused company resources, including by posting numerous non-work-related messages pertaining to the Israel-Palestine conflict.”

Shahrour called it “a blatant act of retaliation designed to silence dissent from Palestinian voices within Amazon and shield Amazon’s collaboration in the genocide from internal scrutiny.”

On Friday, Amazon spokesperson MacLachlan said of that incident, “We don’t tolerate discrimination, harassment, or threatening behavior or language of any kind in our workplace, and when any conduct of that nature is reported, we investigate it and take appropriate action based on our findings.”

No Azure for Apartheid, which includes current and former Microsoft workers, has staged repeated protests of its own, similarly calling on Microsoft to cut ties with Israel.

They set up an encampment on the Redmond campus last year, where 20 people were arrested for trespassing, and later occupied the office of Microsoft President Brad Smith. Microsoft has fired several employees over various protests and activities, citing violations of company policies.

After a Guardian investigation revealed that an Israeli military unit had used Microsoft’s Azure cloud to store millions of intercepted Palestinian phone calls, the company cut off the unit’s access and opened a review that recently led the company to announce that it would tighten its human-rights controls on its work with national security agencies.

An employee group filed a civil rights complaint against Amazon with the City of Seattle on Thursday on behalf of three engineers who allege that the company is wrongly investigating them for testifying before the Seattle City Council in favor of regulating data centers. 

The complaint, filed by Amazon Employees for Climate Justice (AECJ), invokes an unusual Seattle law that bars employers from discriminating against workers based on political ideology. 

Amazon acknowledged the investigations but characterized them differently, citing its policy against employees speaking publicly as representatives of the company without first going through specific procedures. A spokesperson described this as the focus of the internal inquiry, noting that employees are free to discuss working conditions in their individual capacity.

The three engineers — Patrick Schloesser, Darius Irani, and Liesl Wigand — testified June 3 before city council subcommittees in support of regulating data centers. Each opened by noting they were legally protected from retaliation for speaking out.

A week later, Amazon’s Employee Relations team called them into separate meetings and told them they were under a disciplinary investigation, according to the complaint, a copy of which was reviewed by GeekWire.

“After publicly affirming our right to speak freely, Amazon privately interrogated me, asking me the same questions over and over to try to get me to admit to doing something wrong and made me feel like I committed a crime,” Irani said in a statement released by the group. 

The complaint says the engineers were told the investigation could lead to termination. 

Amazon denied that it threatened to fire the engineers or told them they were at risk of termination, saying the reference came up in response to a direct question and was taken out of context in AECJ’s characterization of what happened.

After reviewing the testimony, “it became clear that they may have been speaking in their capacity as Amazonians and not as private citizens,” said Amazon spokesperson Margaret Callahan in a statement. “We believe it’s important to apply our policies consistently so, just as we would with anyone else, we’re investigating whether there was a violation of our policies and may or may not take action based on what we find.”

She added, “It’s important to note that we don’t tolerate retaliatory behavior.”

Under the city’s Fair Employment Practices Ordinance, the Seattle Office for Civil Rights will investigate the complaint and determine whether there is reasonable cause to support the allegations. Remedies can include reinstatement, back pay, and financial damages.

Following testimony by more than 50 people, including members of AECJ, the full Seattle City Council voted unanimously on June 9 to impose a one-year emergency moratorium on new large data centers inside the city limits.

Amazon Web Services is announcing a new set of AI agents for businesses, developers, and individual users, capable of everything from fixing security vulnerabilities to triaging email.

The agents, unveiled at the AWS Summit in New York, reflect an attempt to maximize autonomy while ultimately keeping humans in control of how much the AI does on its own.

It’s part of a broader industry push into agents, with Google, Microsoft, Anthropic, OpenAI and others developing AI that can do more work and increasingly complete tasks on their own.

A new security agent, dubbed AWS Continuum, starts in a supervised “learn mode” and earns the right to act alone only as customers grant it permission, category by category.

The Amazon Quick AI assistant will now let users build their own background agents in plain language to handle tasks like following up on stalled business deals or flagging regulatory changes. 

Amazon gave Quick a redesigned activity feed that triages email, messages, and calendar items into one prioritized view; new links to services including Adobe, Figma, Snowflake, and WhatsApp; and the ability to tap multiple connected services to answer a single question.

On the developer side, AWS is also pushing its coding agents to take on more of the grunt work, checking and testing new code before it ships and cleaning up old code, while leaving the final decision to merge or deploy in the hands of humans. A new iPhone app for Kiro, the company’s AI coding assistant, will let developers start and monitor that work from their phones. 

Deepak Singh, the AWS VP who leads the Kiro team, said the overarching idea is to take the background work AI has piled onto people — reviewing code, triaging security findings, keeping software current — and let agents handle it with minimal human intervention. 

The faster AI writes code and surfaces problems, he said, the more there is for humans to review, test, and maintain: “Those are all good problems to have, but they are real problems.” 

AWS also expanded AgentCore, its platform for building agents, and introduced AWS Context, a service that organizes a company’s data so agents can reason over it. 

Announcing the new Continuum security agent, AWS cited the rise of powerful AI models — most notably Anthropic’s Claude Mythos — that can now find software flaws and chain them into serious attacks faster than any human team can respond. 

Amazon made headlines for raising concerns about those same models, reportedly warning Trump administration officials about security risks in Anthropic’s most advanced AI, before a government order forced the lab to take its two newest models offline. 

Continuum is starting with code vulnerabilities, and AWS says it will expand to other aspects of security in the future. It works through issues the way a human team would, if given the time: triaging the findings, testing whether a vulnerability is exploitable, and then proposing a fix, with an estimate of what else the change might break.

In categories where the customer has granted the agent autonomy, Continuum can apply the fix itself, feeding the change into an existing deployment pipeline. 

Neha Rungta, AWS director of applied science, said in an interview that this kind of speed is necessary given the acceleration of the threats. AI can now chain minor flaws together, she said, combining two medium-severity findings and a low one into something critical.

“That was something that would have taken a lot of effort, expertise, and determination for an attacker to get through — so the floor has been lowered,” said Rungta, who led the work on Continuum. “The goal is to raise that floor up again.”

[Editor’s Note: Agents of Transformation is an independent GeekWire series, underwritten by Accenture, exploring the adoption and impact of AI and agents. See coverage of our related event.]

Amazon is legendary for its process of “working backwards.” Start with a customer problem, imagine a future in which it’s solved, draft a press release and FAQs as if it had already happened, obsess over the document until it’s just right, and then go make it a reality.

But sometime last year, it dawned on Swami Sivasubramanian, Amazon Web Services VP of agentic AI, that new coding tools had suddenly made it easier for his teams to develop a demo — actual working software — than to write the classic six-page Amazon “PRFAQ.”

So they began starting with the prototype instead.

If something is “a low-risk bet where we just want to prove our intuition, then I actually say, let’s first go build the demo, and then iterate,” Sivasubramanian said in an interview last week, in advance of his keynote address Wednesday at the AWS New York Summit.

It’s an illustration of how agentic tools are reshaping even the most entrenched workplace practices and traditions. But it’s just one of the ways that the AWS agentic AI team is departing from the company’s established norms, and in some ways returning to its roots. 

Inside Amazon, CEO Andy Jassy says he wants the company to run like the world’s largest startup. Sivasubramanian’s division may be the closest thing to what that looks like in practice. 

Back to two pizzas

The AWS agentic AI division is organized into dozens of small teams, many of them just large enough to feed with two pizzas. That was the organizing principle that Amazon pioneered in its early days and that much of the company outgrew as it scaled to 1.5 million employees. 

When Matt Garman, the CEO of AWS, carved out agentic AI as its own division last year, Sivasubramanian went with small teams on purpose. It matches the new reality of the AI era: projects that once required 30 to 40 people, he said, can now be done by teams of six to eight.

Case in point: the Amazon Quick desktop app, which connects to a user’s email, calendar, Slack, documents, and other apps in a single workspace, and uses AI to search across them, answer questions, and perform tasks. It’s Amazon’s entry in a market where Anthropic, Microsoft, Google, and OpenAI have captured much of the attention.

It traces its roots to late January of this year, when Sivasubramanian said it became clear to him and others on the team that the underlying models had gotten good enough that the main missing ingredient was connecting them to the systems where people actually work. 

He pulled together a team of about six engineers to build it. Six weeks later, 200 people inside Amazon were using it. Ten weeks in, it was up to 10,000 internally. The team circled back to write the PRFAQ after the product was already in beta, to help refine their approach to the external launch. They shipped on April 28, three months after they got started.

Under the old system — writing the PRFAQ, routing it through layers of review — the paperwork alone could have taken as long as building and shipping the actual product.

Similar stories are playing out across the division. 

• One team open-sourced Strands, an AWS software development kit for building AI agents, after a member of Sivasubramanian’s team messaged him at 7 a.m. with the idea. After a quick call with Garman, they decided to go ahead. Within days, it was done.
• Kiro, the AI coding tool, was built by a deliberately small team, using Kiro itself to build it. One engineer prototyped a complex cross-platform notification feature for Kiro that had been estimated at four weeks of work, and shipped it in a day and a half. 
• The internal Amazon team that rebuilt the inference engine for the company’s Bedrock platform for AI models did it with six engineers in 76 days, a project originally expected to take 30 developers 12 to 18 months. 

Smaller teams everywhere

What’s happening inside Amazon’s agentic AI division is part of a trend across the tech industry toward smaller teams and flatter organizations, driven by AI and agents. 

Microsoft’s 2026 Work Trend Index, a survey of 20,000 workers in 10 countries, found that the biggest factor behind AI’s real impact in the workplace isn’t individual skill but whether the organization has restructured around the new technologies. 

Vijaye Raji, OpenAI’s CTO of applications, said during a recent Technology Alliance event that the company’s “ambitions are growing faster than we can hire people” — but the profile of who gets hired is changing. OpenAI increasingly looks for engineers who work with AI tools natively, and the gap between those who do and those who don’t is stark: the top engineers at OpenAI use roughly 100 times more AI tokens than the median.

All of this leads to a natural question: what does this mean for jobs? Amazon has cut roughly 30,000 corporate jobs since late 2025 as part of what Jassy has described as an effort to reduce bureaucracy. He has said he expects AI to shrink the corporate workforce over time. 

Similar cuts are playing out across the industry, from Meta to Block to LinkedIn, as companies rethink not only the roles they need to fill but also how many people they need overall. 

Bigger goals, same team

Sivasubramanian describes the shift differently: In his division, the same number of people are now pursuing a bigger charter. With the new structure, they’re able to take on more projects, and faster, accomplishing things in weeks that would have taken much longer in the past.

The nature of the roles inside those teams is changing, too. Increasingly, product managers write code, and engineers make product decisions. On the Kiro team, for example, a product manager built the first version of a cost analysis dashboard using Kiro itself. 

This also requires leaders to operate differently. For example, Sivasubramanian said he is careful to monitor which decisions need his approval, even when traveling. At the current pace, even four or five days of delay can add as much as 10% to a team’s shipping timeline. 

Managing these teams also raises new questions. Sivasubramanian said his division has started tracking how much it spends on AI tokens — the basic unit of interaction with an AI model — the way it would track any other operating cost. 

So far, the numbers have been manageable: tools like Kiro invest upfront in defining specs and pulling in the right context before generating code, which makes them more efficient with tokens rather than burning through them in aimless back-and-forth. 

Even the heaviest users consume only a few thousand dollars a month, he said. But he expects that over time, companies will need a full picture of their operating expenses that includes not just headcount but the cost of the AI agents working alongside them.

This gets to a bigger point: “The bottleneck is not about the time it takes to build something,” Sivasubramanian said. “The bottleneck is about crafting the right specification and the tests and the right product and customer experience.”

In a blog post published last week, Sivasubramanian wrote that teams across the company that restructured their workflows around AI saw a median 4.5x productivity gain, with some exceeding 10x gains. The teams that simply added AI tools to their existing way of working didn’t see the same results.

Coding and testing

That shift has created its own challenges. Teams can generate code faster than ever, but if they don’t define what success looks like up front — the specs, the tests, the edge cases — the agents don’t have as much chance of success. 

Amazon is now pushing testing to the moment of coding rather than handling it in stages, so agents can check their own work before anything reaches production.

Sivasubramanian learned this first-hand, the hard way. Earlier this year, jet-lagged and unable to sleep in his hotel room on a trip to India, he decided to try a fun project: He used Kiro to rebuild a piece of AWS infrastructure he’d originally developed by hand nearly 20 years ago — a replication engine that still underpins core services like S3 and DynamoDB.

He and one of Amazon’s earliest distinguished engineers, Allan Vermeulen, had spent four months on the original. Sivasubramanian figured the agent would make quick work of it. Instead, he spent four nights going back and forth, babysitting each step. 

On the fifth night, he realized the problem: he hadn’t given the agent the tools to test its own output. Once he wrote the right spec and set up the testing environment, it was done in about two hours. Asked what he did with his rebuilt version of the engine, Sivasubramanian laughed. He never shipped it. “Maybe I should have,” he said.

With the right team and a couple of pizzas, maybe he still can.

Updated with comments from Seattle Mayor Katie Wilson.

A new Downtown Seattle Association report asserts that Seattle’s signature tax on big employers is backfiring, five years after it went into effect, holding up nearby Bellevue as an example of the jobs and prosperity the city has missed out on in the process.

The report, released Monday afternoon, finds that downtown Seattle has lost about 30,000 jobs since 2020 and that the taxable value of its office buildings has fallen 48% — even as Bellevue, which has no comparable tax, added jobs and saw commercial values rise 7%. 

JumpStart, passed in 2020 and in effect since 2021, taxes the payrolls of Seattle’s largest employers, including Amazon and other big tech companies. It’s projected to raise about $388 million this year, down from earlier forecasts, due in part to the loss of high-paying jobs. 

It is, to a large extent, a tax on big tech. About 70% of JumpStart revenue comes from just 10 companies, most in the technology sector, according to the city’s budget office. That’s a reflection of how heavily Seattle’s economy leans on a handful of large tech employers.

“These were a set of taxes that may have provided some short-term gain to the city coffers, but are inflicting long-term pain,” DSA President and CEO Jon Scholes said in an interview. “We predicted that at the time, and were sort of dismissed and ignored.” 

Seattle Mayor Katie Wilson, in a statement Monday evening, credited JumpStart with helping the city recover from the pandemic and cautioned against blaming downtown’s challenges on any single cause.

The tax has raised far more than originally projected over the past several years, she said, and let the city avoid deep budget cuts that would have dragged on the local economy.

“We should be careful not to oversimplify the challenges facing downtown and our regional economy,” Wilson said, blaming “chaotic and counterproductive national economic policies” for higher costs and interest rates that have slowed investment across the city, region, and country.

Wilson also cited the pandemic, the rise of remote work and broader shifts in the tech sector as forces that have affected cities well beyond Seattle. The city’s recovery, she said, has remained resilient and competitive even as her administration works to diversify the economy for the future.

Amazon had started to expand in Bellevue prior to the JumpStart tax, following the city’s short-lived 2018 “head tax,” a JumpStart precursor that the council at the time passed and quickly repealed. The company has since built its Bellevue workforce to about 15,000 people, part of what it now calls its broader Puget Sound regional headquarters. 

JumpStart was an early example of a wave of new taxes in Washington that has prompted business and tech leaders to warn of an increasingly anti-business climate. Lawmakers have since added a capital gains tax and, this spring, a 9.9% tax on income above $1 million — fueling concerns from some executives about the state’s competitiveness.

The DSA is not calling for outright repeal of the Seattle tax. Scholes said the group wants a “course correction” — incentives and the temporary suspension of payroll or business taxes for companies that invest in Seattle, along with a more welcoming posture from City Hall toward employers. 

The tax was created to fund affordable housing, small-business support, climate programs and equitable development, with the largest share (about 62%) going to housing. But amid recurring budget shortfalls, the city has tapped JumpStart to help prop up its general fund, transferring about $201 million — roughly 47% of the tax’s revenue — to general government operations this year, according to budget documents.

DSA may face a challenge in proving a direct causal link between the tax and the trends in downtown Seattle. Downtowns across the country, including San Francisco, Portland and Chicago, have seen office values fall and vacancies climb since the pandemic with no comparable tax, due to remote work, tech-sector layoffs and AI-driven cuts. 

Scholes asserted that Bellevue has faced similar pressures yet kept growing.

“We think it’s a pretty good control group over there,” he said, attributing the divergence to Seattle’s higher cost of doing business and an unwelcoming “tone and tenor” toward employers. 

Scholes said he was encouraged by early signals from Wilson, who has asked city departments to identify spending reductions ahead of her 2027 budget, due late this summer. He credited the mayor for that but added that the DSA is taking a wait-and-see approach overall. 

In her statement, Wilson said Seattle “remains one of the fastest-growing big cities in the country,” but added that the city needs “to do more to push against the global and national headwinds and build a city with more businesses opening here, thriving here, and providing jobs here.”

The key to improving its economic climate, she said, is addressing homelessness, improving public safety, and making Seattle a better, more affordable place to live and work.

Read the DSA report here.

Amazon CEO Andy Jassy was reportedly among the tech leaders who communicated with senior Trump administration officials about security risks in Anthropic’s most advanced AI models, before a government order forced the AI lab to take its two newest models offline.

The situation puts Amazon in an unusual and potentially awkward position with Anthropic, in which it has invested $13 billion since 2023, with plans to put in as much as $20 billion more. 

The Information first reported the calls between Jassy and senior officials, citing two people familiar with the conversations. The Wall Street Journal reported that Jassy told Treasury Secretary Scott Bessent and others that Amazon researchers had used Anthropic’s Fable 5 to obtain information that could be used in cyberattacks.

Amazon shared those findings with administration officials, according to the reports.

“As a leading cloud provider that serves a large number of private and public sector customers, it’s not uncommon for governments to seek our counsel on potential security risks,” an Amazon spokesperson said in a statement to GeekWire on Monday morning. However, the statement added, the company doesn’t share the details of these discussions when they occur.

The administration’s directive, issued Friday afternoon, cited a method for jailbreaking Anthropic’s Fable 5 — a general-use version of its more powerful Mythos 5 model — to extract information that could aid cyberattacks. The order suspended access for any foreign national, forcing Anthropic to disable both models for all users to comply.

Axios reported that Amazon was among at least five companies that raised concerns with administration officials on Thursday night and Friday before the order came down. 

In a statement Friday evening, Anthropic said it was complying with the government’s legal directive but disagreed that the situation warranted the action. The company said the vulnerabilities identified using Fable were “relatively simple” and could be found using other publicly available models, including OpenAI’s GPT-5.5. 

“If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers,” the company said. 

Independent experts have questioned the severity of the finding. Andrew Morris, founder of the cybersecurity firm GreyNoise Intelligence, told the Journal that Amazon’s report showed Fable could surface security bugs in at least four software programs, but that the information was “still a long way from dangerous cybersecurity information.” 

Fable 5 remains unavailable to Anthropic’s Claude users as of publication time.

It’s the latest twist in a contentious relationship between Anthropic and the Trump administration. Earlier this year, the Pentagon designated the company’s model as a supply-chain risk, after the two sides clashed over whether Anthropic’s models could be used for purposes such as mass domestic surveillance or in lethal autonomous weapons. 

Amazon Web Services on Thursday announced that efforts to curb water use at its data centers have made it seven times more water-efficient than the industry average.

The company says it’s 75% of the way toward its goal of being water positive by 2030, meaning for each gallon consumed at a data center, it will return a greater volume to the same community where it was drawn.

Data center operators are trying to address concerns about water and energy usage as AI adoption drives massive expansion of the facilities.

Even in Amazon’s backyard, resistance is growing. Seattle’s city council this week unanimously approved a one-year emergency moratorium on new large data centers inside city limits.

AWS executives said the reality of these facilities can differ from public perception.

“As we’ve been engaging with our local communities, they’ve been very pleasantly surprised about how little water we are using,” Kerry Person, AWS vice president of Data Center Operations, told GeekWire. “We’re starting to share more and more of this information publicly to really just educate folks.”

Data centers use a variety of strategies to keep their electronics cool. Those include fans, air that’s cooled using evaporated water, air conditioning and direct liquid cooling. The approaches involve resource tradeoffs: air conditioning draws more electricity but saves water, while evaporative cooling is less energy-intensive but consumes more water.

AWS uses fans to cool its facilities about 90% of the time, drawing in outside air, blowing it past server racks and releasing it back outside. The company switches to evaporative cooling when outside temperatures exceed roughly 85 degrees. Another water savings was gained by researching the maximum temperatures its electronics can tolerate, and running machines under warmer conditions.

That allows the company to use 0.12 liters of water per kilowatt-hour of operations, compared to an industry average of 0.84 liters. The rate applies to both Amazon-owned facilities and leased data center space internationally, and has been verified by outside auditors.

While it touts its own accomplishments, Amazon also notes that the global data center industry uses less water than many may realize, accounting for 0.5% of all industrial water use worldwide.

Other tech companies are likewise implementing water-saving strategies and policies. Earlier this year, Microsoft pledged a 40% improvement in water efficiency by 2030 and committed to replenishing more water than it uses in each district where it operates. It also started installing closed-loop systems where water flows past heat-generating processing chips, drawing off heat that it carries to chillers. Then the cooled water starts the journey all over again.

But public concerns persist, particularly in regions facing water shortages. In 2025, Bloomberg reported that nearly two-thirds of the U.S. data centers that were built or are under development in the past three years are located in water-stressed areas.

Simon Hans Edasi, a Seattle-area data scientist and geospatial researcher, has examined data center locations in Washington state relative to water availability, energy access and other factors. He raised concerns about Amazon’s planned $4.8 billion campus in Burbank, near the Columbia River. The industry overall is moving “deeper into arid eastern Washington,” Edasi said.

Without addressing that specific project, Will Hewes, Amazon’s water stewardship lead, said the company focuses on three things at each location: drawing as little water as possible, using recycled water sourced from treatment plants rather than drinking water supplies, and partnering with local organizations to replenish water back into the area.

“For any of those water-stressed basins where we’re operating, we’re making sure that in each of those we’re also putting more back,” Hewes said.

Replenishment efforts vary by location. They can include programs such as helping farmers use wastewater from data centers for irrigation, or working with building managers to fix water loss from running toilets and leaky faucets.

AWS consumed about 2.5 billion gallons of water for its data centers worldwide last year. Through replenishment efforts, the company reports returning 3 gallons for every 4 that it used.

For the first time in 13 years, the Fortune 500 has a new No. 1— and it’s Amazon.

The Seattle-based e-commerce and cloud giant knocked Walmart from its familiar perch atop the ranking of the biggest U.S. companies by revenue.

Amazon surpassed $700 billion in revenue in 2025 with a 12% year-over-year jump. Just over 20 years ago, the company made its debut on the list at No. 492.

Only four companies have ever held the No. 1 spot in the 72-year history of the Fortune 500 list: General Motors, ExxonMobil, Walmart, and now Amazon.

Walmart fell to No. 2 for the first time since 2012. Here is the top 10:

• Amazon
• Walmart
• UnitedHealth Group
• Apple
• Alphabet
• CVS Health
• Berkshire Hathaway
• McKesson
• Exxon Mobil
• Cencora

The companies on the list combined for $21.0 trillion in revenue and $2.1 trillion in profits last year, while employing 30.5 million people worldwide, according to Fortune.

Amazon posted first-quarter sales of $181.5 billion, up 17%, and operating income of $23.9 billion, up 30%, in earnings reported at the end of April. Amazon Web Services growth accelerated to 28% in the first quarter — its fastest pace in nearly four years.

The tech giant, which employs 1.5 million globally, laid off 16,000 corporate employees in January in the second phase in a restructuring that began last October and totals 30,000 positions — adding up to the largest workforce reduction in the company’s history.

Additional tech-focused nuggets from the Fortune 500 list:

• Microsoft climbed one position to No. 11 to hit its highest ranking ever, fueled by heavy integrations with OpenAI, Azure cloud, and Copilot tools.
• Alphabet repeated as “most profitable company,” eclipsing the $100 billion profit mark for the second year in a row. The company’s $132 billion earnings figure is a record for a Fortune 500 company.
• Nvidia (No. 16) became the first $4 trillion dollar Fortune 500 company and moved past Apple to claim the most valuable company title.
• Meta (No. 17) jumped five spots for its highest-ever ranking.

This week marks one of the biggest events on the modern video game calendar, as studios from around the world bring their newest projects to the annual Summer Game Fest event in California. This includes a new look at Amazon Game Studios’ impending reboot of the long-running Tomb Raider franchise, coming in early 2027.

Tomb Raider: Legacy of Atlantis is a ground-up “reimagining” of the original 1996 Tomb Raider, which celebrates its 30th anniversary in October. Once again, it follows British archaeologist Lara Croft (Alix Wilton Regan) on an expanded look into her journey to collect the scattered pieces of an artifact from the lost civilization of Atlantis. Along the way, she’ll solve puzzles, navigate treacherous labyrinths, and fight dinosaurs, as one does.

The original Tomb Raider’s story and environments have been rebuilt with Unreal Engine 5 for Legacy of Atlantis, which turns the game into less of a series of vaguely connected puzzle boxes and more of an open-ended area that you can freely explore. It’s currently planned for release on Feb. 12, 2027.

Legacy of Atlantis is a co-production between the Polish studio Flying Wild Hog (Hard Reset, Shadow Warrior) and Crystal Dynamics, which maintains offices in Texas, California, and Bellevue, Wash.

It’s also the first step in Amazon’s planned franchise reboot of Tomb Raider, which was first announced back in 2022. Legacy will lead directly into a brand-new game, Catalyst, which is planned for release later next year and is a direct follow-up to 2008’s Tomb Raider: Underworld.

The new Legacy of Atlantis trailer premiered as part of Sony’s semi-regular State of Play, a livestreamed showcase of new and upcoming games for the PlayStation platform.

Other Pacific Northwest gaming news out of the State of Play included the official debut of Marathon’s second “season” of content, Nightfall, which resets players’ progress in order to present them with new challenges and an even playing field.

Snowflake on Wednesday committed to spend $6 billion on Amazon Web Services over five years, adding to the cloud giant’s growing roster of AI infrastructure deals.

Snowflake, which sells cloud-based data warehousing and AI tools to big businesses, said the commitment includes the use of Amazon’s custom Graviton processors and other chips to power AI and agentic applications.

The deal expands a relationship that dates back to the data company’s founding 11 years ago. Snowflake’s five-year AWS spending commitment has grown from $1.2 billion at the time of its IPO in 2020 to $2.5 billion in 2023 to the current $6 billion, CNBC reported.

It’s the latest in a series of large-scale commitments for AI-related infrastructure on AWS, including deals of more than $100 billion with Anthropic, $138 billion with OpenAI, accompanying Amazon’s investments in the AI labs. Meta also plans to deploy tens of millions of Graviton chip cores for agentic AI.

Amazon’s custom chips business has become a major revenue driver. Amazon CEO Andy Jassy said in his annual shareholder letter in April that the business generates more than $20 billion a year and is growing at triple-digit rates.

Two large customers asked to buy all of Amazon’s available Graviton capacity for 2026, Jassy wrote in the letter at the time, and the company was compelled to turn them down.

Snowflake is headquartered in Bozeman, Mont., with a large presence in the Bay Area and a significant office in Bellevue, Wash. The company reported strong fiscal first-quarter results Wednesday, with revenue of $1.39 billion, beating analyst expectations.

Its stock rose as much as 33% in extended trading.

Amazon’s cloud division announced an AI shopping assistant for retailers, following the company’s broader blueprint of turning its internal technology into products for others.

The new tool from Amazon Web Services, the AWS Agentic Shopping Assistant, is built on the same technology that powers the Alexa for Shopping assistant on Amazon.com, formerly known as Rufus, which the company says drove nearly $12 billion in incremental sales last year.

It’s designed to let retailers create AI assistants for their own e-commerce sites that can talk with shoppers, answer questions about products, and make recommendations tailored to each store’s inventory and brand. AWS says a retailer can get one up and running in about 60 days.

The announcement is the latest move in the broader competition among tech giants to control different pieces of the AI shopping experience.

• Google is building shopping features into its AI-powered search results, and has also partnered with Shopify on an open commerce standard to let AI agents interact with merchant checkout systems.
• Microsoft has added checkout tools to its Copilot assistant and has been working with retailers like Ralph Lauren on custom AI shopping experiences.
• OpenAI has been working with Shopify and Walmart to surface products in ChatGPT after its initial Instant Checkout feature fell flat.
• Walmart last year launched its own AI shopping assistant, nicknamed Sparky, and has since integrated it into ChatGPT and Google’s Gemini, betting on being present across AI platforms.

With its new release, AWS is betting that retailers will want to build their own AI shopping experiences, while leveraging the experience of Amazon’s own e-commerce platform.

The stakes are significant. Accenture estimates that by 2030, more than 30% of online commerce could run through AI agents, representing about $3.1 trillion in transactions.

Amazon’s pitch requires retailers to trust its cloud division with their AI shopping infrastructure, even as Amazon’s retail arm competes against them for customers. AWS says retailers using the AWS Agentic Shopping Assistant will keep control of their own customer data, product catalogs, and business rules, with each deployment customized to the retailer’s brand.

An early retail customer is Kate Spade, the fashion and accessories brand. Its parent company, Tapestry, used the tool to launch an AI gift concierge in April that engages shoppers in conversation about the occasion, recipient, and style before recommending products.

Amazon says the concierge was built on Anthropic’s Haiku 4.5 model through Amazon Bedrock, and went through roughly 2.5 months of testing before going live.

When CNBC’s Andrew Ross Sorkin described Jeff Bezos’ startup Project Prometheus as being “really about AI robotics” in an interview on Wednesday, the Amazon founder interrupted with a correction.

“It’s a little premature for me to talk about it, but we are not — we have nothing to do with robotics,” Bezos said. He went on to offer the most detailed public description yet of the secretive startup, for which he is co-CEO: Prometheus is developing an “artificial general engineer,” he said, building next-generation tools for designing physical objects. 

Bezos called it “a very, very modern version” of CAD, or computer-aided design, adding that he is “really oversimplifying here,” and reiterating that it’s premature for him to give much detail.

He said Prometheus is “something I got so excited about that I became the co-CEO of the company, putting a lot of time into it, a lot of energy into it.” 

The tools Prometheus is building will “help companies like Blue immensely,” Bezos added, referring to his space venture Blue Origin. However, he said, the company “deserves its own special focus” as “its own big idea” rather than being housed inside Amazon or Blue Origin.

His comments mark a rare hint of what’s happening inside a company that has operated almost entirely in stealth since news of its formation was reported in November 2025.

Project Prometheus launched with $6.2 billion in funding, led by Bezos and Vik Bajaj, a former Google X executive. Bloomberg reported in April that the company closed a $10 billion round at a $38 billion valuation, with JPMorgan and BlackRock among the investors.

The company has hired roughly 120 employees from firms including OpenAI, DeepMind, Meta and xAI, and operates out of San Francisco, London and Zurich.

Much of the outside reporting on Prometheus has characterized it as focused on robotics and manufacturing automation, which came in part from analyzing LinkedIn profiles of its hires. Bezos’ description Wednesday as a design-tools company focused on engineering physical objects suggests a different ambition, or at least a more specific one.

Amazon.com and Alexa are finally talking to each other.

The tech giant on Wednesday announced Alexa for Shopping, a new capability that connects its Rufus e-commerce chatbot with its Alexa+ assistant, aiming to unify product research, user preferences and shopping activities across Amazon’s apps, websites and Echo devices.

The move comes as consumers increasingly turn to popular AI assistants like OpenAI’s ChatGPT and Google’s Gemini for shopping advice. By creating a more integrated AI shopping experience, Amazon is aiming to keep that research and the resulting purchases on its own platforms.

Several of Amazon’s new features reflect the broader push into agentic AI, which takes action on a customer’s behalf. For example, Alexa for Shopping can monitor prices and automatically purchase an item when it hits a target, or restock household essentials on a schedule.

With the integration, Amazon is retiring the “Rufus” name from its shopping interface, replacing the chatbot with Alexa for Shopping branding in its app and on its website. Amazon says Rufus will continue to power parts of the experience behind the scenes. 

Broader landscape: ChatGPT, Gemini and Perplexity have all launched shopping features in recent months, with Google enabling in-chat checkout from retailers like Walmart and Wayfair. (OpenAI pulled back on its in-chat checkout feature in March after it failed to gain traction).

Amazon is also looking to keep rival AI agents off its platform: A federal judge in March blocked Perplexity’s Comet browser from shopping on Amazon on behalf of users, though the order was stayed pending appeal. In a statement at the time, Amazon called it “an important step in maintaining a trusted shopping experience for Amazon customers.”

On the product front, Amazon is betting that a unified and personalized experience will matter more to customers than the ability to compare products across retailers in a general-purpose AI assistant.

Rollout details: Alexa for Shopping will roll out in the U.S. over the coming week, the company says. It will be available for free to customers signed into an Amazon account through the Amazon Shopping app and Amazon.com, with no Prime membership, Echo device or Alexa app required. 

The company is also bringing the full Amazon shopping experience to Echo Show devices, starting with Alexa+ customers on the latest Echo Show 15 and 21, with other devices to follow.

Use cases and features: Rajiv Mehta, Amazon’s vice president of Conversational Shopping, said the company saw customers starting shopping “missions” in one place and restarting them somewhere else because Rufus and Alexa didn’t share memory or context.

The idea is that “the customer doesn’t have to think about where they started a discussion with Amazon,” Mehta said in an interview with GeekWire. The feature uses what customers have already told Amazon once, then makes that context available on other Amazon devices, sites, and apps.

For example, citing his own usage, Mehta said a customer could brainstorm a science fair project with Alexa on an Echo device, then open the Amazon app and ask for supplies without re-explaining the project. Or a shopper could research laptops in the Amazon app, set a price alert, and get notified on their Echo when the price drops, and buy it with a voice command.

Other features in Alexa for Shopping include:

• Asking questions directly in the main Amazon search bar, rather than opening a separate chat window.
• Scheduled actions that automate tasks like restocking household essentials, getting alerts when a favorite author releases a new book, or adding a product to a cart when it drops below a set price.
• Custom shopping guides for big purchases that compare features, prices and reviews across Amazon and the web.
• Product price history expanded to a full year, up from 30 and 90 days.

Privacy and personalization: Amazon says customers can review and manage their Alexa interactions and conversations through the Alexa Privacy Dashboard. They can also ask Alexa for Shopping what it knows about them and update personal details like family members, pets, interests and dietary needs.

Amazon’s evolution: Rufus launched in 2024 and was used by more than 300 million customers in 2025, according to the company. On Amazon’s most recent earnings call, CEO Andy Jassy said monthly active users of Rufus were up more than 115% and engagement was up nearly 400% year over year.

Jassy compared third-party AI shopping agents to the early days of search engines referring business to e-commerce. Those agents lack personalization features and shopping history and often can’t get pricing or product information right, he said, noting that customers who want to shop at a specific retailer will often start with its own assistant as a result.

Amazon’s ambition, Jassy said, is to develop “the best shopping assistant anywhere.”

Amazon Web Services is adding a feature to its Kiro AI coding tool designed to mathematically prove that software requirements are free of contradictions and gaps before any code gets written, addressing one of the core risks of AI-assisted software development.

The feature, called Requirements Analysis, is designed to catch the kind of bugs that can often be the hardest to spot and most expensive to fix — problems that start not in the resulting code but in the initial requirements that define what the software is supposed to do. 

The announcement Tuesday morning comes three months after Amazon publicly pushed back on a Financial Times report that its AI coding tools contributed to AWS outages, an episode that highlighted the risks of giving AI agents too much autonomy in software development. 

It also comes a day after AWS hired former Microsoft exec Shawn Bice to return to Amazon as VP of AI Services leading its Automated Reasoning Group, the team behind the new feature. Bice will report to Swami Sivasubramanian, Amazon’s VP of Agentic AI.

Requirements Analysis combines large language models with an automated reasoning engine called an SMT solver. The LLM translates natural-language requirements into formal logic. 

The solver then checks those requirements by mathematically proving whether they contradict each other or leave gaps that could be filled in erroneously by the AI coding tool — a common problem as AI increasingly generates software faster than developers can review it. 

“Every vague prompt produces a vague spec or plan, and the AI agent implementing that spec produces code full of undisclosed decisions made on your behalf, without your awareness or agreement,” wrote AWS applied scientists in a blog post accompanying the news. 

Kiro competes in a crowded and fast-growing market for AI coding tools that includes Cursor, GitHub Copilot, Anthropic’s Claude Code, Google’s Antigravity, and OpenAI’s Codex. 

While those tools have increasingly added planning and agent workflows alongside code generation, Kiro has built its identity around a spec-first approach that requires developers to formalize their intent before the AI starts building.

AWS also announced two other Kiro features designed to speed up the development process. 

• Parallel Task Execution runs independent coding tasks concurrently rather than sequentially, cutting implementation times for large projects by roughly 75 percent, according to the company. 
• AWS says a new Quick Plan mode lets developers skip the step-by-step approval process for well-understood features, generating a full set of requirements, design, and tasks in one pass. 

PREVIOUSLY: Amazon’s surprise indie hit: Kiro launches broadly in bid to reshape AI-powered software development

Amazon is expanding its 30-minute “Amazon Now” service to dozens of U.S. cities, with plans to reach tens of millions of customers across the country by the end of the year.

The service, which GeekWire first uncovered through permit filings in Seattle’s Ballard neighborhood last November, delivers fresh groceries, household essentials, and other items from small neighborhood fulfillment hubs using its Amazon Flex citizens’ delivery brigade.

It has been limited until now to Seattle and Philadelphia as test markets.

The company announced Tuesday morning that Amazon Now is now widely available in Atlanta, Dallas-Fort Worth, Philadelphia, and Seattle, and is “rapidly expanding” into Austin, Denver, Houston, Minneapolis, Orlando, Oklahoma City, Phoenix, and other cities. 

The expansion is the latest and most aggressive move in Amazon’s push to dominate “sub-same-day” delivery, a category where it’s competing with Gopuff, DoorDash, and Instacart. 

Amazon CEO Andy Jassy highlighted Amazon Now in both his annual shareholder letter and the company’s Q1 earnings call last month, framing it as part of a broader quest for faster deliveries that also includes drones, one- and three-hour options, and same-day shipping. 

Prime members pay a $3.99 delivery fee for Amazon Now, compared with $13.99 for non-Prime customers. Orders under $15 carry an additional small-order fee of $1.99 for Prime members and $3.99 for non-Prime — a change from the flat $1.99 fee when it launched in December.

In two separate GeekWire tests of Amazon Now in Seattle, orders arrived well under the 30-minute promise. Reporter Kurt Schlosser got his delivery in 23 minutes in December, and a separate live test during the GeekWire Podcast in February clocked in at 19 minutes.

Amazon also faces competition from Walmart, which has been quietly hitting similar speeds from its existing store network. Walmart CEO John Furner said in February that the company is averaging under an hour on Express Delivery orders, with a large number arriving in less than 30 minutes — without dedicated micro-hubs.

Amazon isn’t promising the fastest delivery times. Gopuff’s Fam20 promises 20 minutes, and DoorDash piloted 10-to-15-minute deliveries in New York before ending the program. But reliability has been a weak spot for speedy delivery services, and with its logistics expertise, Amazon is betting that consistently hitting 30 minutes matters more than promising 10. 

Walmart is also a competitor in this realm, quietly hitting similar speeds from its existing store network. Walmart CEO John Furner said in February that the company is averaging under an hour on Express Delivery orders, with a large number arriving in less than 30 minutes.

Amazon has struggled to make the economics ultra-fast delivery work in the past.

Its “Prime Now” one- and two-hour delivery service launched in 2014 and was shut down in 2021, and “Amazon Today,” which used Flex drivers to pick up orders from malls and retailers, was discontinued in late 2024 after drivers often left stores with just one or two items.

Amazon Now takes a different approach, using dedicated company-operated hubs designed specifically for rapid fulfillment. 

The service is accessible through the Amazon shopping app and website at amazon.com/now. Amazon declined to provide a full list of cities where the service will soon be available but said customers can check the app to see if Amazon Now is offered in their area.

Microsoft security exec Shawn Bice is returning to Amazon Web Services as VP of AI Services, leading the company’s Automated Reasoning Group as AWS doubles down on making AI agents more reliable. 

Bice will report to Swami Sivasubramanian, Amazon’s VP of Agentic AI, according to an internal AWS email Monday afternoon, viewed by GeekWire. 

“We are at an inflection point with Agentic AI,” Sivasubramanian wrote in the email, explaining that bringing AI and automated reasoning together is essential to building trustworthy agents. 

It’s a full-circle move for Bice. He worked at Microsoft early in his career and spent five years running AWS’s database portfolio before another former AWS leader, Charlie Bell, recruited him back to Microsoft in 2022 to help build the Redmond company’s revamped security organization. 

At Microsoft, Bell stepped down from that security leadership role in February to become an individual contributor.

Amazon’s Automated Reasoning Group uses a discipline known as neurosymbolic AI, which combines traditional AI’s pattern-matching abilities with mathematical techniques that can prove whether software is doing what it’s supposed to do.

In the email, Sivasubramanian wrote that bringing automated reasoning and AI together is “the fundamental premise” behind AWS’s investment in the field, calling it critical to building agents that businesses can trust to act on their own. 

AWS has been facing questions about the reliability of AI agents in its own operations. In February, Amazon pushed back on a Financial Times report that its Kiro AI coding tool had caused AWS outages, though it acknowledged a limited disruption to a single service after an AI agent was allowed to make changes without human oversight.

At Microsoft, Bice’s role had expanded to Corporate Vice President of Security Platform & AI, overseeing Microsoft Security Copilot, Microsoft Sentinel, and AI security research, according to his LinkedIn profile.

Before that, he spent a year as president of products and technology at Splunk, sandwiched between his two stints at the larger companies.

Bice originally joined Microsoft in 1997 and spent more than 17 years there across two stints, in roles that included managing SQL Server and Azure cloud data services. He left for AWS in 2016, where he ran the database portfolio — including Amazon Aurora, DynamoDB, and RDS — for five years before departing in 2021.

He also serves on the board of WaFd Bank, where he chairs the technology committee.

Amazon launched a new business that opens its entire logistics network to outside companies — sending shares of UPS and FedEx tumbling and marking the latest example of the tech giant under CEO Andy Jassy turning its internal capabilities into products and services for sale.

Amazon Supply Chain Services, announced Monday morning, brings together the company’s freight, distribution, fulfillment, and parcel shipping operations into a single offering available to any business, regardless of whether they sell on Amazon’s marketplace.

Initial customers include Procter & Gamble, which is using Amazon’s freight network to transport raw materials; 3M, which is using it to move products to distribution centers; Lands’ End, which is fulfilling orders across sales channels from Amazon’s warehouses; and American Eagle Outfitters, which is using Amazon’s parcel service for last-mile delivery.

The service can fulfill orders placed through platforms that compete with Amazon’s own marketplace, including Walmart, Shopify, TikTok, and others. 

Shares of UPS dropped nearly 10% and FedEx fell more than 9% in trading early Monday. Amazon’s stock rose slightly. Amazon had already surpassed both carriers to become the nation’s largest parcel shipper by volume, according to parcel-analytics firm ShipMatrix.

Peter Larsen, vice president of Amazon Supply Chain Services, compared the launch to the origins of Amazon’s cloud business. Larsen, an 18-year Amazon veteran who previously led internal transportation and delivery technology operations, said Amazon is bringing its supply chain to outside businesses “much like Amazon Web Services did for cloud computing.” 

In addition to putting Amazon in competition with existing players in the logistics industry, the move also raises questions about data privacy. Amazon has faced accusations of using nonpublic seller data to compete against merchants on its marketplace, which it has denied. 

Larsen told the Wall Street Journal that the company prohibits using supply chain customer data for its own marketplace decisions, noting that hundreds of thousands of Amazon sellers already trust the company to fulfill orders placed on rival platforms. 

The launch follows a recent pattern of Amazon reviving its tradition of turning internal capabilities into external businesses. 

• Last week, AWS announced new agentic AI technologies that build on Amazon’s supply chain and hiring expertise for other companies. 
• In his annual shareholder letter, Jassy said the company is also exploring selling its custom AI chips and robotics to outside customers. 

In shipping, the company is not exactly starting from scratch: Amazon’s logistics network includes more than 200 fulfillment centers in the U.S., more than 80,000 trailers, 24,000 intermodal containers, and 100 aircraft. The company says it delivers 13 billion items annually. 

Amazon did not disclose specific pricing for the new Amazon Supply Chain Services, saying costs will vary based on the services businesses use.

Microsoft and Amazon joined other leading artificial intelligence companies in signing deals to deploy their technology in classified Pentagon networks, the Defense Department announced Friday, accelerating a push to build what the military is calling an “AI-first fighting force.”

The agreements — which also include OpenAI, Google, Nvidia, SpaceX and the startup Reflection — will give those firms’ AI systems access to the military’s most classified network environments, known as Impact Level 6 and Impact Level 7. The Pentagon said the technology will be used to analyze data and improve battlefield decision-making.

“Together, the War Department and these strategic partners share the conviction that American leadership in AI is indispensable to national security,” the Pentagon said in a statement, using the Trump administration’s preferred name for the Defense Department.

The Pentagon says the effort is already well underway. More than 1.3 million Defense Department personnel have used GenAI.mil, the military’s official AI platform, generating tens of millions of prompts and deploying hundreds of thousands of agents in just five months, according to the department. Officials say the technology has cut some tasks from months to days.

The deals come as the Pentagon is locked in a legal battle with Anthropic, one of the nation’s leading AI labs, which had sought guarantees its technology would not be used for mass domestic surveillance or fully autonomous weapons. The Defense Department moved to blacklist Anthropic earlier this year, calling the company a national security risk — a designation Anthropic is contesting in court.

On Thursday, Defense Secretary Pete Hegseth called Anthropic CEO Dario Amodei an “ideological lunatic” and slammed the company during a Senate Armed Services Committee hearing.

Bloomberg reported that the Pentagon negotiated its deal with Amazon Web Services late into Thursday night, according to two officials briefed on the talks.

“We look forward to continuing to support the Department of War’s modernization efforts, building AI solutions that help them accomplish their critical missions,” AWS spokesman Tim Barrett said in a statement.

Hundreds of Google employees sent a letter to company leadership this week urging them to refuse to let the Pentagon use its AI on classified data.

“We want to see AI benefit humanity; not to see it being used in inhumane or extremely harmful ways,” they wrote, according to The Washington Post.

Amazon Web Services growth accelerated to 28% in the first quarter — its fastest pace in nearly four years — pushing Amazon’s results past Wall Street’s expectations and validating, at least for now, the company’s controversial $200 billion capital spending plan.

Overall, Amazon posted sales of $181.5 billion, up 17%, and operating income of $23.9 billion, up 30%. Both topped guidance and exceeded Wall Street’s expectations of about $177 billion in revenue.

Profits were $30.3 billion, or $2.78 per diluted share. However, that included a $16.8 billion pre-tax gain on Amazon’s investment in Anthropic, which inflated the bottom-line numbers. Excluding that one-time gain, adjusted earnings per share would have been $1.61, just shy of analyst expectations of $1.62.

Amazon’s advertising business grew 24% to $17.2 billion in the quarter, and the company said advertising revenue topped $70 billion over the past 12 months.

In the core e-commerce business, unit sales grew 15%. Amazon CEO Andy Jassy called it the strongest growth rate since the waning days of the COVID-19 lockdowns. It was boosted in part by faster delivery, with more than 1 billion items shipped same-day or overnight in the U.S. so far this year.

Amazon is in “the middle of some of the biggest inflections of our lifetime,” Jassy said in a release.

The company spent $147.3 billion on property and equipment over the past 12 months, nearly doubling from $88 billion a year earlier and leaving just $1.2 billion in free cash flow.

In other words, Amazon is making more money than ever but plowing nearly all of it back into building out capacity, mostly for AWS and AI infrastructure. The company has said it expects to spend about $200 billion in capital expenditures for the full year.

Shares were down about 2% in initial after-hours trading.

Update: On the earnings call, Amazon disclosed that its AWS revenue backlog jumped to $364 billion, up from $244 billion last quarter. That figure does not include a recently announced deal with Anthropic valued at more than $100 billion. Jassy said the backlog has reasonable breadth beyond just one or two customers.

Jassy also disclosed that Amazon now has more than $225 billion in revenue commitments specifically for Trainium, its custom AI chip. He called Amazon’s custom silicon business one of the top three data center chip businesses in the world and said the chips business grew nearly 40% from the prior quarter.

For the second quarter, Amazon expects revenue between $194 billion and $199 billion, with operating income between $20 billion and $24 billion. The guidance assumes Prime Day will take place in Q2 in most countries, a shift from Q3 last year.

SAN FRANCISCO — Amazon moved quickly Tuesday to capitalize on OpenAI’s new relationship status with Microsoft, launching a preview of OpenAI’s models on its Bedrock platform less than 24 hours after the ChatGPT maker was freed from its previously exclusive cloud arrangement.

The companies described the news as meeting years of customer demand.

“Their production applications run in AWS. Their data is in AWS. They trust the security of AWS, and we’ve forced them for the last couple of years, to get great OpenAI models, to go to other places,” AWS CEO Matt Garman said at an event in San Francisco.

OpenAI CEO Sam Altman appeared at the event via recorded video, noting that his “schedule got taken away” from him for the day, an apparent allusion to the opening of the trial in Elon Musk’s lawsuit against him and OpenAI in nearby Oakland.

“The opportunity ahead of us is enormous, and the most exciting part is that this is not something in the future — it’s starting right now,” Altman said in the video. 

Altman also said OpenAI’s Codex coding tool will be available to AWS customers, and the companies unveiled Amazon Bedrock Managed Agents, powered by OpenAI — the product name for the enterprise agent platform previously described as a “Stateful Runtime Environment” when the two companies announced their partnership in February.

Denise Dresser, OpenAI’s chief revenue officer, said the hundreds of enterprise customers she has met since joining the company four months ago are past the experimentation phase with AI.

“They understand that to do that, they need to have powerful models. But even more importantly, they want those models in a trusted environment that they know and a trusted infrastructure. And so for me, that’s what’s so special about this partnership,” Dresser said.

It has been a whirlwind couple of months for both companies.

Amazon and OpenAI struck a $50 billion investment and cloud deal in February, with OpenAI committing to run workloads on Amazon’s custom Trainium chips and the two companies agreeing to co-build what they called a “Stateful Runtime Environment” on Amazon’s Bedrock platform. The cloud agreement alone is worth more than $100 billion over eight years.

An internal OpenAI memo that surfaced in April touted the Amazon partnership as a key enterprise growth driver. OpenAI chief revenue officer Denise Dresser wrote that the Microsoft partnership had “limited our ability to meet enterprises where they are,” adding that inbound demand for the AWS offering had been “frankly staggering.”

Amazon has also doubled down on its original AI partner, Anthropic, investing up to $25 billion and striking a similar $100 billion-plus cloud commitment in a deal announced earlier this month.

Both OpenAI and Anthropic have committed to running workloads on Amazon’s custom Trainium processors, and Facebook parent Meta recently signed a multibillion-dollar deal to use Amazon’s Graviton chips for agentic AI. 

Amazon CEO Andy Jassy disclosed in his annual shareholder letter that Amazon’s custom silicon business is generating more than $20 billion a year in revenue.

Things changed again for the AWS-OpenAI partnership just this week.

On Monday, Microsoft and OpenAI revamped their partnership, making OpenAI free to serve all of its products — including API-based services previously exclusive to Microsoft Azure — on any cloud provider. Amazon CEO Andy Jassy immediately signaled the impact, posting on LinkedIn that OpenAI’s models would be available soon on Amazon’s Bedrock platform.

Amazon reports first-quarter earnings Wednesday, with the company on pace to spend $200 billion this year on capital expenditures, most of it on AI infrastructure. Jassy has called AI “an extraordinarily unusual opportunity to forever change the size of AWS and Amazon as a whole.

SAN FRANCISCO — Amazon manages more than 400 million products in its supply chain and hired 250,000 seasonal workers last peak season. Now its cloud division is packaging up what the company has learned and getting ready to sell it to other businesses. 

Amazon Web Services on Tuesday announced two new agentic AI products: Connect Decisions, which uses Amazon’s own supply chain models to help companies forecast demand and manage disruptions; and Connect Talent, which conducts voice-based job interviews around the clock and scores candidates on skills rather than resumes.

They’re part of a growing lineup of AWS business applications, which started with the Amazon Connect contact-center platform in 2017, which has since become a billion-dollar business. The lineup expanded more to health care with Connect Health, announced last month. 

AWS event: Amazon is announcing the products at an event in San Francisco where AWS CEO Matt Garman is expected to detail the company’s expanded work with OpenAI, following Monday’s news that OpenAI’s models will be available on Amazon Bedrock for the first time. 

That was made possible by a revamped deal between Microsoft and OpenAI, and builds on Amazon’s earlier investment of up to $50 billion in the ChatGPT maker. 

Separately, AWS is releasing a major update to Amazon Quick, its AI assistant for business users, adding a desktop app, the ability to create custom dashboards and portals, and expanded integrations with Google Workspace, Microsoft 365, Zoom, and Salesforce. 

Expanded footprint: The new Connect apps push AWS further beyond cloud infrastructure and into direct competition with enterprise software companies, including some AWS customers. 

Colleen Aubrey, AWS senior vice president of Applied AI Solutions, acknowledged that selling applications that compete with AWS customers is “a newer dynamic” for the cloud business. 

However, she noted that it’s familiar territory for Amazon overall. She compared it to the way the company sells its own products alongside third-party sellers on its marketplace, or produces original content for Prime Video while also distributing shows from other studios.

Aubrey called the new apps “a day zero” moment for the AWS applications team after spending the past two years assembling the group and doing the work to determine where to focus.

“If we’re lucky, we’ll have some hits in this collection of four,” she said in an interview, acknowledging that building enterprise software products is inherently uncertain.

Asked why companies wouldn’t simply build these capabilities themselves using AWS tools like Bedrock, Aubrey said the complexity of transforming an entire business function, not just an individual task, calls for a purpose-built product that can be used across an organization.

Amazon’s new Connect apps: Connect Decisions draws on more than 25 specialized supply chain models and tools, including one of Amazon’s own foundation models built by its Supply Chain Optimization Technologies (SCOT) team. 

When something goes wrong in a supply chain — a supplier falls behind, or demand spikes unexpectedly — it can figure out what happened, rank the problems that need human attention, and suggest what to do about them, along with the cost and trade-offs of each option. 

Connect Talent is aimed at high-volume hiring in industries like manufacturing, logistics, retail, and hospitality, rather than corporate recruiting. AI agents conduct voice interviews that candidates can take anytime, eliminating scheduling conflicts. The system strips names and resumes from the process; recruiters see anonymized competency scores and transcripts.

One early customer has started bringing Connect Decisions into business meetings to run what-if scenarios in real time, Aubrey said. That company’s procurement team has already asked to expand its use beyond supply chain planning, which AWS ultimately plans to do.

Follow-up: AWS growth climbs to 28% as Amazon’s big AI bets start to pay off

Amazon reports first-quarter earnings Wednesday with more signs than ever that its cloud business is in demand, including a $244 billion revenue backlog, blockbuster deals with Meta,  OpenAI and Anthropic, and a custom chip business that doubled in a matter of months.

The problem: a $200 billion capital spending plan, largely dedicated to new AI infrastructure, that drained Amazon’s free cash flow and sank its stock 10% last quarter. 

Here’s a preview of the key numbers and storylines to watch.

Core expectations: Wall Street expects Amazon to report about $177 billion in first-quarter revenue, up roughly 14% from a year ago, with earnings of $1.65 per share. That’s up just 4%, reflecting the growing depreciation costs from the company’s infrastructure buildout.

Amazon’s guidance for first-quarter operating income ranges from $16.5 billion to $21.5 billion — a $5 billion spread that reflects uncertainty around tariff impacts on its retail business and about $1 billion in new costs from its satellite internet project, Amazon Leo.

AWS growth: But the main event is Amazon Web Services, where analysts expect about $36.8 billion in revenue, up nearly 26% from a year ago. AWS growth has been accelerating for three straight quarters (from 17% to 20% to 24%) and investors are looking for that to continue.

On the fourth-quarter earnings call, CEO Andy Jassy described the AI market as a “barbell” — with AI research labs spending heavily on one end, and enterprises automating routine tasks on the other. The massive opportunity, he said, is in the middle: core enterprise production workloads that haven’t moved to AI yet, for the most part.

“The lion’s share of that demand is still yet to come,” Jassy said.

The question is whether that middle is starting to fill in, or whether AWS growth is still being driven primarily by a handful of giant AI lab deals.

Beyond the cloud: It’s easy to forget in the AI frenzy, but Amazon is also the country’s largest online retailer, and the first quarter brings its own set of pressures. Jassy warned earlier this year that import costs from tariffs were starting to show up in product prices, and the company faces growing competition from Walmart, Temu, and Shein for cost-conscious shoppers. 

Online store sales grew 10% to $83 billion in the holiday quarter, and third-party seller services brought in $52.8 billion. But costs are rising too: Amazon spent $31.5 billion on shipping in Q4, up 10% from a year earlier.

At the same time, the company has been cutting costs aggressively, eliminating about 16,000 corporate jobs in January in what Jassy has described as a campaign against bureaucracy, followed by additional cuts in its robotics unit in March.

Advertising remains a standout, growing 23% to $21.3 billion in the fourth quarter and emerging alongside AWS as one of Amazon’s primary profit engines. 

Amazon won’t be the only tech giant reporting Wednesday. Microsoft, Alphabet, and Meta are all scheduled to release quarterly results the same day, giving investors a chance to compare notes on AI spending and cloud growth across the industry. 

Google Cloud has been growing faster than AWS in percentage terms, adding another dimension to the debate over which company is best positioned to capitalize on the AI boom.

Check back Wednesday afternoon for coverage.

Amazon has landed talk show queen Oprah Winfrey and her video podcasts for its suite of streaming services.

“Long before the term ‘creator’ existed, Oprah was building a direct and deeply personal connection with audiences across generations — and that bond continues to grow. Creators are reshaping entertainment, and Oprah continues to pave the way. We couldn’t be more excited to partner with her on what’s ahead,” said Matt Sandler, general manager of Amazon Creator Services, in a statement.

The multi-year deal gives the tech giant distribution and advertising rights to “The Oprah Podcast” on audio and video. As part of the arrangement, Winfrey will expand her production to two new episodes per week starting this summer. The partnership also includes rights to the 25-year library of her former talk show and her “Oprah’s Book Club” and “Oprah’s Favorite Things” franchises.

The New York Times first reported the deal on Monday. The deal underscores how Amazon is betting on established names to anchor its creator strategy.

“This is the ultimate validation of where the world is going,” Steve Boom, an Amazon vice president, told the Times. “You have the most influential talk show host in history, by orders of magnitude, leaning heavily into this new world.”

Amazon purchased Wondery, a Los Angeles-based podcast studio known for producing several hit shows, in late 2020, aiming to strengthen its original audio offerings against competitors like Spotify. Last August, the company cut around 110 positions at Wondery as part of an effort to fold some of its operations into Audible and roll out its new Creator Services division.

“The podcast landscape has evolved significantly in the past few years, particularly with the rise of video-forward, creator-led content,” an Amazon spokesperson said in an August statement.

“By making these changes, we can better support creators in monetizing their content across multiple channels, help them expand their brand IP, and simplify the process for advertisers while making content more accessible to audiences wherever they prefer to consume it,” the company added.

The Winfrey deal is one of several high-profile moves in that effort. Another is the Kelce Clubhouse — a dedicated Amazon hub featuring brothers Jason and Travis Kelce, the football stars whose profiles soared after Travis’s engagement to Taylor Swift. The site brings together their video podcast, merchandise, a documentary about the duo, and promoted Audible content.

While Winfrey’s content will be available across Prime Video, Amazon Music, Fire TV Channels and Audible, her shows will also stream on YouTube and other podcast platforms.

“Expanding our reach globally is an opportunity I embrace, as we continue to connect through stories that invite new ways of seeing, and hopefully deepen, understanding,” Winfrey said in a statement.

The dreamy part of Mikaela Kiner‘s life is easy to picture. She has spent her recent winters working from a small Costa Rica beach town, taking surfing lessons before dawn, sunset walks in the sand, and Zoom calls with real palm trees swaying in the background.

But “The Reverb Way,” her new book about building and running the Seattle-based HR consulting firm of the same name, is not the postcard version of the story. 

Kiner describes what happened when new business dropped to half its usual volume, as tech layoffs, a rocky economy, and the rapid rise of AI hit Reverb’s client base. She battled insomnia so severe she couldn’t get through a workday without napping. Her daughter, watching her scramble through a client crisis, told her she’d never seen her this stressed. 

The book is a candid account of the ups and downs, detailing what Kiner has learned in a decade of reorienting her work to support the life and the company she wanted to create.

“I didn’t want to give the impression that owning a business is easy,” Kiner said in a recent conversation about the book on the porch of a Seattle coffee shop. “You can still be tired, you can still be overworked, you can still be drained, and you can still struggle.”

At the same time, she wanted to convey the fun and joy that comes from the freedom of doing your own thing. Kiner spent 15 years in HR leadership at companies including Microsoft, Amazon, and Starbucks, often working 60 to 80 hours a week, before starting Reverb in 2015.

“I made a choice to try and do something different,” she said. “And I’m so happy I did. Really, really happy. The key words there being made a choice.”

“The Reverb Way” is her second book, following “Female Firebrands” in 2020.

The new book is part memoir and part leadership guide. It draws on Kiner’s corporate career and her decade running Reverb to offer insights on everything from hiring and delegation to performance management and company values, and the daily mechanics of productivity and protecting your time.

Practical takeaways

Here are some of the insights from the book that resonated with me: 

Park your ideas. Instead of chasing every good idea the moment it came up, Kiner started logging them in a “Future Goals spreadsheet” and reviewing the list during quarterly business reviews. Some items got done as part of other initiatives. Others became irrelevant. But the team stopped getting pulled in a dozen directions at once.

Use your freedom. Reverb takes Fridays off from Memorial Day through Labor Day, with one person on call to check email a few times in case clients need help.

When Kiner offered to go further and adopt a formal four-day work week, the team turned her down. They already had the flexibility they needed. One employee had been going to a rock climbing gym at 3 p.m. every day, and Kiner never knew, because the work was getting done.

Don’t apologize for your schedule. Kiner writes about watching male executives cancel meetings for their kids’ soccer games without explanation or apology, and realizing she’d been justifying every time she was unavailable. Her rule now: no meetings before 9 or after 5, and no explanation necessary.

Build your own community. After being rejected from a business accelerator — possibly, she suspects, because she’d listed family time as a personal value — Kiner created her own informal group of women CEOs called WISE. They meet quarterly, share business insights, and support each other. Some are direct competitors. Friendship comes first.

Celebrate more than you think you need to. Kiner describes herself as a recovering perfectionist who used to hesitate to praise someone doing one thing well if they were struggling in another area.

For leaders who struggle with this, she suggests a simple tracking method: write down your team members’ names and add a checkmark each time you recognize them. Her point: everyone needs to hear they’re on the right track, probably more often than you think.

In that spirit, while the book is about Kiner’s experience, it also puts a big focus on the team that makes Reverb work, including co-owner and COO Sarah Wilkins, whom Kiner describes as the person who kept the company running during the worst stretches.

What’s happening now

As candid as the book is about the downturn, things have shifted since Kiner finished writing. In the weeks before our recent conversation, she said, new deal volume had jumped 50%, across tech, nonprofits, and small businesses. Reverb is hiring consultants again.

“I literally can’t explain it,” she said, noting that the turnaround has been happening despite inflation, gas prices, and geopolitical turmoil such as the war in Iran. 

AI is a frequent backdrop and topic of conversation in their work. Kiner writes in the book, for example, about teams at some companies being told to double productivity with AI but getting little support. 

In our conversation, she described a split: companies using AI as a way to demand more, and those actually bringing people along, showing them how to save time.

She’s not worried about AI replacing the human side of her work. One of her advisors uses a term she likes: “connective labor,” referring to empathy, conflict resolution, and the work of helping people and teams get unstuck. That part, she said, isn’t going away.

“I think there’s room for all of us,” she said. “Us and the agents, too.”

“The Reverb Way” is available in paperback and e-book versions.

Editor’s Note: GeekWire is a Reverb client.

Amazon added a new member to its senior leadership team Wednesday, naming AWS infrastructure chief Prasad Kalyanaraman to the group known as the S-team or “steam,” while also promoting cloud computing and AI services leader Dave Brown to senior vice president.

CEO Andy Jassy announced the changes internally, according to a memo viewed by GeekWire, and the company updated its public list of S-team members to reflect the changes.

Kalyanaraman oversees AWS infrastructure, including data centers, networking, and supply chain. He has been with the company for more than 20 years, starting in Amazon’s fulfillment and supply chain operations before moving to the cloud division in 2012.

Jassy’s memo praised his “customer obsession, high standards, ability to be right often, delivery, and missionary approach (always focusing on what’s best for customers — and the company as a whole vs. just his own area),” alluding in part to Amazon’s leadership principles. 

Brown leads AWS compute services (EC2) along with fast-growing AI services including Bedrock and SageMaker. He has been on the S-team since 2023, previously as a vice president.

“There are several reasons for his promotion, but chief among them are his outstanding delivery, propensity to look around corners and deliver services customers want, being right a lot, obsessing about customers, and continuing to develop strong teams,” Jassy wrote.

The addition of Kalyanaraman brings the S-team back up to 28 members. That’s still down from more than 30 when the last big round of additions was made in September 2023. 

In the meantime, the group has seen departures including Adam Selipsky as AWS CEO (replaced by Matt Garman); longtime devices chief Dave Limp, (succeeded by former Microsoft executive Panos Panay); artificial intelligence leader Rohit Prasad; grocery head Tony Hoggett; and device software leader Rob Williams. 

Here’s the full list as it stands now.

• Aicha Evans, CEO, Zoox
• Amit Agarwal, SVP, Emerging Markets and Selling Partner Services
• Andy Jassy, President and CEO
• Beth Galetti, SVP, People eXperience and Technology
• Brian Olsavsky, SVP and CFO
• Candi Castleberry, VP, Amazon eXperience and Upskilling
• Christine Beauchamp, SVP, North America Stores
• Colleen Aubrey, SVP, AWS Applied AI Solutions
• Dave Treadwell, SVP, eCommerce Foundation
• David Brown, SVP, AWS Compute & ML Services
• David Zapolsky, Chief Global Affairs & Legal Officer
• Doug Herrington, CEO, Worldwide Amazon Stores
• Drew Herdener, SVP, Communications & Corporate Responsibility
• James Hamilton, SVP and Distinguished Engineer, Amazon
• John Felton, SVP, AWS CFO
• Matt Garman, CEO, Amazon Web Services
• Mike Hopkins, SVP, Amazon Video and Studios
• Neil Lindsay, SVP, Amazon Health Services
• Panos Panay, SVP, Devices & Services
• Paul Kotas, SVP, Advertising, IMDb, and Grand Challenge
• Peter DeSantis, SVP, Foundational AI Models, Custom Silicon, Quantum Computing
• Peter Krawiec, SVP, Worldwide Corporate and Business Development
• Prasad Kalyanaraman, VP, AWS Infrastructure Services
• Russell Grandinetti, SVP, International Stores
• Steve Boom, VP, Audio, Twitch, and Games
• Steve Schmidt, Chief Security Officer
• Swami Sivasubramanian, VP, Agentic AI
• Udit Madan, SVP, Worldwide Operations

This week on the GeekWire Podcast, a week of Seattle-area startup news shows how the AI era is reshaping the regional tech scene. Q1 venture numbers reveal bigger checks going to fewer companies, with Seattle slipping behind the likes of Austin and Miami on deal volume.

And yet the distributed nature of modern startups is complicating what it even means to be a regional tech hub. (Does a mailbox in Pioneer Square really count as a Seattle headquarters?)

Founders and CEOs are navigating this in different ways. Those with enough cash are eyeing strategic acquisitions, including opportunities to absorb startups caught up in the AI shakeout.

Many are also rethinking how they hire and expand. More than a third of the GeekWire 200, our ranking of top Pacific Northwest startups, saw year-over-year employment declines, as agents boost individual productivity and reshape the workforce.

Plus: Andy Jassy’s shareholder letter signals Amazon is making bets again, in areas including chips and robotics. Driving home the point, the tech giant’s Amazon’s ambitious Globalstar acquisition effectively means it’s inheriting Apple’s satellite roadmap.

Of course, we have to talk about Allbirds. The sustainable shoe brand, which once challenged Amazon over knock-off sneakers, pivoted to AI infrastructure and saw its stock soar.

And in our final segment, a trivia challenge on the No. 1 companies in GeekWire 200 history.

With GeekWire co-founders John Cook and Todd Bishop. Edited by Curt Milton. 

Subscribe to GeekWire in Apple Podcasts, Spotify, or wherever you listen.

Amazon paid about $1.8 billion last year to Blue Origin, the space company owned by its founder and board chair Jeff Bezos — nearly triple the amount the year before — as the tech giant prepared to ramp up deployment of its own low-Earth orbit satellite constellation. 

The increase comes as shareholders weigh a proposal calling for a mandatory independent board chair, citing Bezos’ business interests outside Amazon as potential conflicts of interest. 

Bezos stepped down as Amazon’s CEO in 2021 but remains executive chairman.

According to the filing, the company paid approximately $2.2 billion total under satellite launch agreements during the past fiscal year, with an estimated $1.8 billion going to Blue Origin. The prior year’s proxy showed Blue Origin receiving about $578 million out of $1.7 billion total. 

Amazon is building a constellation of 3,236 low-Earth orbit satellites under the Amazon Leo program, formerly known as Project Kuiper, to beam broadband internet to consumers and businesses. The company has deployed 243 satellites so far and has asked the FCC for a two-year extension on a July deadline to launch roughly half of the fleet. 

The company this week also announced a $10.8 billion deal to acquire Globalstar, a satellite operator that has used SpaceX as its primary launch provider. 

Blue Origin’s New Glenn rocket made its debut flight in January 2025 but has not yet reached the launch cadence needed for the rollout. In addition to Blue Origin, Amazon has launch agreements in place with United Launch Alliance and Arianespace, and has also tapped Blue Origin rival SpaceX’s Falcon 9 for some launches, as Reuters reported this week. 

Bezos is also co-founder and co-CEO of AI startup Project Prometheus, a venture focused on applying AI to manufacturing and engineering across a variety of commercial sectors. 

The shareholder proposal calling for a mandatory independent chair, submitted by the AFL-CIO Reserve Fund, points to Bezos’ expanding role outside Amazon as cause for concern. 

“As a technology company, Project Prometheus could be a potential competitor or a business partner with our Company, raising potential conflicts of interest,” the proposal states, also citing Amazon’s multibillion-dollar launch agreements with Blue Origin as a potential conflict.

It notes that Amazon also has done business with the Bezos-owned Washington Post.

Amazon’s board recommends voting against the proposal, arguing that its lead independent director structure provides sufficient oversight. The role is currently held by Jamie Gorelick, a former U.S. Deputy Attorney General. The company’s annual meeting is set for May 20. 

The Blue Origin contracts have drawn scrutiny before. A shareholder lawsuit filed in 2023 alleged Amazon’s board spent less than 40 minutes approving the launch agreements without considering SpaceX as an alternative. Delaware’s Court of Chancery dismissed the case, and the state Supreme Court affirmed that ruling in November 2025.

Keith Alexander, a retired four-star Army general and former director of the National Security Agency, is leaving Amazon’s board of directors after more than five years.

Alexander, 74, informed the company April 7 that he wouldn’t stand for re-election at its annual meeting next month, according to the company’s proxy statement. 

“We’re grateful to General Alexander for his service on our Board since 2020 and for the many contributions he’s made to our company, and we wish him every success in the future,” a spokesperson said in a statement, responding to GeekWire’s inquiry.

No reason was given for his departure. Amazon’s board, which has fluctuated by one or two directors over time, will consist of 11 people after his departure.

Alexander joined Amazon’s board in September 2020, when Jeff Bezos was still CEO and the company was navigating a massive surge in demand during the early days of the pandemic. He previously chaired the board’s Security Committee, which oversees Amazon’s cybersecurity policies, data protection compliance, and response to significant cyber incidents.

Alexander served as commander of U.S. Cyber Command and led the NSA from 2005 to 2014, a tenure that included the surveillance disclosures of former NSA contractor Edward Snowden.

After retiring from the military, Alexander founded IronNet, a cybersecurity company, serving as CEO and president from 2014 to July 2023 and as board chair until February 2024. 

With his departure, eleven members of the board are up for re-election.

• Jeff Bezos, founder and executive chair
• Andy Jassy, president and CEO
• Edith W. Cooper, co-founder of Medley Living and former EVP of Goldman Sachs
• Jamie S. Gorelick, lead independent director; senior counsel at WilmerHale
• Daniel P. Huttenlocher, dean of MIT Schwarzman College of Computing
• Andrew Y. Ng, managing general partner of AI Fund; founder of DeepLearning.AI
• Indra K. Nooyi, former chair and CEO of PepsiCo
• Jonathan J. Rubinstein, former co-CEO of Bridgewater Associates
• Brad D. Smith, president of Marshall University; former CEO of Intuit
• Patricia Q. Stonesifer, former president and CEO of Martha’s Table
• Wendell P. Weeks, chairman, CEO, and president of Corning

Amazon’s annual shareholder meeting will be held virtually May 20.

Amazon isn’t just buying Globalstar — it’s inheriting Apple’s satellite roadmap.

The Seattle-based company’s agreement to acquire the satellite operator behind Apple’s iPhone Emergency SOS feature promises to give it a new constellation of operating satellites, a key slice of mobile spectrum, and Apple as a flagship partner.

The cash-and-stock deal, announced Tuesday, will help the Amazon Leo satellite broadband business press fast-forward in its attempt to catch up with Elon Musk’s Starlink.

Amazon told GeekWire the transaction was valued at approximately $10.8 billion as of April 9, when the exchange ratio was fixed. This differs from the slightly higher figure reported by multiple news outlets Tuesday. The value will fluctuate with Amazon’s share price until closing, capped at $90 worth of Amazon stock per Globalstar share.

Apple is already Globalstar’s biggest customer. In 2024, it committed about $1.5 billion to the company — a combination of prepayments for satellite services and a 20% equity stake in a Globalstar subsidiary — in exchange for the right to most of its network capacity.

Under a separate long-term agreement announced along with the deal, Amazon Leo will power satellite features on future iPhone and Apple Watch models, including Emergency SOS, messaging, Find My location sharing, and roadside assistance. Amazon will also continue supporting the Apple devices that already rely on Globalstar’s existing network.

Amazon renamed its satellite venture from Project Kuiper to Amazon Leo in November, a move the company framed at the time as a key step toward commercial service. 

On Monday, a day before the Globalstar announcement, Amazon Leo unveiled a new aviation antenna capable of delivering gigabit download speeds to aircraft, part of the build-out for its Delta and JetBlue in-flight Wi-Fi deals.

The acquisition agreement values Globalstar at $90 per share in cash or stock and is expected to close in 2027, pending regulatory approval. Thermo Funding, which controls 57.6% of Globalstar, has already agreed to the deal, according to an SEC filing, meaning no shareholder vote is required.

Globalstar, based in Covington, La., currently operates about two dozen satellites in low Earth orbit. It is in the middle of a major expansion, backed by Apple, that will grow the fleet to 54 satellites. It also holds licensed mobile-satellite spectrum — a scarce and tightly regulated asset that is difficult for newer entrants like Amazon to acquire.

Starlink operates about 10,000 satellites and serves more than 9 million subscribers. Amazon has launched about 200 satellites and has yet to begin consumer service. 

OpenAI’s partnership with Microsoft helped launch the generative AI era. Now Amazon is bringing the ChatGPT maker further into the booming market for enterprise AI.

That’s the takeaway from an internal memo distributed over the weekend by Denise Dresser, OpenAI’s chief revenue officer. The memo touted the Amazon Web Services alliance as a key enterprise growth driver for the ChatGPT maker, according to a CNBC report.

“Our Microsoft partnership has been foundational to our success. But it has also limited our ability to meet enterprises where they are — for many that’s Bedrock,” Dresser wrote, referring to the AWS AI model platform. “Since we announced the partnership at the end of February, inbound demand from our customers for this offering has been frankly staggering.”

It’s a high-profile endorsement for AWS, the leading cloud platform by market share, and perhaps Amazon’s best answer yet to the persistent perception that it was caught flat-footed by the generative AI boom sparked by the launch of ChatGPT in November 2022.

The OpenAI-Amazon relationship has come full circle. As GeekWire previously reported, AWS was actually OpenAI’s first cloud partner, providing computing resources at the lab’s founding in 2015, before Microsoft swooped in.

A decade later, Amazon and OpenAI in February struck a $50 billion investment and a cloud deal worth more than $100 billion over eight years.

Microsoft’s deal with OpenAI remains intact. Azure is still the exclusive host for OpenAI’s core APIs, and Microsoft retains its intellectual property license and revenue-sharing arrangement, including a share of revenue from OpenAI’s partnerships with other cloud providers.

Dresser’s memo also took aim at Anthropic, whose Claude model has emerged as the popular favorite in enterprise applications and software development. She said Anthropic made a “strategic misstep to not acquire enough compute,” echoing a separate OpenAI memo to investors that characterized Anthropic as “operating on a meaningfully smaller curve.”

Amazon also has a major investment in Anthropic, having committed $8 billion to the Claude maker. Both Seattle-area tech giants now hold stakes in the two leading AI labs, showing how quickly the era of exclusive AI partnerships has given way to a messier set of alliances.

Andy Jassy’s new letter to Amazon shareholders is a data-heavy defense of the tech giant’s biggest bets — from AI and custom chips to satellite internet and 20-minute delivery.

In the process, the Amazon CEO discloses that AI revenue for AWS has hit a $15 billion annual run rate, that Amazon’s internal chips business is generating over $20 billion a year, and that two large customers asked to buy all of Amazon’s available Graviton chip capacity for 2026.

Amazon said no, but Jassy says it gives a sense for the demand.

“We’re not investing approximately $200 billion in capex in 2026 on a hunch,” he writes.

That is basically the thesis statement for this year’s letter, released Thursday morning. It continues a tradition that stretches back nearly three decades, to Amazon founder Jeff Bezos’s first shareholder letter in 1997, which introduced the world to the “Day 1” mindset and is appended to the latest letter every year in an attempt to show its enduring relevance.

The evolution: This is Jassy’s fifth installment since succeeding Bezos as CEO in 2021. His letters have gone from establishing his management philosophy and navigating a post-pandemic cost hangover to laying out the frameworks Amazon uses to invent and build. 

This year he touches on progress in businesses including grocery (where Amazon says it’s now the second-largest U.S. grocer), satellite broadband (Amazon Leo is set to launch commercially in mid-2026), Amazon Now delivery (expanding from India to the U.S. and Europe), Alexa+, and Zoox, its autonomous ride-hailing service now starting commercial service.

His overarching message: progress won’t be a straight line (here, Jassy’s letter riffs on the title of an album by the New Zealand indie rock band The Beths) but Amazon is placing big bets on many fronts simultaneously, as it has throughout its history, and the results will come. 

The unstated plea: have patience, folks, we’ve been here before, and look how it turned out.

The AI bet: Nowhere is this appeal more important than in AI, given investor concerns about the massive investments being made across the industry. Throughout the letter, Jassy makes the case that the company’s huge capital outlays are backed by real demand and realistic economics.

Jassy readily acknowledges that Amazon’s free cash flow (FCF) dropped from $38 billion to $11 billion last year, driven by a $50.7 billion increase in capital spending, primarily on AI infrastructure. That was despite revenue overall growing 12% from $638 billion to $717 billion last year.

“AI is a once-in-a-lifetime opportunity where the current growth is unprecedented and the future growth even bigger,” he writes, adding later, “We’re not going to be conservative in how we play this—we’re investing to be the meaningful leader, and our future business, operating income, and FCF will be much larger because of it.”

AWS AI revenue: The disclosure about the annual revenue run rate for AI in AWS ($15 billion as of Q1 2026) is a preview of sorts of Amazon’s upcoming quarterly earnings, which won’t be reported for a few weeks. Amazon hasn’t previously reported this type of AI revenue figure. 

To put the figure in context, AWS overall had a $142 billion dollar revenue run rate as of Q4 25.

“We have never seen a technology more quickly adopted than AI,” he writes, adding later, “Amazon is smack in the middle of this land rush, and companies are choosing AWS for AI.”

Jassy compares the current AI wave to the early days of AWS, noting that three years after AWS launched commercially the cloud division had a $58 million revenue run rate. Three years after generative AI took off, the company’s AI business is nearly 260 times greater than that.

Future external businesses: Jassy points to two areas where Amazon may open up internal capabilities to outside customers.

• It’s “quite possible” Amazon will sell racks of its internally developed chips to third parties in the future, he writes.
• In addition, he says, Amazon will explore building and selling its robotics solutions to other industrial and consumer customers. 

Both follow the Amazon playbook of building something internally, then offering it as an external service — the same pattern behind businesses such as AWS and Fulfillment by Amazon, the company’s logistics services for third-party sellers.

Chips economics: Jassy says Amazon currently only monetizes its custom chips through its own EC2 cloud service, but if the chips business were standalone and sold externally like Nvidia, the annual run rate would be roughly $50 billion.

He projects that at scale, Trainium will save “tens of billions of capex dollars per year” and provide “several hundred basis points of operating margin advantage” versus relying on third-party chips for inference.

Trainium2 has largely sold out. Trainium3, which started shipping in early 2026, is nearly fully subscribed. Trainium4, still about 18 months from broad availability, has already been significantly reserved.

“Our chips business is on fire,” Jassy writes, noting that the business “changes the economics for AWS, and will be much larger than most think.”

Across Amazon’s business, he writes, “It’s hard to overstate my optimism for what’s ahead.”

Read the full letter here.

Amazon Web Services is making it possible to access data stored in its S3 cloud storage service as a traditional file system, bridging a divide between two types of storage that has frustrated developers and data scientists for nearly two decades.

The new capability, called Amazon S3 Files, lets applications running on AWS access an S3 storage bucket as if it were a local file system, reading and writing data using standard file operations rather than specialized cloud storage commands.

In practice, this means a machine learning team can run a training job directly against data in S3 without first copying it to a separate file system. Or, perhaps more importantly these days, an AI agent can read and write files in S3 using the same basic tools it would use on a local hard drive.

S3, launched 20 years ago, holds a huge amount of the world’s cloud data. S3 Files promises to open the door for a much broader range of apps and AI systems to work directly with that data.

The backstory: In an unusually candid essay coinciding with the news, Andy Warfield, a vice president and distinguished engineer who leads S3 engineering at AWS, described the technical and philosophical challenges of making the feature work, and why the first approach failed.

The core issue, Warfield wrote, is that files and objects are fundamentally different. 

Files can be edited in place and shared across applications in real time, working the way most software has always expected. Objects in S3 work differently: they are designed to be stored and retrieved as complete units, and millions of applications are built around that assumption. 

So they “did the only sensible thing you can do when you are faced with a really difficult technical design problem: we locked a bunch of our most senior engineers in a room and not let them out till they had a plan that they all liked,” Warfield wrote.

“Passionate and contentious discussions ensued,” he said. “And then finally we gave up.”

But ultimately, the team found its answer by no longer trying to hide the boundary between files and objects and instead making it a deliberate part of the design. 

The approach: S3 Files uses a “stage and commit” model, borrowing the concept from version control systems like Git: changes accumulate on the file system side and are pushed back to S3 as whole objects, preserving the guarantees that existing S3 applications depend on.

Google and Microsoft offer their own tools for accessing cloud object storage through file system interfaces, but AWS is positioning S3 Files as a deeper integration, backed by a fully managed file system rather than a simple adapter.

S3 Files is available today in AWS regions worldwide, built on Amazon’s Elastic File System. The company says it has been in customer testing for about nine months.

A trio of YouTube producers filed a class action lawsuit against Amazon alleging the tech giant illegally used content from the video platform to train and improve its Nova Reel generative AI model.

The suit, filed Friday in U.S. District Court for the Western District of Washington in Seattle, describes how Amazon allegedly used datasets earmarked only for academic use, circumvented YouTube’s copyright protection measures, and scraped video content. KING5 first reported on the suit.

“In a world where Defendant and others can circumvent technological protections to exploit copyrighted works without authorization with impunity, creators will be less likely to make their creations available on YouTube and other similar platforms, for fear of losing all control of them,” the plaintiffs state in their suit. “The world will be poorer for it.”

Plaintiffs are seeking damages, restitution and injunctive relief, claiming Amazon violated the Digital Millennium Copyright Act.

An Amazon spokesperson declined to comment on the matter, citing ongoing litigation.

Amazon released its Nova foundation models in 2024 via AWS Bedrock. The Nova Reel model can take text prompts and images and turn them into short videos, with features including watermarking.

According to the suit, Amazon deployed automated download tools paired with virtual machines that cycled through IP addresses to avoid being blocked, enabling the unauthorized extraction of data from millions of videos.

The named plaintiffs include:

• Ted Entertainment, Inc. (TEI), a California-based media company owned by Ethan and Hila Klein with more than 5,800 videos on YouTube with a combined total of more than 4 billion views. TEI channels include h3h3 Productions and H3 Podcast Highlights.
• Matt Fisher, a California-based YouTuber who runs the MrShortGame Golf channel that provides instructional videos and has more than 500,000 subscribers.
• Golfholics, a golf-focused YouTube channel with more than 130,000 subscribers and millions of views.

The suit argues the plaintiffs have no way to recover intellectual property already used to train Amazon’s models. “Once AI ingests content, that content is stored in its neural network and not capable of deletion or retraction,” it states.

Dozens of similar cases are working their way through courts nationwide. Among them: the New York Times’ lawsuit against OpenAI and Microsoft, a class action by authors against Microsoft, and a suit from musicians with YouTube content against Google.

Separate lawsuits against Anthropic and music-generation startup Suno over the alleged unauthorized use of books and music in AI training have since settled. A case brought by authors against Meta was dismissed.

For a while now, since the closure of the Amazon Fresh Pickup in Seattle, I’ve been complaining about having to drive across the Ballard Bridge to Whole Foods to do my Amazon returns. 

So when news emerged that FedEx Office locations are now part of Amazon’s drop-off network, I jumped at the chance to try it. Turns out there’s a FedEx Office on the way to GeekWire HQ, near the PCC on NW 46th Street (across from the “Up” house in the Ballard Blocks complex).

I walked in with a microchip pet door (long story), showed the QR code on my phone, got it scanned, handed over the unpackaged item, and walked out with a receipt. No box, tape, or label required, just as with other drop-off locations. There was no line.

The refund hit my account the same day.

The one thing that made me scratch my head is that, unlike returning something at a Kohl’s or Whole Foods, there’s no real ancillary benefit for FedEx Office. I dropped off the package and there was nothing else to do in the store. I had no copies to make, nothing to ship, and no need for any of the miscellaneous supplies in their limited displays.

However, I was in need of traditional U.S. Postal Service stamps, so I asked if they sold them, and the guy looked at me like I was a complete idiot. Fair enough.

But for pure convenience, it seems like a win for Amazon customers. 

Amazon and FedEx severed their logistics relationship back in 2019 as Amazon built out its own delivery network. Now they’re patching things up, and more than 1,500 FedEx Office locations are accepting returns as part of a network of over 10,000 drop-off points nationwide. 

We discussed this (and much more) on this week’s GeekWire Podcast. Listen above, and subscribe to GeekWire in Apple Podcasts, Spotify, or wherever you listen.