Amazon posted 1,193 AI-specific jobs in the Seattle metro on Glassdoor as of early June 2026 — against a backdrop of 30,000 corporate layoffs that began in October 2025. The company simultaneously committed $200 billion in capital expenditure this year, the majority directed at AI data centers and infrastructure. What that math reveals is not contradiction but architecture: Amazon is deliberately dismantling one workforce and building a different one, and Seattle is where the new version is being assembled.
The restructuring accelerated in December 2025, when Andy Jassy announced that Rohit Prasad — the SVP who led Amazon's AGI group for two years and spent twelve years building Alexa — would depart. Peter DeSantis, a 27-year Amazon veteran and cloud infrastructure SVP, stepped in to run a newly consolidated division that merges AGI model development, the Trainium and Graviton silicon teams, and quantum computing under a single org. Pieter Abbeel, who joined Amazon in 2024 when it acquired robotics startup Covariant, now leads the frontier model research team within that structure. The reorganization folded the AGI SF Lab — a separately operated research operation that had been recruiting from a "few dozen" carefully selected scientists — into the broader DeSantis org. What Amazon had previously run as parallel AI operations is now a unified stack with a single reporting line.
How Amazon Is Rebuilding Its Seattle AI Workforce in 2026
The hiring math begins with the 11,000 figure. At the AWS What's Next event earlier in 2026, AWS CEO Matt Garman announced Amazon would hire approximately 11,000 software engineering interns across the company in 2026 — in line with recent prior years, and even as it continued eliminating corporate headcount. "We're hiring just as many software developers as we ever had inside of Amazon," Garman said, "and demand for that talent is really accelerating." The 11,000 figure specifically describes the intern pipeline, which at Amazon historically converts substantially into full-time offers; it is not a net new full-time hire count. Per coverage in People Matters and Storyboard18, the emphasis is on engineers and scientists capable of working across AI, cloud infrastructure, and large-scale systems. It is not a campus hiring surge — it is a targeted rebuild of the technical layer Amazon cut out from legacy product lines.
The Alexa+ side of the equation is numerically significant on its own. Amazon made Alexa+ available to all US users in February 2026, roughly a year after its initial rollout. The company auto-enrolled Prime subscribers — a base of over 200 million globally — in January 2026, pushing the Alexa+ user count past 1 million within its first months. The Alexa AI Engineering team in Seattle is actively hiring Software Development Engineers, Technical Program Managers, Applied Scientists for conversational AI, and Product Managers. Publicly posted salary ranges for those Alexa AI roles in Seattle run from $136,000 for a Language Data Scientist to $252,000 at the upper end of advertised compensation — but those figures are the posted floor, not the total package. Levels.fyi data for Amazon Machine Learning Engineers in the US shows median total compensation of $265,000, and L6 Software Engineering Manager packages in Seattle have been reported at a median of $414,000 total comp on Levels.fyi's Seattle dataset (the $414K figure specifically applies to the L6 SDM track; L6 SDE medians run lower).
On the AGI side, the DeSantis restructuring is the story. The consolidation of AGI model work (Amazon Nova foundation models), custom silicon (Trainium 2 and Graviton), and quantum computing under one SVP is a structural signal that Amazon is treating its AI compute stack as a single integrated capability, not a collection of adjacent bets. Amazon Nova 2, the multimodal reasoning model, was published through Amazon Science in early 2026. The AGI SF Lab — which had been hiring specifically for research roles combining large language models with reinforcement learning — has now been absorbed into this broader group. Job postings across Amazon's AGI pages list open research scientist and applied scientist roles working directly on Nova model development, dataset design, and pre/post-training optimization, with Trainium infrastructure as the underlying compute context.
Q1 2026 earnings provided the financial frame. Amazon reported $181.5 billion in worldwide revenue, up 17% year-over-year, and $43.2 billion in capital expenditure in a single quarter — primarily AWS and generative AI infrastructure. AWS revenue reached $37.6 billion in Q1, growing 28% YoY, with an AI revenue run rate above $15 billion. The Trainium custom silicon business alone carries an annual revenue run rate over $20 billion, growing triple-digit percentages YoY. Against those numbers, the 11,000-hire commitment looks less like a moonshot and more like maintenance.
Why It Matters
Seattle's AI talent market is being pulled in three directions simultaneously, and Amazon is the largest gravitational force at the table.
The first pull is Amazon itself. With 998 total open positions in Seattle as of May 2026 and 1,193 AI-specific roles visible on Glassdoor in early June, Amazon is running the single largest concentrated AI hiring operation in the city. The company is paying for it: AI Engineer total compensation at Amazon in Seattle runs from $241,000 at the low end to $1.4 million at the top of the range, per 6figr's 2026 Seattle dataset. For an L6 Software Engineering Manager in the Greater Seattle Area, Levels.fyi reports a median total comp of $414,000 (L6 SDE medians sit below this figure). The Microsoft AI division, whose two-track hiring framework this publication covered on June 7, offers packages reportedly in the $700,000 to $1.4 million range for senior research engineers. Those two comp bands are now converging at the top — and candidates know it.
The second pull is Microsoft Copilot, which is recruiting from the same Seattle AI engineering pool. Both companies are competing for a specific profile: engineers who can work at the intersection of LLMs, retrieval systems, and agentic product design. Microsoft packages now top $330,000 for some AI engineering and research roles, per ComputerUser's reporting on H1 2026 data. Amazon's restructured AGI-plus-silicon org is countering not with comparable base salaries — historically Amazon has lagged Microsoft on base, compensating with equity — but with the DeSantis pitch: that the combination of AGI model work, Trainium silicon, and AWS scale creates a technical challenge that neither Copilot nor any other product org can replicate. Whether that pitch holds for senior researchers who could alternatively work at OpenAI or DeepMind is the open question.
The third pull is Project Prometheus, Jeff Bezos's stealth AI lab, which closed a $10 billion funding round at a $38 billion valuation in April 2026. Prometheus recruited from OpenAI, Meta, Anthropic, xAI, Nvidia, and Google DeepMind to build a team of approximately 120. Its geographic footprint — San Francisco, London, and Zürich — does not directly compete with Seattle's AI talent pool, but the existence of a Bezos-backed AI lab at $38 billion means the broader market for senior AI researchers has a new benchmark that reshapes how every other organization in his orbit positions compensation. Amazon cannot be indifferent to that signal.
The Seattle City Council tension adds a public dimension that is unusual for a talent briefing to track. On June 3, Amazon software engineers — including Patrick Schloesser, a software engineer at AWS — testified at a Seattle City Council hearing in support of a proposed moratorium on new mega data center development within city limits. Council passed the one-year moratorium unanimously. "Amazon is spending $200 billion on capital this year, with most of it going to data centers and AI," Schloesser said at the June 3, 2026 hearing. "You've got to provide good jobs building these things, and you've got to pay a new tax that funds city jobs every time you conduct a large layoff." Amazon's response was to note it has no current plans to build data centers within Seattle city limits. The exchange is a recruiting signal in itself: the engineering workforce Amazon is trying to attract is the same one asking publicly why 30,000 jobs were cut while the capital budget tripled.
What's Next
Three things to watch in H2 2026.
The DeSantis integration speed. Peter DeSantis has inherited a newly unified org that covers AGI model development, Trainium silicon, and quantum computing — three previously separate hiring pipelines with different recruiting profiles, different comp bands, and different technical cultures. The speed at which DeSantis consolidates those pipelines into a coherent recruiting pitch will determine whether Amazon's H2 hiring targets are met. Pieter Abbeel's frontier model research team is the most talent-constrained node: it requires researchers who can work at the intersection of LLMs and reinforcement learning, and the competition for that profile — from OpenAI, Google DeepMind, and Anthropic — is the most acute anywhere in the market.
Alexa+ subscriber conversion. Amazon auto-enrolled Prime members in Alexa+ in January 2026, but converting passive auto-enrollment into active daily use is the product problem that will drive or retard the Alexa AI Engineering hiring plan. If Alexa+ engagement metrics disappoint through Q2 and Q3, the Alexa AI team in Seattle is a cost center under pressure. If Alexa+ retention holds — particularly given that customers report talking to Alexa twice as much and for longer durations since the upgrade — it justifies the double-digit SDE headcount expansion that the current job posting volume suggests. Amazon's Q2 2026 earnings call, expected late July, will be the first moment any quantified Alexa+ engagement data enters the public record.
The Seattle comp floor. Amazon's L6 Software Engineering Manager total comp in Seattle runs at approximately $414,000 median, per Levels.fyi. Microsoft AI packages for senior research engineers sit at $700,000 to $1.4 million. The gap between those two figures — which did not look like this in 2021 — is the single most measurable indicator of whether Amazon is winning or losing Seattle's AI talent contest. If Amazon raises its L6 and L7 AI-specialist bands materially before year-end, it will signal that DeSantis's consolidation pitch is not closing candidates on its own. If the gap holds, Amazon is betting that scale, silicon, and AGI scope compensate for below-frontier comp. That bet has been wrong before.
Amazon is not losing Seattle's AI talent market. But for the first time since the 2021 comp surge, it is not obviously winning it either.