Amazon has disclosed that its chips business now has an annual revenue run rate of more than $20 billion, giving investors one of the clearest signs yet that the company’s years of spending on custom silicon are becoming a meaningful commercial engine. The figure was revealed by CEO Andy Jassy in his annual shareholder letter and covers Amazon’s Graviton processors, Trainium AI chips and Nitro networking cards. Reuters reported that the run rate has doubled from the $10 billion figure the company previously disclosed alongside its fourth-quarter results.
The disclosure matters because it shows Amazon is no longer just building chips to lower internal costs or optimize AWS performance. It is now operating a semiconductor platform large enough to be discussed as a major business in its own right. That is especially important at a time when investors are scrutinizing whether Big Tech’s massive AI capital spending is producing tangible returns. This is an inference based on Amazon’s new revenue disclosure and the broader investor debate over AI capex.
Jassy Is Making the Case That AI Spending Is Paying Off
The chip milestone came alongside a broader defense of Amazon’s AI spending. Reuters reported that Jassy said Amazon’s AI services within AWS are now generating an annualized revenue run rate of more than $15 billion, the first time the company has publicly disclosed a figure for that business. Reuters said the number represents roughly 10% of AWS’s $142 billion revenue run rate.
Together, the AI services figure and the chip run rate are central to Amazon’s argument that its enormous infrastructure buildout is not speculative excess. Jassy’s message was effectively that customer demand is already visible, monetization is already happening and the spending is being backed by real commercial commitments. Reuters reported that Amazon shares rose 4.5% after the disclosures, suggesting investors took some reassurance from the update.
Why the Chip Number Is So Significant
The over-$20 billion figure stands out because Amazon’s chip portfolio has historically received less attention than AWS itself or the company’s retail empire. Yet Graviton, Trainium and Nitro have become increasingly important to Amazon’s AI and cloud strategy. Jassy said in his shareholder letter that the business is “on fire,” underscoring how central custom silicon has become to the company’s infrastructure ambitions.
Strategically, this gives Amazon a stronger position in a market where access to compute and control over hardware design have become major competitive advantages. Companies that can build their own chips can reduce dependence on third-party suppliers, improve cost-performance for internal workloads and potentially create new external sales opportunities. That interpretation is an inference from Amazon’s business model and Jassy’s comments, but the underlying revenue milestone is directly sourced.
Capex Remains Huge
Amazon’s spending plans are still enormous. Reuters reported in February that the company projected about $200 billion in capital expenditures in 2026, up from $131 billion in 2025, as it races to build more AI data centers and cloud infrastructure. That forecast initially alarmed investors and sent the stock down 11.5% in after-hours trading at the time.
What changed this week is that Amazon offered a clearer answer to the question that has hung over the stock for months: what is the company getting in return? The new disclosures suggest that AWS AI services and Amazon’s custom silicon business are already large enough to justify more confidence in the spending cycle. That does not mean the debate is over, but it does give Amazon a stronger factual basis for defending one of the biggest capex programs in corporate America. This is an inference based on the February capex reaction and the new April disclosures.
Thinking Beyond Internal Use
Another notable point from Reuters’ report is that Jassy suggested Amazon could eventually sell its chips to outside customers. Reuters said he drew a parallel with Google’s success using custom chips in external partnerships, including its supply relationship with Anthropic.
That possibility is strategically important because it would expand Amazon’s chip effort from an internal cloud advantage into a broader commercial semiconductor business. If Amazon ever scales third-party chip sales, investors may begin valuing the business less as a support function inside AWS and more as a standalone growth engine. That remains speculative for now, but Reuters reported that Jassy explicitly raised the possibility.
The Bigger Message for the AI Infrastructure Race
Amazon’s update also says something broader about the AI race. The battle is no longer only about models and software. It is increasingly about who can finance, build and monetize the underlying compute stack. Reuters reported that AWS still has capacity constraints and unserved demand even after adding 3.9 gigawatts of new power capacity in 2025, while Jassy said Amazon expects to double total power capacity by the end of 2027.
That means the company is trying to convince the market that its giant capex plans are not a warning sign, but a signal of durable demand. The combination of a $15 billion AI services run rate and a more than $20 billion chip run rate strengthens that case materially. It does not eliminate execution risk, but it makes the bull case on Amazon’s AI infrastructure business easier to quantify. This is an inference supported by the reported figures.
Conclusion
Amazon’s disclosure that its chips business now exceeds a $20 billion annual revenue run rate is a major milestone for the company’s AI and cloud strategy. Combined with a more than $15 billion AI services run rate inside AWS, the figures give CEO Andy Jassy stronger evidence that Amazon’s huge AI infrastructure spending is already producing real commercial returns. The company is still investing at an extraordinary pace, but the latest numbers suggest Amazon’s custom silicon push is no longer just a strategic bet. It is becoming a business large enough to shape how investors think about AWS, AI and Amazon’s long-term valuation.
FAQ
What did Amazon disclose about its chip business?
It said its chips business now has an annual revenue run rate of more than $20 billion, covering Graviton, Trainium and Nitro products.
Who revealed the figure?
CEO Andy Jassy disclosed it in his annual shareholder letter.
How does this relate to the company’s AI spending?
Reuters reported that Jassy used the disclosure alongside an AWS AI services run rate of more than $15 billion to argue that Amazon’s heavy AI capex is already generating returns.
How much capex is the comapany planning for 2026?
It is reported in February that projected are about $200 billion in capital expenditures for 2026, up from $131 billion in 2025.
Could Amazon sell its chips to outside customers?
Possibly. It is reported that Jassy suggested the company could eventually sell its chips externally, similar to how Google has pursued outside partnerships.
Disclaimer
This article is for informational and journalistic purposes only and does not constitute investment advice, financial advice or a recommendation to buy or sell any security. Large-cap technology stocks can react sharply to AI revenue disclosures, capital spending plans and changes in investor expectations around infrastructure returns.
