Amazon Web Services’ two-track approach to artificial intelligence came into sharper focus Tuesday as the world’s largest cloud moved closer to Nvidia in developing its own custom chips. At Amazon’s annual conference AWS Re:Invent 2025 in Las Vegas, Amazon Web Services CEO Matt Garman announced Trainium3, the latest version of the company’s in-house custom chip. It has 4x more computing performance, energy efficiency, and memory bandwidth than the previous generation. AWS announced that early results from customers who tested Trainium3 showed AI training and inference costs reduced by up to 50%. Custom chips like Trainium are becoming increasingly popular among large technology companies that can afford to manufacture them. And that use case is growing. For example, Google’s Tensor Processing Unit (TPU), co-designed with Broadcom, has also received a lot of attention since the launch of its well-received Gemini 3 artificial intelligence model last month. Equipped with TPU. There were even reports that Meta Platforms was considering TPUs in addition to Nvidia’s graphics processing units (GPUs), the gold standard for general-purpose AI workloads. At the same time, Amazon also announced that it is deepening its collaboration with Nvidia. In Tuesday’s keynote, Garman introduced AWS Factories, which provides on-premises AI infrastructure that customers can use in their own data centers. The service combines Trainium accelerators and Nvidia graphics processing units, giving customers access to Nvidia’s accelerated computing platform, full-stack AI software, and GPU-accelerated applications. By offering both options, Amazon aims to continue accelerating AWS cloud capacity and, in turn, revenue growth, and stay on top in the fierce competition with Microsoft’s Azure and Alphabet’s Google Cloud, which are vying for second and third place by revenue in the AI race. Earlier this year, investors were concerned that AWS’s second-quarter revenue growth wasn’t matching its closest competitors. Amazon went a long way toward allaying those concerns with its third-quarter earnings release in late October. “AWS is growing at a pace not seen since 2022, reaccelerating to 20.2% year over year,” Amazon CEO Andy Jassy said at the time. He added: “We are focused on accelerating capacity, adding more than 3.8 gigawatts (GW) in the past 12 months.” Tuesday’s announcement comes at a pivotal time for AWS, as the company looks to rapidly expand its computing capacity after a year of supply constraints that halted cloud growth. More efficient chips are good, but they can’t make up for the capacity demands businesses are facing as AI adoption increases. That’s why Wall Street is focused on adding more gigawatts of capacity. Fortunately, Wall Street insists that the capacity headwinds should turn into tailwinds. Wells Fargo said Trainium3 is “essential to complement Nvidia GPUs and CPUs in this capability build” to close the gap with rivals. In a note to investors on Monday, analysts predict that Amazon will add more than 12 gigawatts of compute by the end of 2027, increasing total AWS capacity and supporting up to $150 billion in annual AWS revenue growth if demand remains strong. In a separate memo, Oppenheimer said Monday that AWS has already proven its ability to increase capacity, which has already doubled starting in 2022. Amazon plans to double that number by 2027. Analysts said such expansion could add 14% to AWS’s revenue in 2026 and 22% in 2027. Analysts said each gigawatt of computing added in recent quarters was worth about $3 billion annually. cloud income. Conclusion While the new chips are welcome news that will help AWS move deeper into the AI chip race, investors are paying more attention to Amazon’s capacity investment and when that capacity will be released, as it is how it will meet demand. This issue is not a demand issue. It’s a supply issue. We are confident in AWS’s ability to add capacity. In fact, no company in the world is better able to handle these types of logistics problems at this scale than Amazon. Amazon shares rose nearly 14% to $254 each in two sessions following the cloud and e-commerce giant’s earnings release late Oct. 30. Stocks have since recouped those gains and then some. As of Tuesday’s close, the stock is up 6.5% since the beginning of the year, lagging among its Magnificent Seven peers and lagging the S&P 500’s roughly 16% gain in 2025. (The Jim Cramer Charitable Trust is long on NVDA’s AMZN. See the complete list of stocks here.) As a subscriber to Jim Cramer’s CNBC Investment Club, you will receive trade alerts before Jim makes a trade. After Jim sends a trade alert, he waits 45 minutes before buying or selling stocks in his charitable trust’s portfolio. If Jim talks about a stock on CNBC TV, he will issue a trade alert and then wait 72 hours before executing the trade. The above investment club information is subject to our Terms of Use and Privacy Policy, along with our disclaimer. No fiduciary duties or obligations exist or arise from your receipt of information provided in connection with the Investment Club. No specific results or benefits are guaranteed.
