Amazon (AMZN +1.25%) Web Services, Microsoft (MSFT +1.62%) Azure, and Alphabet's (GOOG +0.34%) (GOOGL +0.23%) Google Cloud dominate the cloud computing infrastructure market. But artificial intelligence (AI) marks a paradigm shift in the industry. Established players and newcomers alike needed to redesign and build new data centers with high-powered processors and networking hardware, while also navigating AI energy bottlenecks.
The opportunity is coming at a steep price, with Amazon, Alphabet, and Microsoft spending record capital expenditures (capex) on AI data centers. But the prospect of accelerated growth has investors generally optimistic.
Alphabet and Amazon stocks are up 23.1% and 16.4%, respectively, year to date as of market close on May 1. But Microsoft stock is down 14% year to date despite reporting record earnings on April 29 for its third quarter of fiscal 2026.
Here's why Microsoft is out of favor and if the "Magnificent Seven" stock is worth buying anyway.
Image source: Getty Images.
The custom chip race is intensifying
In Microsoft's latest quarter, capex surged to $31.9 billion, two-thirds of which was allocated to short-lived assets, primarily graphics processing units (GPUs) and central processing units (CPUs). For context, Microsoft's capex in fiscal 2021 was $20.6 billion. This means that Microsoft is now spending roughly the same on AI chips in a single quarter as its entire annual capex budget from five years ago. And spending is only going up, as Microsoft is guiding for $40 billion in fiscal 2026's Q4 capex and a jaw-dropping $190 billion for calendar year 2026.
Hyperscalers are designing custom AI accelerators to reduce dependence on Nvidia. Broadcom has been vocal about the limitations of general-purpose GPUs -- advocating separate custom AI chips for training and inference.
In January, Microsoft introduced its new Maia 200 chip, specifically designed for AI inference. On the April 29 earnings call, Microsoft said that the Maia 200 is now live in two of its major data centers and is achieving a 30% improvement in tokens per dollar. Tokens are the currency of AI inferencing, so hyperscalers want to process as many tokens as possible to reduce latency and boost efficiency. Microsoft also said that its Cobalt server CPU is deployed in nearly half of its data center regions, and that its largest AI customers are increasingly running workloads on Cobalt.
These developments are a step in the right direction, but Microsoft is nowhere close to Alphabet's and Amazon's integrated AI architecture. Alphabet's Tensor Processing Units (TPUs) -- co-designed with Broadcom -- are helping scale models, agents, and enterprise applications for high-profile companies like Anthropic, the maker of Claude large language models.
The advancement of TPUs was on full display during Alphabet's April 29 earnings call, where Alphabet said that its TPU 8t for AI training delivered 3x the processing power and double the performance of the prior generation, while the TPU 8i for AI inferencing as 80% better performance per dollar than the prior generation. Alphabet also said it will begin delivering TPUs to a select group of customers in their own data centers, a notable step change from its prior strategy of renting out TPU capacity.
Similarly, Amazon's custom chips and specialized hardware are showing clear cost and efficiency advantages, dramatically reducing its reliance on Nvidia. Amazon CEO Andy Jassy said on Amazon's April 29 earnings call:
Our chips business continues to grow rapidly and is larger than what a lot of folks thought. We saw nearly 40% quarter-over-quarter growth in Q1, and, our annual revenue run rate is now over $20 billion, and growing triple digit percentages year over year. But, this somewhat masks the size. If our chips business was a stand-alone business, and sold chips produced this year to AWS and other third parties (as other leading chips companies do), our annual run rate would be ~$50 billion. As best as we can tell, our custom silicon business is now one of the top three data center chip businesses in the world. And, the speed at which we've gotten here is extraordinary.
Amazon is committing $200 billion in 2026 capex and noted on its April 29 earnings call that capex growth is outpacing revenue growth, which will strain free cash flow. However, Amazon compared this investment cycle to the early days of AWS and expressed confidence that this infrastructure expansion has much larger revenue and FCF potential than the first AWS build-out.
NASDAQ: MSFT
Key Data Points
Microsoft is a good value
Microsoft has more uncertainty than its cloud peers. Microsoft is generating record profits now, but higher spending will strain its near-term results. And Microsoft's AI tools are largely built around OpenAI's ChatGPT large language models, which are going through a period of rapid change now that the two companies have revised their once exclusive agreement.
Outside of the cloud, Microsoft's business is doing pretty well, but its More Personal Computing segment (Windows, Xbox, devices, etc.) reported a 1% revenue decline in its latest quarter.
By comparison, Amazon and Alphabet are showing a much clearer path to efficiency improvements by gaining greater control over their data center supply chains through custom chips, as well as the opportunity to monetize the sale of those chips. Amazon and Alphabet's non-cloud business segments are also arguably doing better than Microsoft. Amazon's online retail business is booming, and so is Alphabet's high-margin Google Search and YouTube, as well as Waymo's massive potential.
Microsoft trades at just 24.4 times forward earnings compared to 34.2 for Amazon and 34.9 for Alphabet. So investors can buy the stock at a discount compared to its peers. While the discount is somewhat justified, given that Microsoft is trailing Alphabet and Amazon in custom silicon, it's a mistake to assume Microsoft won't make a big push to unlock similar efficiency improvements.
That possibility, combined with favorable new terms from OpenAI, makes Microsoft an excellent buy for investors seeking an AI stock at a compelling valuation.
