Is Google about to eat Nvidia’s lunch?

This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.

Advances in the field of artificial intelligence (AI) are having a profound impact on the technology landscape. The ability of these cutting-edge algorithms to automate repetitive chores, streamline tasks, and generate original content is saving time, boosting productivity, and freeing users for higher-value work.

One of the undisputed beneficiaries of the AI revolution has been Nvidia (NASDAQ: NVDA). The company pioneered the graphics processing units (GPUs) that have become the gold standard for AI, providing the computational horsepower necessary to run these advanced generative AI systems. Demand for these chips has fueled a meteoric rise in Nvidia’s sales and profits, driving its stock price higher.

However, reports have emerged that Alphabet (NASDAQ: GOOGL) (NASDAQ: GOOG) wants a piece of the action and has fired a shot across Nvidia’s bow.

A shot across the bow

Reports emerged this week that Meta Platforms (NASDAQ: META) is in talks with Alphabet to deploy Google’s Tensor Processing Units (TPUs) to run AI models in its data centers as early as 2027. Google began designing these specialized processors in 2018 for use in its own cloud computing operations and has launched numerous upgrades to the TPUs in the ensuing years, which have since been adapted to facilitate AI. Google has never sold these processors, which have only been used in the company’s own data centers.

However, recent reports suggest Google is considering selling TPUs to Meta, which would mark a significant shift in the company’s strategy and could spell trouble for Nvidia if true.

Raw, number-crunching power, but at a price

Nvidia has been the primary beneficiary of the AI revolution. This is in part due to the mass appeal of its GPUs. Not only are these chips arguably the gold standard for AI processing, but they have a distinct advantage compared to many of Nvidia’s rivals. Years ago, the company developed its CUDA architecture, a library of software tools that allowed developers to harness the raw, number-crunching power of GPUs when applying them to their own computationally intensive applications.

This speedy processing comes at a cost, as the immense computational demands of AI tend to consume a great deal of energy in the process. On the other hand, Google’s TPUs were designed to be more specialized than GPUs, making them more energy-efficient. Until now, Google has kept these specialized chips to itself, but the company may be shifting gears as the AI revolution plays out.

Billions of dollars at stake

The fact that Google is considering a change to its strategy may simply come down to dollars and cents. Big tech companies have been shelling out billions of dollars on capital expenditures (capex) in order to position themselves to profit from the proliferation of AI — and that spending continues to escalate.

• Alphabet plans to spend $92 billion on capex in 2025, up from $52 billion in 2024.
• Amazon expects to spend $125 billion, up from $83 billion.
• Meta plans to spend roughly $71 billion, up from $37 billion.
• Microsoft is expected to spend $94 billion in fiscal 2026, up from $65 billion in fiscal 2025 (which ended in July).

The year isn’t over, so the numbers could still increase. All told, big tech is expected to spend as much as $405 billion on AI capex in 2025, which helps to illustrate just how high the stakes are. If Google could capture just a small percentage of that spending, it could boost its own results at the expense of Nvidia.

The fine print

While the news of Google’s possible strategy shift has huge implications, it’s important to put this in context.

While Nvidia doesn’t provide any specific details about its biggest customers, Wall Street has done some digging, and it is widely believed the list includes — you guessed it — Alphabet, Amazon, Meta, and Microsoft. Despite Alphabet developing and running its home-grown TPUs since 2018, it continues to be Nvidia’s biggest customer.

This means that Google still needs Nvidia GPUs, used in tandem with its own TPUs, to get the combination of speed and energy efficiency it needs to compete. It further suggests that, even if the reports are true, and Google’s power-miserly chips cut into Nvidia’s business, the company will still be the dominant player in the data center GPU space.

Current estimates put Nvidia’s market share at 92% of the data center GPU market, according to IoT Analytics. Even if Google succeeds in slicing off some share in this ever-growing market, Nvidia is well-positioned to continue to benefit from the secular tailwinds of AI.

This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.

The post Is Google about to eat Nvidia’s lunch? appeared first on The Motley Fool Australia.

This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.

This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.

More reading

• If you’d invested $100 in Nvidia 10 years ago, here’s how much you’d have today
• 3 ASX thematic ETFs that could boom over the next decade
• Will Alphabet be the world’s next $5 trillion stock?
• Why Nvidia could be a bigger winner in quantum computing than you might think
• Is Nvidia in an AI bubble? Here’s what Jensen Huang says.

Danny Vena, CPA has positions in Alphabet, Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool Australia’s parent company Motley Fool Holdings Inc. has positions in and has recommended Alphabet, Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool Australia’s parent company Motley Fool Holdings Inc. has recommended the following options: long January 2026 $395 calls on Microsoft and short January 2026 $405 calls on Microsoft. The Motley Fool Australia has recommended Alphabet, Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool has a disclosure policy. This article contains general investment advice only (under AFSL 400691). Authorised by Scott Phillips.