Google began using its first Tensor Processing Unit (TPU) in early 2015, several months before the company restructured to create a parent organization, Alphabet (GOOG 0.84%) (GOOGL 0.59%). For the next 11 years, Google kept TPUs to itself. The chips powered many applications hosted by Google Cloud, including Google’s own AI products.
While Google continues to use its TPUs internally, it’s no longer keeping them to itself. Alphabet CEO Sundar Pichai revealed during his company’s first-quarter earnings call last week that Google is selling AI chips for the first time. Could this surprising decision shake up the market that Nvidia (NVDA +0.04%) has dominated for years?
Image source: Getty Images.
Why Alphabet’s move is a big deal
Pichai explained in the Q1 call that there’s a simple reason why Alphabet is now selling TPUs: significant demand. He noted that “AI labs, capital markets firms, and high-performance computing applications” are driving demand for TPUs. As a result, Alphabet will deploy its TPUs to “a select group of customers in their own data centers.”
Importantly, Pichai also acknowledged that selling TPUs will expand Alphabet’s market opportunity. Alphabet CFO Anat Ashkenazi stated that the company will recognize a small percent of revenue from its TPU agreements later in 2026, with most revenue to be realized next year.
Investors shouldn’t have to worry that TPUs sales will negatively affect Google Cloud’s sizzling growth. The types of customers Pichai mentioned have specialized application needs that aren’t a good fit for cloud-based hosting due to data privacy or latency concerns.
TPU sales could even strengthen the investment case for buying Alphabet stock. The company will now be the most vertically integrated AI infrastructure play, with chips used internally and sold externally, a top-tier cloud platform, a powerful AI model in Gemini, and a myriad of AI-powered applications.
Today’s Change
(-0.59%) $-2.29
Current Price
$383.40
Key Data Points
Market Cap
$4.6T
Day’s Range
$379.80 – $387.34
52wk Range
$147.84 – $387.38
Volume
940K
Avg Vol
32M
Gross Margin
60.43%
Dividend Yield
0.22%
What’s the impact on Nvidia?
In December, D.A. Davidson analyst Gil Luria wrote to clients that Google’s TPUs could gain around 20% market share if Alphabet chose to sell them to other companies. Is Nvidia worried? Apparently not.
Nvidia CEO Jensen Huang recently said that Google’s TPUs don’t present a significant threat to his company’s GPU business. That’s in line with Nvidia’s post on X (formerly Twitter) last November in response to reports that Meta Platforms (META +0.28%) could use Google’s TPUs in its data centers: “Nvidia is a generation ahead of the industry — it’s the only platform that runs every AI model and does it everywhere computing is done.”
To be sure, Nvidia’s GPUs offer greater flexibility than TPUs. However, TPUs are often more cost-effective for AI workloads. In particular, TPUs are more energy-efficient — a huge advantage when electric power is a major constraint for AI developers.
Of course, Nvidia already faces competition in the AI chip market. For example, Advanced Micro Devices (AMD 5.22%) has stepped up its game. However, Alphabet could present an even more formidable rival because of its ability to pioneer the next big AI advances — and develop chips specifically tailored to support them.
Today’s Change
(0.04%) $0.07
Current Price
$198.52
Key Data Points
Market Cap
$4.8T
Day’s Range
$194.76 – $201.73
52wk Range
$110.82 – $216.82
Volume
4.7M
Avg Vol
173M
Gross Margin
71.07%
Dividend Yield
0.02%
“Move over” doesn’t mean “roll over”
Will Nvidia need to move over as Alphabet begins to compete in the AI chip market? To some extent, yes. Although Luria’s estimate that Alphabet could capture roughly 20% of the market may be overly optimistic, the company’s TPUs will likely gain some business at Nvidia’s expense.
However, this scenario doesn’t mean that Nvidia must roll over. I don’t expect that it will. Instead, I predict that intensified competition will prompt Nvidia to accelerate its development efforts to deliver more powerful, cost-effective GPUs.
AI developers will win as a result of this increased competition. Customers who use AI applications will win. And I think that both Alphabet and Nvidia will also win — just as they have been doing for years.
