GOOGL

Google challenges Nvidia in the AI chip market

Google is changing the rules of the game in the AI chip market with a new strategic move that allows direct deployment of TPU chips at customer sites, threatening Nvidia's monopoly.

By the SpyStocks desk · 9mo ago · 4 min read

$GOOGL Google Google Google

Much talk about Google ($GOOGL) recently – not that this is new to us. We wrote to you that everyone had written it off, and it continues to surprise?

But last night, a huge turn in a niche that might seem rather marginal, but places Google at the heart of the news

Google's chip industry - the TPU

$GOOGL stock rises about 4% in after-hours trading.

- Nvidia stock falls over 2% in after-hours trading.

And all this because investors are starting to ask themselves - is Google the next Nvidia?

Is a company threatening Nvidia's market share for the first time in history?

Let's try to understand what happened here.

For years, Nvidia (NVDA) held undisputed dominance in AI model training, but it seems this monopoly is facing its toughest test yet - and it's coming from Google (GOOGL).

For those who don't know, Google is one of the largest chip manufacturers in the world - and as of yesterday, the rules of the game are changing - completely.

Google is radically changing the rules of the game and declaring war on Jensen Huang's AI chip market share.

Note the change in strategy:

The TPU is coming down from the cloud - straight to the customer's pocket!

Until today, Google offered its TPU (Tensor Processing Unit) chips only as a cloud service (Google Cloud), and succeeded in stealing customers from all the other big players, including Amazon, and even an attempt to steal a large customer from Nvidia!

Google Cloud has become a huge cloud provider. As of September, 9 out of 10 of the world's largest research institutions were Google Cloud customers!

Google's cloud customer growth stood at almost 30% quarter-over-quarter! While Google Cloud's share of total revenue is "only" 14%, it is currently the arm with the strongest growth.

Google's advantage?

Custom silicon, control over the chip (TPU), training (AI), and distribution via the cloud, removes a significant expense that others, who only use Nvidia's ($NVDA) chips, still incur.

The cloud went from breakeven around early 2023 to a quarterly operating profit of about $3 billion in 2025!

This is what happens when you have your own silicon...?

As of last night, this chapter is over.

The new move is: direct deployment of the TPU in customer data centers (On-Premise).

In other words, Google is starting to sell chips to customers, just like Nvidia.

Overnight, it was reported that Google is in negotiations with Meta (META) for a multi-billion dollar mega-deal, yes, not one billion, but many billions.

According to the proposal, Meta will lease TPU in the cloud starting in 2026, and Google will perform extensive direct TPU deployment in Meta's data centers starting in 2027!

And that's not all

There's also a revolution in the purpose of the chips, and that's actually the big surprise...

According to the publication, the use of TPU is not only for simple "Inference" but also for training AI models – a field previously considered "exclusive to Nvidia."

A paradigm shift?

Google Cloud executives declare: "We want to take 10% of Nvidia's annual revenue."

This represents potential additional revenue of billions to tens of billions of dollars for Google.

Google is entering head-to-head competition with Nvidia, with a set of tools that cannot be ignored...?

Google's 3 secret weapons: price, security, and convenience

A winning price? - The cost of using TPU is significantly lower than that of Nvidia's GPU chips.

In a price war, customers will benefit.

Security and regulation: On-Premise deployment allows customers to keep sensitive data in their own data centers, a critical solution for organizations with strict security and regulatory requirements.

Developer convenience: Strengthening support for PyTorch (instead of Jax) and improving usability with the new 'TPU Command Center' software.

The fact that Meta created PyTorch makes this move particularly interesting for them!

Additionally, Google is using tactics similar to Nvidia (such as insuring up to $3.2 billion for small cloud providers) to distribute the TPU as quickly as possible.

The huge question here is...

What will be Nvidia's next move?

The last time Google tried to steal a customer from Nvidia (Anthropic), Nvidia announced a multi-billion dollar investment in the company to ensure they would use their GPUs.

How will it react now?

That's all for this morning.

Soon... more updates on Google....

Related Stories

Will Pfizer do a Moderna for investors?

PremiumInsidersPFE1d ago4 min read

Major move at Somnigroup International: CEO buys shares for almost $2 million one day after major deal completion

PremiumInsidersSGI8h ago3 min read

IREN: options whales predicted the decline

Option WhalesIREN11h ago1 min read