Nvidia had better watch its back…

That might sound like an audacious claim about the premier AI chipmaker and world’s biggest company.

But hear us out.

Nvidia (NVDA) appears to have the biggest defensive moat in market history.

It posts continually record-setting revenue (it’s doubled to a forecasted $130 billion this year from $61 billion last). It holds 94% of the market share for graphics processing units (GPUs), which are used to train and run AI models. And it’s among the most productive tech companies ever, making more than $4 million per employee.

It’s hard to imagine anyone coming for Nvidia.

But that’s exactly what’s happening.

Allow me to turn your attention to a Daily from August 2024. There, I shined the spotlight on an alternative type of AI chip called tensor processing units (TPUs).

Without getting too into the technical details, you can think of GPUs as being like a bulldozer. They use sheer power to solve problems, and they may not always be the most efficient choice. But with brute force, they get the job done.

TPUs, on the other hand, are more specialized for a certain type of AI integration that’s extremely promising. It’s called “edge computing.” And it proposes to take all that GPU power that’s currently hosted in a datacenter and integrate it into the device itself.

Edge computing is a big piece of the TPU trend, but it’s just one of them.

TPUs are also much more power- and bandwidth-efficient – they use less energy and more easily connect with each other than GPUs. That makes them especially useful in the huge datacenter builds you keep hearing about.

So why aren’t they already dominant?

Simply put, by the time the AI trend really kicked off, they weren’t as readily available as NVDA’s GPUs. TPUs only came around in 2016. NVDA has been building GPUs for decades. That’s how NVDA was able to grab so much market share in such a short time.

In short, TPUs are a significant evolution of the dominant AI chip of today. They’re more efficient at handling AI tasks. And while most companies aren’t using TPUs yet, they’re starting to… and that threatens to make NVDA’s AI dominance obsolete.

But who makes these TPUs?

The Magnificent 7 stock that’s running away from the pack in 2025, up 68% and more than double Nvidia’s gain…

I’m talking about Google (GOOGL)…

Google first announced its own TPU back in 2016. And until now, it’s largely been working on them in the background.

But news broke yesterday that Meta Platforms (META) was in talks to spend billions on Google’s TPUs for its data centers in 2027.

This comes after another leading AI company, Anthropic, bought 1 million of Google’s TPUs. And a year after Apple secured TPUs for its Apple Intelligence efforts.

It’s a potentially seismic shift in how we think about the AI trade.

See, Nvidia’s GPUs are not purpose-built for running AI models. They were originally built to render graphics for video games. They just also happened to be good at running AI models.

TPUs are specifically made for running AI models. Nvidia may have a stranglehold on the market for GPUs. But instead of trying to compete directly with its own GPUs, Google has come up with an entirely new kind of chip.

And investors are recognizing this potential shift. As you can see in this chart, GOOGL has surged higher over the past few weeks:

NVDA had a handsome lead over the summer – but now, even after a strong earnings report, it’s struggled so far downward that GOOGL is now beating it 2 to 1 for the year.

If you think it’s too late to buy Google, check this out…

It’s easy to think you’ve missed the boat on this trade. But if the last couple years have taught us anything, it’s that the size of these tech companies doesn’t limit how high they can fly.

And according to our own AI-based forecasting algorithm, GOOGL is a strong buy right now.

If you aren’t aware, Predictive Alpha is TradeSmith’s proprietary AI-based trading model. Where ChatGPT is a large language model that tries to guess the next word in a sequence, you can think of Predictive Alpha as a large numbers model. It uses AI to guess the next number in a sequence.

The end result is a clear-cut forecast for your stock. Check out what Predictive Alpha Prime expects for GOOGL – 7.8% upside by mid-December:

We launched Predictive Alpha with a slightly longer, 21-trading-day rolling forecast. But what we found in our research is that some stocks move more consistently across different timeframes. That led us to the Prime forecast, which shows you the best holding time for any AI-based trade we track in our system.

In the case of GOOGL, its Prime forecast is for 17 trading days from today. That takes us out to Dec. 18. It forecasts GOOGL will be 7.8% higher by that time, and it’s been accurate on similar calls 81.3% of the time in the past.

So if you think you’ve missed the boat on GOOGL and its TPU advances, think again. Our AI forecast shows this is just the start of a long and strong trend.

And if the next 17 days are as strong as Predictive Alpha suggests, you’ll want to get in ahead while the trend is still accelerating.

— Michael Salvatore

The Guy Who Beat Buffett Shocks the World Again… [sponsor]
The New York Times called him "an icon among growth stock investors." The Wall Street Journal said "Most money managers can only dream of having the same success" he's had. His investment fund produced a 4,000% return over a 15-year span, outperforming even the great Warren Buffett. His name is Louis Navellier. And he's just issued an astonishing financial prediction. Click here to see it.

Source: TradeSmith