Massive chip news from Google… what it means for Nvidia and the AI buildout… how Luke Lango saw it coming… another example where Luke was there first… where to invest now for the next market surge
This morning brought a huge shakeup to the tech world, and there could be portfolio-rattling consequences.
We learned that Meta (META) is in talks with Alphabet (GOOGL) to potentially buy billions of dollars’ worth of Alphabet’s TPU chips rather than the industry standard GPU chips produced by firms such as Nvidia (NVDA) (disclaimer: I own GOOGL).
This could be a huge shift for the tech sector – and investor portfolios. As I write on Tuesday, GPU giant Nvidia is down 5%.
Let’s back up and fill in a few details…
Think of AI like a city that’s exploding in size. Nvidia has been selling the construction machinery – the cranes, the bulldozers, the equipment that builds the skyscrapers.
Well, with this morning’s news, Google walked in and said, “Actually, we make our own custom machinery now – and it might even be better for certain kinds of buildings.”
If Google’s right, then the most critical part of AI’s future – the infrastructure – might shift away from Nvidia’s long-standing dominance.
This “machinery” involves TPUs and GPUs
Nvidia’s GPUs (Graphics Processing Units) have been the standard for training and running large AI models. They’re flexible, powerful and have a strong ecosystem behind them.
Google’s TPUs (Tensor Processing Units) are different: They’re custom silicon (often called ASICs) optimized for machine-learning-specific workloads – especially inference, large-language models, etc.
The big news this morning is that Google is moving to commercialize its TPUs more widely, expanding access beyond limited partners like Anthropic.
Here’s Bloomberg:
Meta’s likely use of Google’s TPUs, which are already used by Anthropic, shows third-party providers of large language models are likely to leverage Google as a secondary supplier of accelerator chips for inferencing in the near term.
Meta’s capex of at least $100 billion for 2026 suggests it will spend at least $40-$50 billion on inferencing-chip capacity next year, we calculate.
If Google succeeds, the value of the “platform” or “ecosystem” in AI hardware shifts.
It won’t only be about GPUs from Nvidia anymore; there could be meaningful alternative architecture in play – which can affect margins, growth, and competitive advantage.
Does this mean dump Nvidia?
No.
Just because a company gets challenged doesn’t mean it’s dead. Nvidia still has massive scale, ecosystem, customers, software support – those don’t evaporate overnight.
Plus, Alphabet’s TPU chips aren’t a sure thing yet.
Here’s Bloomberg:
Much depends on whether the tensor chips can demonstrate the power efficiency and computing muscle necessary to become a viable option in the long run.
Still, we must understand this sudden risk and then reevaluate how much exposure we want and why.
On one hand, Nvidia still has the world’s deepest ecosystem of tools, developers, and software built specifically to work with their hardware. That’s not something Google can replicate in a year or two.
But Google doesn’t need to “win” outright to change the landscape. It only needs to prove that its TPUs are a viable alternative.
All it takes is one major customer – such as Meta – shifting part of its AI workload to Google silicon, and suddenly the ground underneath Nvidia’s throne starts to rumble.
But this story is way bigger than Nvidia alone
TPUs may be significantly cheaper to operate than GPUs. And if that proves true at scale, the effects go far beyond GOOGL and NVDA trading blows in the stock market.
Nvidia’s GPUs are incredibly powerful, and also extremely flexible – designed to handle a wide range of computing tasks.
But that flexibility comes at a high cost because you’re paying for capabilities that aren’t always necessary for AI workloads.
Google’s TPUs, on the other hand, are highly specialized chips built almost exclusively for one thing: accelerating the matrix math that powers large-scale AI models.
By stripping out everything nonessential, Google can make these chips simpler, more power-efficient, and most importantly, cheaper to run. Lower component costs and reduced energy usage result in lower total compute costs for customers.
If TPUs take major market share, it would change the entire industry – and potentially, make Google the new AI king…
If TPUs deliver similar performance per dollar – or better – the economics of AI change overnight.
The biggest technology companies spend billions running AI models. If Google offers a significantly cheaper alternative, these companies have a financial incentive to shift more of their workloads away from Nvidia hardware. And that has ripple effects…
Cloud platforms would need to adjust pricing… AI startups would rethink their cost structures… and investors would have to reassess Nvidia’s future growth assumptions.
To be clear, my point is not that Nvidia is suddenly in trouble. It’s that the market is suddenly facing the possibility that AI compute – the most expensive raw ingredient in this entire revolution – may not be as locked-in or as expensive as it appeared last week.
Consider the implications for all those data centers popping up everywhere…
Consider the implications for all that capex spend that the hyperscalers have already dropped…
Consider how this news could redirect trillions of dollars’ worth of capital flows over the next few years…
Bottom line: Cheaper AI would change everything. And Google just cracked that door open.
Here’s where things get even more interesting…
While Wall Street is only waking up to the TPU narrative today, some of the sharpest people inside our own research group were mapping this out months ago.
Back in early June, an internal email thread started circulating between Mike Merson, editor of the TradeSmith Digest, Luis Hernandez, InvestorPlace’s Editor in Chief, Senior Analyst Brian Hunt, and our technology expert Luke Lango, editor of Early Stage Investor.
What stood out wasn’t just that they were talking about TPUs… but how they were talking about them.
Luke, in particular, was out in front of the curve. His core insight came down to this evolving split inside AI:
GPUs > TPUs on AI training… but TPUs > GPUs on AI inferencing.
That line captured the quiet turning point that Wall Street hadn’t seen yet. But Luke laid it out even more clearly:
We are going from AI training to inferencing, and that requires a new type of AI chip.
GPU usage should fall… TPU usage should rise.
And then the kicker – the part today’s market action is validating:
NVDA is behind on TPUs. GOOGL is the only company making TPUs at scale right now.
So, the play is to invest in the GOOGL TPU supply chain.
Since that email thread on June 4, 2025, GOOGL is up 93% – nearly 4Xing NVDA’s 24% return over the same period.
Stepping back, this is hardly the first time that Luke has seen big moves coming long before Wall Street.
Let’s look at another recent example – because the opportunity to profit here is just getting started…
Between July 1, 2022, and July 1, 2025, MP Materials (MP) was dead money
During a period in which the S&P jumped 64% and the Nasdaq climbed 83%, MP didn’t just trail the major indexes, it destroyed investor wealth.
Take a look…
If you’re a money manager who put clients into MP, you’re likely out of a job with this kind of performance. And for individual investors, this is a huge drag on retirement timing.
But as you’re likely aware, this wasn’t the end of the story.
Here in 2025, the U.S. government has launched a massive, multitrillion-dollar effort to secure America’s dominance in artificial intelligence and other critical technologies.
Behind this effort is a simple logical progression…
We’re in an AI race against China… Beijing controls 80% of the rare earth elements (REEs) that are critical to building out our cutting-edge AI, robotics, and defense systems… so, President Trump has been on a warpath to secure these REEs… a handful of specific REE stocks – MP being one of them – stood to benefit.
Sure enough, in July, MP Materials announced that the Pentagon would purchase $400 million of its stock, taking a 15% ownership stake through structured agreements. The stock surged, putting subscribers of our technology expert Luke Lango up about 3X on their MP investment in short order.
Now, I’m not telling this to brag on Luke, but to call your attention to something important…
The same methodology Luke used to find MP before it received that firehose of federal dollars is the same one Luke is using today to find the next target of the government.
So, if you weren’t a part of that MP surge, don’t worry – we’re still in the beginning of this government spending spree.
How to spot the government’s next move
So, what was Luke’s methodology that enabled his early recommendation of MP?
It’s the same one he just codified and offered as a blueprint for investors today.
From Luke:
Here’s a simple three-part screen for predicting future winners:
- Right Industry: Is it in a sector Washington deems essential (AI, semiconductors, critical minerals, drones, nuclear, cybersecurity, etc.)?
- Right Company: Is it the clear leader with the best shot at solving the national-security problem?
- Right People/Relationships: Do they have U.S. operations, D.C. ties, or credible strategic backers?
If a company checks these three boxes, it’s a strong candidate for federal support… and often just one announcement away from a major stock move.
Luke says MP was only the beginning
According to him, we’re entering Act 2 of a massive, government-driven tech push – one that could send a new class of companies soaring as Washington deploys more capital, more contracts, and more strategic investments.
We’re talking rare earths, semiconductors, small nuclear reactors, robotics, advanced software, and drone technologies – it’s an enormous opportunity set with triple- even quadruple-digit return opportunities.
Back to Luke:
There are dozens of companies that haven’t yet received federal backing… but almost certainly will.
And investors who buy those companies before Washington strikes will capture the biggest gains of the entire AI Boom.
I’m talking about potential 500%… 1,000%… even 2,000% returns.
This is Act 2.
Over the last six months, Luke has done a deep dive into this “Act 2” opportunity, and he’s put together his U.S. Government AI Shortlist. These are the highest-probability companies in line for investment, partnerships, or guaranteed contracts.
There’s one in particular that Luke says stands out. It sits at the very center of the AI supply chain – and Luke is giving it away.
It’s in his latest briefing in this broader opportunity. You’ll get its name, ticker, and Luke’s explanation for its 10X return possibility when you review the broader research package. It’s totally free.
Now, a quick timing clarification…
If you read yesterday’s Digest, you might be asking yourself, “Now? Didn’t Luke just advise investors not to rush back into today’s market?”
Well, a few hours after yesterday’s Digest hit your inbox, Luke sent out a flurry of “Buy Alerts” from his various investment services.
Here he is from his Early Stage Investor Daily Notes explaining:
For months, we’ve walked you through every AI bear talking point — overspending, circular financing, accounting noise, valuation panic — and showed why none of it breaks the AI Boom. But we also warned: we don’t catch falling knives. We wait for strength.
Well… the turn just happened.
Sentiment is turning. The Fed is turning. The AI narrative is turning. And the next leg of the AI Boom is setting up right in front of us.
This is the moment to buy the dip in AI stocks.
Wrapping up
Luke says the uncertainty is behind us.
This is the same analyst who spotted the TPU shift months before today’s headlines…and who handed subscribers triple-digit gains in MP…
He’s now saying the next phase of the AI Boom is officially underway.
Consider yourself ahead of the curve.
Here’s Luke’s to take us out:
The White House is actively picking and funding the companies it deems crucial to winning the 21st-century tech race.
The investors who recognize this shift will be the ones who get rich from it.
Have a good evening,
Jeff Remsburg
