Alphabet (GOOGL) shares jumped, while Nvidia (NVDA) and AMD (AMD) took a tumble. The headline? Google is now pitching its custom Tensor Processing Units (TPUs) directly to Meta Platforms (META) and other big players, aiming to grab a slice of Nvidia's AI chip dominance. But let's peel back the layers and see what the numbers really tell us.
The core of the story is Google's strategic shift. They're no longer just renting out TPU infrastructure; they're selling the chips outright, touting enhanced security and compliance. Google Cloud executives are even projecting they could snatch up to 10% of Nvidia's annual revenue. That's a bold claim. Is it plausible? Well, Nvidia's current dominance is significant, so a 10% dent would be a major win for Google.
Meta is reportedly in talks to spend billions on Google's TPUs starting in 2027 and might start renting TPU capacity as early as next year. This is where things get interesting. Meta is a massive consumer of AI compute, currently relying heavily on Nvidia GPUs. If Meta switches, even partially, to TPUs, it's a clear signal that Google's chips are a viable alternative. The article suggests Google's TPUs are cheaper and were used to train Gemini 3, their most advanced AI model. Cheaper is good, but the key question is performance. Can TPUs deliver the power efficiency and computing strength Meta needs for its AI workloads?
Here's where the data gets a little thin. We don't have hard numbers comparing the performance of Google's TPUs versus Nvidia's GPUs in Meta's specific use cases. We're relying on Google's claims and Meta's potential interest. That's not nothing, but it's not a done deal either. What specific benchmarks are Meta using to evaluate the TPUs? What kind of cost savings are they projecting? These are the details that would truly validate Google's claims.
Soros Fund Management (which George Soros remains chairman of) has been making big bets on both Amazon (AMZN) and Alphabet, increasing its GOOGL holdings substantially in Q3. While Soros has warned of the dangers of AI, his fund has not been shy of picking up shares of the sector’s leaders. In Q3, Soros Fund Management bought 631,397 shares of Alphabet, bringing its total stake to 658,367 shares, currently worth $210 million.

The narrative is that companies are seeking to reduce their reliance on Nvidia. Is this just wishful thinking, or is there real evidence? The fact that Google already has a deal to sell up to 1 million TPUs to AI startup Anthropic PBC (PC:ANTPQ) suggests there's demand. But Anthropic is just one company. Meta is the whale Google is hunting. Nvidia shares fall 3% on report Meta will use Google AI chips
The Arete research firm recently upped its price target for GOOGL, and several other firms have reiterated "buy" ratings. But here's a crucial point: the average Alphabet price target of $312 implies a potential downside of 2.1% from current levels. That's not exactly a ringing endorsement. And insider activity shows John Kent Walker sold 17,816 shares of GOOGL stock. (He sold at an average price of $247.42, for a total value of $4,408,034.72.) Insider selling doesn't always mean trouble, but it's worth noting.
I've looked at hundreds of these filings, and while I wouldn't call it unusual, the timing is interesting given the TPU news.
Now, the cookie notice from NBCUniversal might seem irrelevant at first glance, but it highlights a critical issue: data privacy and control. Google is pitching TPUs as a solution for companies with strict data security needs. This is a smart move, as concerns about data sovereignty and compliance are only going to increase. If Google can convince companies that on-site TPUs are more secure than cloud-based GPUs, they have a compelling value proposition.
Google's TPU strategy is a calculated gamble, not a guaranteed win. They're taking on Nvidia in a market where Nvidia has a significant head start. But Google has deep pockets, cutting-edge AI technology, and a strong incentive to diversify its revenue streams. The Meta deal, if it happens, would be a major validation of their strategy. But even without Meta, Google's push into the AI chip market is a move that Nvidia can't afford to ignore. The proof will be in the performance data—and the long-term contracts.
Google's definitely making a play, but the market hasn't decided yet if it's a checkmate or just a clever feint. The next few quarters will give us the cold, hard numbers to decide.
Look around you. Everywhere you turn, there’s a hum of anticipation, a palpable energy in the air. W...