A piece of news rocks the market, the competitive landscape of AI is changing! Nvidia plunges 7% intraday, while Google is poised to reach a new high.

```

During early trading this Tuesday, Nvidia's share price plunged over 7% at one point, erasing nearly $350 billion in market value. The drop later narrowed to below 5%, but it is still on track to close at its lowest in more than two months. Meanwhile, Alphabet, Google’s parent company, saw its shares rise further, at one point up more than 3% in early trading. Although it later gave up most gains, it still looks set to hit a record high for a third consecutive trading day.

Behind the market’s sharp reaction, investors are sensing subtle changes in the competitive landscape of artificial intelligence—Google is beginning to take the lead. The tech giant, once considered to be lagging in the AI race, now shows signs of a comeback.

On the surface, this shake-up was triggered by a report from The Information after markets closed on Monday: Meta is considering using Google’s Tensor Processing Units (TPUs) instead of Nvidia chips in its data centers by 2027. At a deeper level, it’s because Google’s newly released Gemini 3 model has immediately won acclaim, being considered superior to OpenAI’s ChatGPT in reasoning and coding skills—crucially, it was trained on TPUs rather than Nvidia chips.

The impact on the market landscape was immediate. Nvidia’s plunge dragged down the entire tech sector: Super Micro Computer (SMCI), a key partner, fell over 6% in early trading Tuesday; data center operator CoreWeave, another holder, dropped 10% in early trading; even AMD, Nvidia’s main rival, was down nearly 9.7% intraday. Meanwhile, Broadcom, which helped Google design the TPU, saw its shares soar 11% on Monday and rise more than 2% in early trading Tuesday before reversing. Alphabet’s market value is now close to breaking the $4 trillion mark for the first time.

Mike O'Rourke, chief market strategist at Jones Trading, said that the release of Gemini 3 “could prove to be a subtler but more significant event than the DeepSeek shock,” and the market is embracing the view that “Google is the clear AI leader.” Nomura strategist Charlie McElligott believes Alphabet’s latest model has “reset” the “AI hierarchy chessboard,” ushering the market into a “new DeepSeek moment.”

Google’s Awakening: From Underperformer to Leader

Since the launch of ChatGPT three years ago, analysts and experts—including Google engineers and its former CEO—once claimed Google was falling behind in the risky AI race. Today, that assertion is outdated.

Media outlets note that Google’s recent AI software releases and deals have convinced investors the company won’t easily lose to OpenAI, the creator of ChatGPT, or other competitors. Neil Shah, an analyst and co-founder at Counterpoint Research, commented: “You could say Google has always been the dark horse in the AI race, a sleeping giant that is now fully awakened.”

The numbers tell the story. Since mid-October, Alphabet’s share price has added close to $1 trillion in market value, helped partly by Buffett holding $4.9 billion in shares during Q3, and overwhelmingly positive sentiment from Wall Street surrounding its AI efforts. On Tuesday, SoftBank Group, a major backer of OpenAI, saw its shares drop to a two-month low—driven by concerns over Google Gemini’s competition.

Google’s advantage lies in its years of “full-stack” development. CEO Sundar Pichai said last quarter: “We’ve taken a holistic, deep, full-stack approach to AI, and it’s really working.” Unlike OpenAI, Google owns a ready data corpus for model training, has steady profit streams, and controls its own compute infrastructure.

The Unique Edge of TPU: Specialization vs Generalization

In architecture, Google’s TPU and Nvidia’s GPU differ fundamentally—a difference now translating into market advantage.

Media reports note GPUs were originally designed to render video game graphics realistically, processing multiple tasks in parallel via thousands of compute “cores.” This architecture lets them execute AI tasks at speeds unmatched by other technology. In contrast, TPUs were built specifically for matrix multiplication—the main operation in neural network training, involving repeated, sequential computation rather than parallelization.

This specialization brings tangible advantages. Jay Goldberg, an analyst at Seaport, said that TPUs outperform GPUs for some AI jobs because Google can “strip down much of the chip,” removing features not custom-tailored for AI. TPUs are seen as less flexible and specialized than Nvidia’s GPUs, but are more energy efficient at running these operations. Nvidia’s GPUs are considered more adaptable and programmable—but that flexibility can make them costlier to operate.

Ben Barringer, head of tech research at Quilter Cheviot, noted: “Many other companies have failed in the quest to build custom chips, but Google evidently adds another string to its bow here.”

Customer Validation: From Exclusive Use to Market Expansion

Google began developing its first TPUs in 2013, launching them two years later. Initially, they accelerated the company’s web search engine and improved efficiency. In 2018, Google first offered TPUs on its cloud platform, letting customers sign up for the same computational services powering its search engine.

For years, Google was essentially the sole customer for its custom processors. That’s now changing. AI startup Anthropic announced in October it would use up to one million Google TPUs in a multibillion-dollar deal—a major shift from internal tool to market product.

According to The Information, Meta plans to use Google’s chips in its data centers by 2027. Google declined to comment on specific plans but said cloud demand for custom TPUs and Nvidia GPUs is “accelerating.” “For years, we’ve been committed to supporting both,” a Google spokesperson wrote. Meta declined to comment on the Monday night report.

Current TPU customers also include Safe Superintelligence, the startup founded last year by OpenAI co-founder Ilya Sutskever, as well as Salesforce and Midjourney. These high-profile AI companies provide major validation for TPU’s market potential.

Market Prospects: Complementary, Not Replacement

While TPUs show strong momentum, no one—not even Google—now seeks to fully replace Nvidia GPUs. The pace of AI development means this is currently unattainable.

Gartner analyst Gaurav Gupta said that even with its own chips, Google remains one of Nvidia’s biggest customers, since it must stay flexible for client needs. If clients’ algorithms or models change, GPUs are better suited to diverse workloads.

Meryem Arik, CEO of AI startup Doubleword, points out that Google’s TPUs mainly appeal to a handful of companies with huge compute bills, like Meta and Anthropic. There’s another key limitation: “Once you use TPUs, you’re locked into Google’s cloud ecosystem.” AI developers can only access TPUs through Google’s own cloud, whereas using Nvidia GPUs is much more flexible.

Barringer noted that the chip industry is “not a zero-sum game with only one winner.” In fact, even tech firms signing for TPUs are still heavily investing in Nvidia chips. For instance, Anthropic announced a major deal with Nvidia just weeks after its TPU agreement with Google.

According to Bloomberg Intelligence analysts, Anthropic’s deal makes it likelier that TPUs will expand to other clouds. For Google’s TPU, the best outlook may be as part of a package of products underpinning AI’s explosive growth.

Forrester analyst Thomas Husson summed it up: “It’s safe to say that Google, with Gemini 3, is back in the race. In fact, to borrow a phrase from Mark Twain, reports of Google’s death have been greatly exaggerated—perhaps even meaningless.”

Risk Warning and DisclaimerThe market involves risks and investments should be made cautiously. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial condition, or needs of individual users. Users should consider whether any opinions, views or conclusions in this article suit their particular circumstances. Investments made on this basis are at your own risk. ```