Sergey Brin, co-founder of Google, and Director of Google X and Special Projects, attends an event debuting the new Google self driving car outside the Google X labs in Mountain View, CA. (Photo by Brooks Kraft LLC/Corbis via Getty Images)
Corbis via Getty Images
Shares of Alphabet are up 66% in 2025.
One reason could be the ChatGPT-beating performance of Gemini 3 Google’s latest AI chatbot, reported the Wall Street Journal.
Another may be a possible “multi billion dollar contract” with Meta to rent, then buy Tensor Processing Units from Google to train the Facebook-parent’s superintelligence models, reported TechRadar.
This news do not sound great for Nvidia which currently dominates the market for Graphics Processing Units for training and operating the large language models powering AI chatbots.
Nvidia both competes and cooperates with many of the participants in the generative AI value network. Indeed, Open AI, Google and Meta Platforms are Nvidia customers, reported the Wall Street Journal. The success of Gemini has prompted OpenAI to declare a ChatGPT ‘Code Red,’ noted the Journal while Meta aims to compete with both of them.
For investors, the most significant consideration is whether Google’s TPUs will take a significant share of the AI chip market which is expected to grow at a 25% average annual rate to $440 billion by 2030, noted Mordor Intelligence.
By 2030, Nvidia’s dominant share of the AI chip market could fall significantly from 90% to 70%, estimates Kearney. During this time, Google’s share could increased from 5% to 25%, according to a Vikas Sah Medium post.
Read on for what these changes mean for investors in $GOOGL and $NVDA.
Google’s Gemini Catching Up With ChatGPT
OpenAI’s ChatGPT – with a growing user base topping 800 million weekly users – is facing serious challenges. Since July when Google added an image generator dubbed Nano Banana, Gemini’s user base has risen 44% to 650 million in October while Anthropic – which may go public next year – is popular among business users, noted the Journal.
Google has a considerable financial advantage over OpenAI. That’s because the ChatGPT parent will need to grow its revenue to some $200 billion to turn a profit in 2030, according to its own financial projections. In 2025 and 2026, OpenAI projects net losses of $9 billion and $14 billion respectively, noted The Information, which puts pressure on the company to keep raising capital.
The AI chatbot industry’s competitive dynamics prompted OpenAI CEO Sam Altman to declare a “code red” effort to improve ChatGPT’s “personalization features for users, increasing its speed and reliability, and allowing it to answer a wider range of questions,” the Journal reported.
Google’s TPU Aiming At Nvidia’s AI Chip Lead
Google and Nvidia both have ample operating profits with which to fund a competition for the gigantic and rapidly growing market for AI chips. While Nvidia’s operating income over the last four quarters topped $110 billion, Google’s led the S&P 500 at $151.4 billion, noted the Journal.
But Google’s possible deal to sell TPUs to Meta could be the beginning of a significant source of revenue for the search giant. Meta would rent Google Cloud TPUs during 2026 and transition to direct purchases in 2027, TechRadar reported, adding the possibility of a “multibillion-dollar agreement” – which sent Alphabet’s market capitalization up to nearly $4 trillion last week.
Google’s TPUs could end 2025 controlling 8% of the AI chip market. This estimate adds the one million TPUs Google has reportedly reserved for Anthropic, according to Reuters, to the 5% market share forecast by ByteBridge.
By 2030 Google’s TPU Could Control 25% Of AI Chip Market
By 2030, Google’s TPUs could control 25% of the AI chip market, according to Vikas Sah. Google’s TPUs will likely win customers – beyond Anthropic and Meta — for many reasons: lower cost-per-compute, strong performance and efficiency, integration with a leading cloud platform, and relief from the GPU supply shortage, noted ByteBridge and Vikas Sah.
While Nvidia still offers greater flexibility – due to the the developer community’s widespread adoption of its CUDA software, as a I noted in my book Brain Rush – TPUs shine in well-defined large scale deep learning tasks. Buyers will continue to use both; however, TPUs are emerging as a credible threat to Nvidia’s GPU dominance.
The competitive pressure is real enough to affect pricing. OpenAI negotiated a 30% discount on its Nvidia compute fleet by threatening to purchase TPUs—without deploying a single one, according to a SemiAnalysis report report from Futu News.
Is $GOOGL A Better Bet Than $NVDA?
Wall Street sees Alphabet shares as slightly over-valued while estimating Nvidia stock is cheap.
How so? Google’s stock trades a fraction above the $314.70 a share average price target, according to TipRanks, while Nvidia stock – up 45% in 2025 – would need to rise 46% to reach its price target of $257.72.
I am not so sure. With Google’s generative AI ecosystem initiatives, I see its new growth from TPUs and possibly Gemini making $GOOGL the better bet.