Amid concerns about an "AI bubble" that have led to a divergence in the performance of US tech stocks, Alphabet, Google's parent company, has seen its stock price surge recently, rising nearly 20% in the past month and becoming the hottest large-cap AI stock.
On the one hand, Google's latest large-scale model, Gemini 3, has received rave reviews; on the other hand, its self-developed AI chip... —TPU (Tensor Processing Unit) is attracting attention from the capital market.
Reports this week indicated that Google is marketing its TPUs to customers, and Meta is considering spending billions of dollars to purchase Google TPUs for use in data centers. Construction has begun. This marks a significant shift in Google's strategy, as its TPUs were previously primarily used within its own cloud data centers .
Morgan Stanley This is believed to provide a modest boost to the tech giant's sales and profits .
"The extent to which Google sells TPUs through a first-party model does indeed have a significant impact on its performance. Our sensitivity analysis shows that for every approximately 500,000 TPU chips Google sells, it could add about $13 billion (about 11%) upside to its 2027 cloud revenue forecast and boost its 2027 earnings per share (EPS) by about 3% (i.e., $0.37)," wrote a report to clients by a team led by Morgan Stanley analyst Brian Nowak.
Analysts explained that if Google Cloud's growth accelerates and the company successfully enters the semiconductor market , [the situation could improve]. The market believes this could very well drive (or support) Google to achieve a higher valuation multiple.
Furthermore, analysts pointed out that, considering Nvidia... Google expects to ship approximately 8 million GPUs (graphics processing units) in 2027, and its prediction of selling 500,000 to 1 million TPUs annually is "not unreasonable" if production capacity allows .
However, analysts also emphasize that there is still significant uncertainty surrounding Google's strategy for marketing its TPUs, and investors need to focus on three core issues: its business model, pricing strategy, and the types of workloads that TPUs can handle.
Regarding the impact of Google's sales of TPUs on the semiconductor industry, analysts believe it will benefit Broadcom. The impact on Nvidia and AMD is minimal, as the company co-designed the TPU chip with Google .
This has somewhat alleviated market concerns surrounding Nvidia , as investors worried that Nvidia's market share in AI chips would be eroded with the rapid rise of Google's TPU chips. Google's TPU outperforms GPUs in certain scenarios—a field in which Nvidia holds a dominant position.
Morgan Stanley analysts pointed out that Google has spent about $20 billion on Nvidia chips this year to advance its large language model, while spending only a little over $1 billion on TPUs. This spending structure may be slightly adjusted next year, but the overall growth in AI chip demand is unlikely to result in a "winner-takes-all" situation.
“Even if the Gemini model surpasses GPT-5 to become the leading model, it doesn’t mean that competitors will slow down—as technology industry leaders have repeatedly emphasized, this race is crucial, and we expect the competition to remain fierce, especially as the law of scale still applies,” they said.
Dongcai Illustrated Guide: Some Useful Tips
Is Nvidia panicking? A graphic explaining what Google TPU is.
(Article source: CLS)
