Morgan Stanley, a global investment bank, says investors may be overlooking a significant long-term opportunity for Alphabet: its Tensor Processing Units (TPUs). The bank's analysts argue that the market is underestimating the potential of these custom AI chips, even as the timeline for Alphabet's next major AI models remains unclear.
What Are Tensor Processing Units?
Tensor Processing Units are specialized computer chips designed by Alphabet to accelerate machine learning tasks. Unlike general-purpose chips from companies like Nvidia, TPUs are built specifically for the kind of calculations that power large AI models. They are used both inside Alphabet's own data centers and offered to cloud customers through Google Cloud.
Morgan Stanley's note highlights that TPUs could become a meaningful revenue stream for Alphabet if demand for custom AI hardware continues to grow. The bank sees this as a potential growth driver that many investors have not fully factored into their valuations.
The Gemini Model Pipeline
At the same time, Morgan Stanley is watching Alphabet's next generation of AI models closely. The bank says the upcoming Gemini 3.5 Pro and the Gemini 4 family—which is expected to include cheaper, more accessible versions—will be critical proof points. These models will show whether Alphabet can remain competitive at the cutting edge of AI development.
However, the launch schedule remains uncertain. Morgan Stanley notes that Alphabet's management wants to ship models more frequently as it gets closer to Gemini 4, but there is no firm timeline. This lack of clarity has left some investors cautious, especially as rivals like OpenAI and Meta continue to release new models at a rapid pace.
The broader context is that AI spending has been a major theme in markets this year, with companies across sectors investing heavily in infrastructure and models. Oil Hits $100 as Red Sea Attacks and AI Spending Jolt Markets highlighted how AI-related demand is affecting commodity prices and supply chains.
What It Means for Investors
For everyday investors, the key takeaway is that Alphabet's AI story has two parts: the software side (Gemini models) and the hardware side (TPUs). While much of the market's attention has been on the race to build better AI models, Morgan Stanley is pointing to the hardware opportunity as a potentially overlooked value driver.
If TPUs gain wider adoption among cloud customers, they could provide Alphabet with a new, high-margin revenue stream. This would be similar to how Amazon's AWS and Microsoft's Azure have built profitable businesses around their cloud infrastructure. However, the success of TPUs is tied to the performance of Alphabet's AI models—if Gemini 4 proves to be a leader in the field, it could drive more customers to use Google Cloud's TPU-powered services.
The uncertainty around launch dates is a reminder that AI development is still unpredictable. Companies are racing to release new models, but delays and shifting timelines are common. Investors should watch for concrete announcements about Gemini 3.5 Pro and Gemini 4, as these will be important signals of Alphabet's competitive position.
In the broader market, AI-related stocks have been a major driver of returns, but some analysts are warning that the sector may be due for a rotation. Small Caps and Emerging Markets Surge Past Big Tech in 2026 Market Shift noted that investors are increasingly looking beyond the largest tech companies for growth opportunities.
What to Watch Next
Morgan Stanley's report suggests that Alphabet's next earnings call will be closely watched for updates on both TPU adoption and the Gemini model roadmap. Investors will also be paying attention to any announcements about new cloud customers using TPUs, as well as benchmark results comparing Gemini 4 to competing models from OpenAI and others.
For now, the bank's message is clear: don't overlook Alphabet's hardware business, even if the software timeline is fuzzy. The combination of custom chips and advanced AI models could give Alphabet a unique advantage in the cloud computing market, but the proof will be in the execution.
