Amid Broadcom's sharp decline, this investment bank made a bold claim: channel research indicates that Google's TPU shipments will reach 35 million units in 2028, far exceeding expectations!

Amid Broadcom's sharp decline, this investment bank made a bold claim: channel research indicates that Google's TPU shipments will reach 35 million units in 2028, far exceeding expectations!

As Broadcom's stock comes under pressure, a recently circulated channel survey by Mizuho shows a strong bullish signal from the bank regarding Broadcom. There is only one core data point, but it’s explosive enough: It predicts that Google’s TPU shipments will exceed 35 million units in 2028.

What does this number mean? Google’s TPU shipments in 2025 are about 2.4 million units, estimated at about 4.3 million in 2026, then jumping to 35 million in 2028—an eightfold increase in two years, implying a compound annual growth rate of about 185%.

Google’s TPU is a custom AI chip (ASIC), and Broadcom is its core design partner. For every TPU shipped, Broadcom receives income from its ASIC business. Simply put, Google provides the demand and architecture, Broadcom is responsible for chip design services and packaging, and TSMC handles manufacturing. For each TPU shipment, Broadcom gets the corresponding design service revenue.

However, the credibility of this number has been directly questioned by market observers: The entire industry's ASIC shipment forecast for 2028 is only 15 million units; Google alone is supposed to produce 2.3 times that? "I'm really curious, did some Mizuho intern get Claude to extrapolate a trend line, and then nobody checked the output?"

35 million units—how unbelievable is this number?

The numbers from Mizuho’s channel survey have sparked notable controversy in the industry.

Let’s compare with a few reference points:

  • Counterpoint Research predicts that by 2028, the total ASIC shipments of all hyperscale cloud providers like Google, Amazon, Microsoft, and Meta will be about 15 million units.
  • Wolfe Research channel checks indicate Google’s TPU shipments to be about 7 million units.
  • Mizuho’s own model previously predicted ASIC shipments to be about 7 million units.

Yet this channel survey gives a number of 35 million units—it’s five times Mizuho’s own model, and 2.3 times the total industry forecast.

In other words, per Mizuho’s survey, Google alone in 2028 would ship more TPUs than all other ASIC projects on Earth combined—by double.

Doubt: "Calculable" doesn’t equal "achievable"

Market observer @LinQingV directly called out this survey on X platform.

"Have you really thought through what 35 million TPUs shipped in 2028 would mean?" he wrote.

His logic is clear: going from 4.3 million to 35 million implies a two-year CAGR of about 185%. "Even with dual suppliers like MediaTek, and Anthropic’s 3.5GW agreement, there’s no supply chain in the world that can expand at this rate."

He also noted that Mizuho's calculation for Micron’s HBM demand is mathematically consistent—cumulative TPU shipments from 2026 to 2028: 4.3 million + 10.7 million + 35 million = 50 million, mathematically correct. But, his conclusion is: "Internal consistency" and "physical feasibility" are two different things.

He ended with: “I'm really curious, did some Mizuho intern get Claude to extrapolate a trend line, and then nobody checked the output?”

Besides Google TPU, what else did Mizuho mention?

This channel survey isn’t just about Google.

Mizuho also mentioned that OpenAI's ASIC chip is in development, tied to Broadcom’s "Nexus" project, which is up to 10GW and will supplement Google TPU, Meta MTIA, and ARM chip roadmaps.

For ARM, calculations show its AI ASIC chip development is progressing, expected to launch between late 2026 and early 2027.

The impact on Micron was also singled out: if cumulative TPU shipments reach 50 million units between 2026 and 2028, ASIC will take an increasing share in the HBM market, especially as the next-gen ASICs move to HBM4e specifications.

Risk warning and disclaimerThe market has risks, and investment should be cautious. This article does not constitute personal investment advice, nor does it consider the special investment goals, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their specific circumstances. Investing accordingly is at your own risk.