Evercore ISI Projects AI Chips Will Generate 35 Percent of Google Cloud Revenue by 2027
If the Evercore ISI estimate proves accurate, Alphabet's internal chip business would represent a structural shift in how the company monetizes its cloud infrastructure, moving hardware from a...
Alphabet's artificial intelligence chip sales could account for 35 percent of Google Cloud revenue by 2027, according to a research note from Evercore ISI cited by Investors Business Daily. The projection marks a significant focus on Alphabet's Tensor Processing Units (TPUs) and related AI silicon as a standalone commercial product, not merely infrastructure supporting other services.
Google Cloud reported $11.35 billion in revenue for the second quarter of 2026, according to Alphabet's most recent quarterly earnings filing with the Securities and Exchange Commission. A 35 percent share attributed to AI chip sales would imply chip-related cloud revenue approaching or exceeding $15 billion annually if overall cloud growth continues at its current trajectory.
Alphabet has developed multiple generations of custom AI accelerators under the TPU brand. The sixth generation, known as Trillium, was announced by Google in late 2024 and is available to cloud customers through Google Cloud's on-demand and reservation pricing models, as disclosed in Google Cloud product documentation. Evercore ISI's projection rests in part on the assumption that enterprise and hyperscaler demand for these chips will scale materially over the next 12 to 18 months.
The commercial chip market for AI workloads is currently dominated by Nvidia, whose H100 and B200 GPU series have been the primary hardware choice for large language model training and inference, according to Nvidia's fiscal 2026 annual report filed with the SEC. Alphabet's strategy differs in that it designs chips primarily for internal use and then offers access to those chips through its cloud platform, rather than selling hardware directly to end customers.
This model means that Google Cloud's revenue from AI chips is recognized as cloud services revenue, not hardware revenue, a distinction that affects how analysts and investors should interpret the Evercore ISI figure. The 35 percent estimate reflects cloud billings attributable to TPU-powered compute instances, not a separate hardware line item on Alphabet's income statement.
For context, Google Cloud's operating income reached $1.17 billion in the second quarter of 2026, per Alphabet's SEC filing, compared to an operating loss in the same segment as recently as 2022. The segment has moved toward consistent profitability over the past two years, and the Evercore ISI thesis suggests AI chip demand could accelerate that margin expansion if TPU instances carry higher average selling prices than general-purpose compute.
Alphabet faces competition in the cloud AI chip space from Amazon Web Services, which offers its own Trainium and Inferentia chips through AWS, and Microsoft Azure, which has partnered with OpenAI and also offers Nvidia GPU instances at scale. Each of the three major US cloud providers is investing in proprietary silicon as a means of reducing dependence on Nvidia and improving unit economics, according to public statements from each company's most recent investor days.
The Evercore ISI projection is a forward estimate and not a guarantee. What would confirm or refute it is Google Cloud's quarterly revenue disclosures for fiscal years 2026 and 2027, along with any segment-level breakdowns Alphabet chooses to provide regarding TPU utilization. Alphabet does not currently break out chip-specific revenue in its SEC filings, so the 35 percent figure cannot be verified independently until the company either discloses that data or third-party analysts reconstruct it from cloud pricing and usage data.
For US investors holding Alphabet stock, the relevance of the Evercore ISI note is that it reframes Google Cloud not only as a software and services competitor to AWS and Azure, but as a hardware-enabled platform where proprietary silicon may become a durable competitive advantage. Alphabet's class A shares trade on the Nasdaq under the ticker GOOGL. As of the most recent trading session prior to publication, the stock's performance relative to this projection will depend on whether management provides additional disclosure on TPU economics at its next earnings call.