Evercore ISI says Alphabet’s AI chips could account for a growing share of Google Cloud revenue by 2027

Alphabet’s effort to shift more of its artificial intelligence computing stack toward custom hardware could translate into a noticeably larger contribution from “AI chips” to Google Cloud revenue, according to Evercore ISI, as described by Yahoo Finance in a market note.

The report’s central figure is a projection that sales of AI chips could represent as much as 35% of Google Cloud revenue in 2027. The same note frames that outlook as potentially supportive of Alphabet’s stock performance, tying hardware monetization to the broader cloud growth story.

The logic behind such estimates typically rests on two linked developments: AI workloads increasingly require specialized accelerators, and hyperscalers can capture more value when they supply both infrastructure and the underlying compute capacity rather than relying only on third-party chips. For Alphabet, that means customers purchasing cloud services that include compute built around AI chips could indirectly boost revenue attribution to those chips over time.

Google Cloud has positioned its AI and machine learning offerings as a key growth area, and the industry-wide trend is that customers are buying end-to-end systems for training and inference. In that environment, analysts often build scenarios in which custom chips become embedded in the cost and pricing of cloud services, raising the share of revenue that maps to chip-related supply and capacity.

While the Yahoo Finance note highlights the 2027 contribution rate, it does not, in the information provided here, spell out the underlying assumptions. That includes how the “AI chips” revenue is defined (for example, whether it is direct product revenue, internal transfer pricing, or how cloud service economics are modeled), and whether the estimate assumes a specific ramp in demand, production capacity, or customer adoption.

For investors, the key angle is the potential effect on cloud economics. If a larger fraction of cloud sales can be tied to chips that are designed and optimized by the cloud operator, it can change how margins evolve, how quickly capacity can be deployed, and how resilient pricing may be during periods of chip supply constraints.

Alphabet did not provide additional details in the supplied material beyond the Yahoo Finance framing of the Evercore ISI projection. Without further disclosure from the analyst or additional primary documentation, it remains unclear whether the 35% figure reflects consensus expectations, a single-house scenario, or a broader set of sensitivities such as competitor offerings and overall cloud spend.

What to watch next is whether Alphabet expands disclosures around its AI hardware roadmap and capacity plans, and whether sell-side research following Evercore’s view confirms similar assumptions for 2027 cloud revenue mix. Any changes in how cloud customers adopt AI services, and any updates on custom chip deployment in production environments, would be the most direct reality check on the forecast.