Stock Markets July 24, 2026 10:11 AM

Barclays Says TPU-as-a-Service Could Meaningfully Lift Alphabet’s 2028 Profits

Broker projects a roughly 15% boost to 2028 gross profit and substantial operating income upside if TPU sales scale as expected

By Marcus Reed
Share
Twitter Reddit Facebook LinkedIn
GOOGL

Barclays projects that Alphabet’s planned TPU-as-a-service offering could be a material driver of profit and revenue by 2028. The bank expects the initiative to initially support Anthropic, broaden compute availability across AI developers, and enable Google to sell Tensor Processing Units directly, improving capital efficiency and expanding operating income.

Barclays Says TPU-as-a-Service Could Meaningfully Lift Alphabet’s 2028 Profits
GOOGL
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • Barclays projects TPU-aaS could raise Alphabet’s 2028 gross profit by about 15% and substantially increase operating income.
  • The model would begin by supporting Anthropic and could expand to other AI developers, allowing Google to sell TPUs directly rather than solely via Google Cloud.
  • Barclays estimates roughly $250 billion in TPU-aaS revenue by 2028 and more than 11 gigawatts of external TPU capacity if deployments scale as expected.

Barclays is forecasting that Alphabet’s move to offer Tensor Processing Units (TPUs) through a TPU-as-a-service (TPU-aaS) model could materially enhance the company’s financial profile by 2028. The brokerage estimates the initiative could increase Alphabet’s gross profit in 2028 by about 15% while also lifting operating income by a substantial amount.

Under the plan described by Barclays, Google would extend access to its TPUs beyond the traditional Google Cloud delivery channels and sell the hardware directly to external AI developers. The bank highlighted Anthropic as the initial customer for the program, and said deployments could later expand to other AI labs and developers.

Barclays argued the TPU-aaS approach would broaden AI compute capacity across the industry and improve Alphabet’s capital efficiency. The firm described the shift as a move toward a more asset-light infrastructure model, with the potential to alter how AI firms access and control compute resources.

The broker raised its forecasts for Google Cloud revenue and operating income after incorporating the TPU-aaS opportunity into its financial model. Barclays kept an Overweight rating on Alphabet and maintained a $425 price target, saying that investor expectations do not yet fully reflect the potential of the new offering.

Barclays also set out estimates for the scale of the opportunity. If deployments scale as the bank anticipates, it estimates that TPU-aaS could generate roughly $250 billion in revenue for Google by 2028, with external TPU sales exceeding 11 gigawatts of capacity. The firm said this outcome could represent more than one-third upside to the current 2028 revenue consensus and materially enhance Google Cloud’s long-term growth profile.

The brokerage pointed to ecosystem benefits stemming from the TPU-aaS model. AI labs such as Anthropic would gain more direct control over computing infrastructure and reduce reliance on hyperscaler cloud agreements. Meanwhile, partners in the broader ecosystem - including Broadcom, Blackstone and infrastructure provider Fluidstack - could see increased demand for hardware components, financing solutions and data center management services.

Barclays’ view positions TPU-aaS as a potentially significant strategic and financial lever for Alphabet, contingent on the scale and pace of external deployments. The bank’s updated forecast adjustments and maintained positive rating indicate it sees meaningful upside to consensus expectations if the program realizes the scale it projects.

Risks

  • The financial upside relies on TPU deployments scaling as Barclays expects - if deployments fall short the revenue and profit gains would be reduced (impacts Google Cloud, AI infrastructure suppliers).
  • The shift to selling TPUs directly presumes adoption by AI labs and developers - slower or limited customer adoption would constrain projected benefits (impacts AI developers and hardware partners).
  • Ecosystem partner benefits depend on expanded demand for hardware, financing and data center services - variability in that demand could alter projected gains for suppliers like Broadcom, Blackstone and Fluidstack (impacts hardware and infrastructure sectors).

More from Stock Markets

Scribe Therapeutics Shares Rally Sharply in IPO Debut as Pharma Backing Boosts Demand Jul 24, 2026 Tech Giants Urge Lawmakers to Preserve Open-Source AI Models Jul 24, 2026 Piper Sandler Starts Coverage on Integrated Oils, Favors Chevron Over Peers Jul 24, 2026 Disney options action concentrates on July calls as volume surges Jul 24, 2026 SanDisk, American Express Lead Swings Among Friday Market Movers Jul 24, 2026