Demis Hassabis Turns DeepMind Into Alphabet's $205B AI Engine

Google Cloud revenue reached $24.8 billion in Q2 2026, an 82 percent jump from the same quarter a year earlier, according to Alphabet's Q2 earnings report on July 22. The result was driven by enterprise demand for Gemini AI models built at Google DeepMind, the research organisation Demis Hassabis has led since 2023.

DeepMind Becomes the Engine of Google Cloud

When Google merged DeepMind with its Google Brain unit in 2023 and handed Hassabis leadership of the combined entity, the organisational shift was read primarily as a research play. The Q2 2026 numbers reframe it as a revenue play. Google Cloud grew 82 percent year-on-year to $24.8 billion in a single quarter, and the earnings call transcript released by Alphabet credited Gemini models, custom TPU chips, and expanded data centre capacity, all of which sit inside the DeepMind umbrella.

The commercial pipeline behind that revenue is large. Alphabet reported a $514 billion cloud backlog, up from $106 billion in Q2 2025, and enterprise AI customer acquisition doubled year-on-year. Multiple contracts in the $1 billion-plus range were signed in the quarter. For context, Alphabet's entire 2022 revenue was $282 billion; the cloud backlog alone now exceeds that figure.

Bloomberg reported in February 2026 that Google AI Studio and the Gemini API team formally moved under Google DeepMind, completing a consolidation that puts Hassabis in operational control of both the research pipeline and the developer-facing product layer. The result is an unusually short path from a research breakthrough to a billable cloud service, one that competitors running separate AI labs and cloud units do not have.

A $44.9 Billion Quarterly Capital Commitment

Google Cloud's growth comes at a specific cost. Alphabet spent $44.9 billion in capital expenditure in Q2 2026 alone, more than double the $22.4 billion it spent in Q2 2025, according to Fierce Network's analysis of the earnings release. For the full year, Alphabet raised its capex guidance to as high as $205 billion, up from the $180-190 billion range it had set just one quarter earlier. Management signalled that 2027 spending would increase further.

The investment is concentrated in custom TPU silicon, data centre construction, and network infrastructure. Google's TPU chips, designed inside DeepMind and Google Research and manufactured externally, allow the company to train and serve Gemini models at a cost structure that differs from buying Nvidia H100 or H200 units at market rates. That hardware bet is partly what allows the Alphabet AI business to maintain margins as compute demand scales.

The market reacted to the scale of the commitment: Alphabet shares fell 6.6 percent on July 22 despite Alphabet reporting earnings per share of roughly $9.11, more than three times the year-earlier figure. The gap between profit growth and investor reaction reflects concern that free cash flow turned negative in the quarter, a consequence of the capex surge. Hassabis has not commented publicly on the market reaction, but at the Fortune 500 podcast in February 2026 he stated: "In 10, 15 years' time, we'll be in a kind of new golden era of discovery that is a kind of new renaissance."

Isomorphic Labs: The $2.1 Billion Scientific Arm

Alongside DeepMind, Hassabis oversees Isomorphic Labs, an Alphabet spinout using AlphaFold to design drug candidates. In May 2026, Isomorphic closed a $2.1 billion Series B round, bringing total outside capital raised to approximately $2.6 billion, according to Forbes. Thrive Capital led the round for the second consecutive time. New investors included Abu Dhabi's MGX, Singapore's Temasek, Alphabet's growth fund CapitalG, and the UK Sovereign AI Fund.

Isomorphic's core product is now the Isomorphic Drug Design Engine, referred to internally as IsoDDE. The company says IsoDDE more than doubled the accuracy of AlphaFold 3 on a protein-ligand generalization benchmark, a key test for predicting how candidate drug molecules bind to protein targets. The 2024 Nobel Prize that Hassabis shared was for AlphaFold's foundational protein structure prediction work, which Isomorphic is now applying to the drug design step specifically.

At the World Economic Forum in Davos in January 2026, Hassabis updated Isomorphic's clinical trial timeline, moving the target for its first human patient dosing from the end of 2025 to the end of 2026. No human patient has been dosed with an Isomorphic-designed compound to date. The company's stated ambition, which Hassabis first articulated when launching the spinout, is to "solve all disease." At the India AI Impact Summit in 2026, he framed the broader AI opportunity in similar terms: "It's going to be something like ten times the impact of the Industrial Revolution, but happening at ten times the speed, probably unfolding in a decade rather than a century."

What It Means

Hassabis is simultaneously managing two distinct bets at scale. The commercial bet, Gemini converting Alphabet's $44.9 billion quarterly capex into a $514 billion cloud backlog, is tracking in one direction. The scientific bet, Isomorphic translating a Nobel Prize-winning protein prediction tool into a drug that enters a human clinical trial by year-end, remains unresolved. Most researchers would consider either task a career-defining achievement. Hassabis, uniquely positioned inside Alphabet's infrastructure and budget, is pursuing both at the same time.

Sources

• CNBC: Alphabet earnings takeaways, Q2 revenue beats, GOOGL stock sinks on 2026 capex hike
• Yahoo Finance: Alphabet Inc Q2 2026 Earnings Call Highlights
• Fierce Network: Google Cloud surges, but Alphabet's AI capex problem grows
• Forbes: Isomorphic Labs' $2.1 Billion Fundraise Is The Biggest Bet Yet On AI Drug Discovery
• Nobel Prize.org: Demis Hassabis, Nobel Prize in Chemistry 2024 facts
• Bloomberg: Google Upgrades Gemini Deep Think AI for Advanced Math, Science Research

Editorial standards: every claim is sourced. Tips: [email protected]