Google Cloud CEO warns AI market will be reshaped in 1 to 2 years
Google Cloud CEO Thomas Kurian said the company's strategy of building chips, data centers, foundation models and products in-house puts it in a stronger position than competitors who rely on external suppliers. In a recent interview, Kurian said Google avoids handing "80 percent of every $1 of revenue to model or chip suppliers," allowing it to invest more heavily in its own infrastructure.
Google Cloud's cloud market share has grown to 14 percent from 7 percent over the eight years since Kurian joined from Oracle. Fourth-quarter 2025 revenue surged 48 percent year-over-year, and the company expects 2026 revenue to exceed $70 billion, up from $43 billion in 2024.
The company remains third in the $418 billion cloud market behind Amazon Web Services and Microsoft Azure. Google also trails OpenAI and Anthropic in chatbots and coding tools.
The economics problem for AI startups
Kurian said the market will face a reckoning as OpenAI and Anthropic burn billions annually while searching for computing power. "These AI suppliers are relying on a private capital market that is becoming saturated," he said. Neither company can go public while losing money, and they cannot keep raising venture capital indefinitely.
"There will be a reshaping of the market within the next 1 to 2 years," Kurian said. "Whether a particular supplier survives ultimately depends on economics."
Google's hardware and software advantage
Google recently released its eighth-generation TPU chips in two configurations: one for AI model training and another optimized for faster inference through increased memory. Kurian said Google's TPU chips and Gemini models outperform Amazon's Trainium chips and Nova models, as well as Microsoft's Maia chips.
According to Epoch AI estimates, Google controls about one quarter of global AI computing power, operating roughly 3.8 million TPUs and 1.3 million GPUs. Kurian said only Nvidia currently competes with Google as a company that produces both AI chips and integrated chip software.
For executives evaluating AI strategy and vendor relationships, read more on AI for Executives & Strategy and Generative AI and LLM.
Your membership also unlocks:
