Despite their rivalry in AI, OpenAI will tap Google Cloud for computing power as capacity demands surge.
OpenAI has signed a deal to use Google Cloud’s infrastructure to help meet its skyrocketing computing demands, in a move that underscores how intensifying hardware requirements are redrawing traditional competitive lines in the artificial intelligence sector.
As first reported by Reuters, the agreement—finalised in May—will see Google provide OpenAI with additional cloud computing capacity to train and deploy its AI models, including ChatGPT. The deal marks a notable shift for OpenAI, which has until recently relied almost exclusively on Microsoft Azure as its cloud provider, following Microsoft’s multibillion-dollar investment in the company.
Strategic shift amid compute arms race
The deal is especially significant given that Google and OpenAI are direct competitors: OpenAI’s ChatGPT has become one of the biggest existential threats to Google’s search business, and Google’s DeepMind unit competes head-on with OpenAI in developing next-generation AI models.
Despite this rivalry, the agreement reflects a larger trend in AI infrastructure: as demand for high-performance chips and computing capacity outpaces supply, even the fiercest competitors are being forced into pragmatic partnerships.
OpenAI, which recently revealed an annualised revenue run rate of US$10bn, is pursuing multiple strategies to diversify its infrastructure. These include:
- The US$500bn Stargate initiative with SoftBank, Oracle, and Abu Dhabi’s MGX, with its first megacampus under construction in Texas.
- A US$4bn cloud deal with CoreWeave, building on an earlier US$11.9bn agreement.
- Efforts to design its own in-house AI chips to reduce dependency on third-party hardware.
The collaboration with Google is thus the latest step in OpenAI’s plan to reduce its reliance on Microsoft, with whom it is reportedly renegotiating the terms of its investment, including Microsoft’s equity stake in the company.
Analysts: a win for Google Cloud, with risks
While terms of the Google deal remain undisclosed—including whether OpenAI will use Google’s proprietary tensor processing units (TPUs) or just GPU capacity—it is widely seen as a coup for Google Cloud, which has been seeking to position itself as a neutral infrastructure provider to the fast-growing AI sector.
Alphabet shares rose 1.34% on the news, while Microsoft slipped 0.4%.
Scotiabank analysts described the deal as “somewhat surprising,” highlighting the growth opportunity for Google’s cloud unit but also the strategic risk: Google is now providing critical infrastructure to a company whose products could erode its dominance in search.
Alphabet’s cloud division, which generated US$43bn in 2024 revenue, has already signed deals with other AI challengers, including Apple, Anthropic, and Safe Superintelligence. However, the OpenAI agreement may further strain Google’s already limited chip capacity—an issue CFO Anat Ashkenazi raised in the company’s April earnings call.
