Google is paying cloud providers to switch to its AI accelerators

The battle for dominance in the AI accelerator market is entering a new phase. Google no longer relies solely on the technical strengths of its chips: it is now prepared to pay cloud providers extra if they choose its custom accelerators over competitors' products. A strategy that would have seemed unthinkable for a company of this size a couple of years ago now looks like a forced but logical move amid sharper competition.

To understand the context, it helps to recall how the market for AI computing infrastructure has changed over the past two years. NVIDIA still holds the lion's share of the GPU market for training and inference of neural networks. Its H100 and B200 accelerators have become the de facto standard, and delivery queues stretch for months.

In this situation Google, which develops its own line of TPU tensor processors, finds itself in a paradoxical position: it has a technically competitive product, but the ecosystem and market habits work against it. Cloud providers and their customers have spent years building workflows around CUDA and NVIDIA's architecture, and simply offering an alternative chip is not enough: they need a compelling reason to switch.

This is where financial incentives come in. According to available information, Google offers cloud providers various forms of financial support, from direct subsidies to favorable lending terms and discounts on cloud services, in exchange for a commitment to buy and deploy its accelerators. The model resembles the so-called "circular deals" that OpenAI actively practices: the company invests in startups, and they in turn spend the money on cloud computing from OpenAI's partners, primarily Microsoft Azure. The money, in effect, goes around in a circle and returns to the ecosystem, but along the way it creates the appearance of organic demand and strengthens the market positions of everyone in the chain.

Google has evidently decided to adapt this scheme to its own needs. The difference is that the company from Mountain View is promoting not a software platform but specific hardware: its TPUs, which until now were available mainly through Google Cloud. Expanding the presence of TPUs at third-party cloud providers could radically change the balance of power in the market, creating a real alternative to NVIDIA's monopoly and giving developers a choice they badly lack.

However, competitors, above all NVIDIA itself, view Google's initiative with skepticism, and not without reason. First, financial incentives work only as long as the company is willing to sustain them. Subsidies are by definition a temporary measure, and once the cash flow dries up, providers may quickly return to familiar solutions. Second, moving to a new hardware platform is not just swapping one card for another. It means retraining engineers, adapting the software stack and testing compatibility with existing models. The cost of such a transition often exceeds any subsidy.

There is also a third, purely practical problem: the global shortage of components. Even if cloud providers want to buy TPUs en masse, Google may run into manufacturing constraints. The company orders its chips from TSMC, whose capacity is booked years ahead among Apple, NVIDIA, AMD and dozens of other customers. Scaling up TPU supply when every silicon wafer is worth its weight in gold is a nontrivial task.

Nevertheless, Google's strategy deserves attention because it reflects a fundamental shift in the industry. The AI infrastructure market is so overheated that the largest tech companies are willing to literally pay extra to expand their share. This is no longer just competition between products; it is competition between ecosystems and financial resources. Whoever builds the broadest partner network and creates a critical mass of users around its platform will gain a strategic advantage for decades to come.

For end users, meaning developers of AI systems and companies deploying artificial intelligence, this clash of giants brings mostly positive consequences. More competition means more choice, more flexible terms and, ultimately, more affordable computing resources. If Google manages to break NVIDIA's monopoly even partially, the whole industry will win. The only question is whether Mountain View has enough patience and money to see this game through.