G42 and Cerebras to deploy 8 exaflops of compute in India

Eight exaflops. To grasp the scale, consider that just a few years ago reaching a single exaflop was considered a historic milestone in supercomputing. Now Abu Dhabi's G42 and US chipmaker Cerebras plan to deploy a system eight times past that threshold, and not in Silicon Valley but in India.

The partnership, announced in February 2026, is more than another infrastructure deal. It is a geopolitical signal that touches three vectors at once: the Middle East's ambition to be a global technology hub, India's drive not to be left on the sidelines of the AI revolution, and Cerebras's attempt to mount a serious challenge to NVIDIA's dominance in AI computing.

Over the past few years, G42 has grown from a little-known Emirati player into one of the most active investors in global AI infrastructure. Backed by Abu Dhabi sovereign capital and closely tied to the ecosystem around the national AI champion Technology Innovation Institute, G42 has already drawn Washington's attention, and not only in a positive way. In 2024 the company had to wind down a number of projects with Chinese partners under US pressure, after which Microsoft invested $1.5 billion in it, effectively anchoring it in the Western technology bloc. A partnership with Cerebras, an American company, fits that strategy: G42 shows loyalty to Western technology standards while building its own influence.

Cerebras, for its part, gets what it critically needs from the deal: scale. The Sunnyvale company is known for a radical approach to AI chip architecture. Instead of conventional GPUs, it makes so-called wafer-scale processors, giant chips the size of an entire silicon wafer, each containing hundreds of thousands of compute cores. The technology is impressive, but until now Cerebras has remained a niche player, unable to compete with NVIDIA on deployment volume. An 8-exaflop project in India is a bid of a completely different caliber. If the system is deployed successfully, it will be one of Cerebras's largest installations in the world and proof that wafer-scale architecture can work at the scale of national-level data centers.

India is more than a place to host servers in this triangle. A country of 1.4 billion people with a fast-growing technology sector urgently needs its own computing infrastructure for AI. Narendra Modi's government has consistently pushed an "India as an AI nation" strategy, but until now the country has depended heavily on the cloud capacity of Western hyperscalers. Exaflop-class infrastructure of its own changes the situation fundamentally: Indian companies, research institutes and government bodies gain the ability to train and run large AI models without sending data abroad. For a country where digital sovereignty is an increasingly pressing issue, that argument carries as much weight as raw performance.

The broader context matters too. The world has entered an era in which AI computing power is a strategic resource, on par with oil or rare earth metals. Countries and corporations are increasing investment in data centers at unprecedented speed, while access to advanced chips is increasingly regulated through export controls. In these conditions, the role of intermediaries, countries and companies able to redistribute computing resources between regions, becomes key. G42 is deliberately occupying that niche, and the India project is a vivid illustration of the strategy.

The question of execution remains. Deploying 8 exaflops is a colossal task, not only in terms of hardware but also power supply, cooling and engineering infrastructure. For all its ambitions, India faces chronic problems in its energy sector, and securing stable power for a data center of this size will require separate investment. Neither G42 nor Cerebras has yet disclosed details on the timeline, location or financial terms of the project, which leaves room for healthy skepticism.

Still, the partnership itself marks an important shift. The global map of AI computing is no longer a two-pole story about the US and China. The Middle East, India and other regions are joining the race, forming new alliances and supply chains. And if this project goes ahead, it will be one of the clearest examples of AI infrastructure becoming an instrument of geopolitical influence.