Nvidia Vera: Modal Tests Show Up to Double AI Task Throughput

Cloud-computing company Modal plans to add Nvidia Vera CPUs for AI-agent workloads, supporting Nvidia’s expansion beyond graphics processors. Modal says early benchmarks showed up to twice the throughput under concurrent load.

What doubled in the tests?

Modal’s October 19, 2026 technical post compares completed tasks per unit of time against an equivalent x86 GNR machine in its benchmark. The result does not mean every individual task became twice as fast, or that all workloads will behave identically.

Serial tests on an unloaded system showed more mixed advantages. The clearer gains emerged as concurrent work increased. Supporting Arm requires software changes, and Reuters reports that Modal will retain Intel and AMD hardware in a mixed fleet.

A growing customer, with no disclosed deal value

Reuters reports that Modal raised $750 million at a $15.75 billion valuation and said it was on track for $1 billion in sales over the next year. Those figures describe the customer’s financing and outlook, not Nvidia’s revenue from the arrangement.

Neither a financial value for the agreement nor a processor quantity has been disclosed. A direct contribution to Nvidia’s profit or margins therefore cannot be calculated from this announcement.

The wider CPU opportunity

Nvidia launched Vera in March 2026 for integrated systems and standalone CPU deployments. In its May earnings call, Jensen Huang clarified that the cited $20 billion figure concerned standalone CPUs. It provides context for the wider business outlook, rather than realized revenue from Modal.

What NVDA investors can monitor

Market News analysis: the commercial test is whether benchmark advantages lead to repeat orders. Order volumes, deliveries, total operating costs and competing processors will matter alongside measured performance.

Vera could broaden revenue sources, but this announcement alone establishes neither a stock-price direction nor a guaranteed earnings benefit. Follow Market News for coverage connecting semiconductor technology with business results.

Related coverage: Microsoft, Nscale and the power constraints on AI data centres, and AI spending and the test of financial returns.

اقرأ المقال بالعربية