DensityAI nears $10B valuation with hundreds of millions in funding and AWS deal

A startup that barely existed 18 months ago is now worth as much as some publicly traded semiconductor companies. DensityAI, the AI chip venture founded by former Tesla Dojo engineers, is closing a funding round that values the company at roughly $10 billion, with Andreessen Horowitz leading the deal.

The round is expected to bring in hundreds of millions of dollars in fresh capital. More importantly, DensityAI has locked in a conditional procurement agreement with Amazon Web Services, meaning AWS will buy DensityAI’s chips if they hit certain performance benchmarks.

From Tesla’s basement project to a $10B valuation

DensityAI was founded in 2025 by Ganesh Venkataramanan, who previously led Tesla’s Dojo supercomputer initiative. He brought roughly 20 engineers from the Dojo team with him.

The company’s trajectory since then has been aggressive. DensityAI closed a seed round in September 2024, followed by an early-stage venture round in September 2025. Now, roughly a year later, it’s commanding a valuation that puts it in the upper echelon of private AI companies globally.

The AWS agreement and what it actually means

The conditional purchase agreement with AWS is the centerpiece of DensityAI’s pitch to investors. Under the deal, AWS commits to buying DensityAI’s custom AI chips, but only if those chips meet predefined performance standards.

AWS has been diversifying its chip supply chain for years, developing its own Graviton and Trainium processors while also maintaining its relationship with Nvidia. Adding DensityAI to the mix would give Amazon another option for specialized AI workloads, reducing dependence on any single supplier.

A crowded market with very high stakes

What separates DensityAI, at least on paper, is its focus on building integrated solutions spanning chips, hardware, and software for data centers. The company targets sectors including automotive, robotics, and broader AI applications, areas where the Dojo team’s experience with Tesla’s compute-intensive workloads could provide a genuine edge.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.