Nvidia In Talks To Buy Inception For $500M+ | Menlo Led Its $50M Seed

Nvidia is in advanced talks to acquire Inception, the Palo Alto maker of Mercury diffusion language models, for more than $500 million plus employee earnouts, The Information reported; neither company has confirmed a deal.

Trace Cohen
By Trace Cohen · Value Add Pulse AI Desk
Founder & Editor-in-chief, Value Add VC
3 min read
Nvidia In Talks To Buy Inception For $500M+ | Menlo Led Its $50M Seed: key figures
ShareXLinkedInEmail

THE RUNDOWN

1

A price above $500 million would be a large exit for a two-year-old startup with more than 30 employees and a $50 million seed.

2

The deal could give Nvidia diffusion language models for its Nemotron line of open models, per Tech Funding News.

3

Nvidia's own venture arm NVentures is already on Inception's cap table, alongside Microsoft's M12, Snowflake Ventures and Databricks.

4

Microsoft and SpaceX had also courted Inception, Reuters reported in May, so the talks show how fast labs buy model teams.

The VC Read

Value Add VC analysis

Our read: at $500M+ for a team of more than 30 led by three co-founders, Nvidia is paying for diffusion research talent and a head start in an unproven architecture. If it is signed, Menlo and the seed syndicate get a fast exit; for founders building alternative model architectures, it shows the buyers are strategics, and the earnout terms will decide how much of the price the team actually keeps.

Analysis

As of November 20, 2026, Nvidia is in advanced talks to acquire Inception, a Palo Alto startup that builds diffusion-based language models, for more than $500 million, The Information reported on November 19, citing people familiar with the discussions. The deal would also include earnout awards for employees tied to performance targets, Tech Funding News reported, and the price and terms could still change. Neither company has confirmed an agreement, and Seeking Alpha carried the same more-than-$500 million figure from the report.

What does Inception do?

Inception makes the Mercury family of diffusion large language models. Mercury generates and refines multiple tokens in parallel rather than one at a time, per Tech Funding News. The company calls Mercury the first commercially available diffusion LLM and prices it at $0.25 per million input tokens and $1.00 per million output tokens, per Inception's own blog. Inception says Mercury runs 5 to 10 times faster than the speed-optimized models of the three big labs at matched accuracy, a company claim that Tech Funding News relays.

Who founded Inception?

Inception was founded by Stefano Ermon, a Stanford professor who serves as CEO, with Aditya Grover and Volodymyr Kuleshov, per Tech Funding News.The deal would bring Inception's three co-founders into Nvidia, Seeking Alpha reported. Tech Funding News describes the company as two years old, with more than 30 employees in Palo Alto.

“Inception says Mercury runs 5 to 10 times faster than the speed-optimized models of the three big labs at matched accuracy, a company claim that Tech Funding News relays.”

How much has Inception raised?

Inception's only round we could verify is its seed. Menlo Ventures led a $50 million round in November 2025, with Mayfield, Innovation Endeavors, Microsoft's M12, Snowflake Ventures, Databricks Investment and Nvidia's NVentures participating, plus angels Andrew Ng and Andrej Karpathy, per Tech Funding News and The Decoder. Tech Funding News says Inception was valued higher earlier this year but gives no figure, so the size of any markup is not known.

Value Add VC data

- Nov 2025, Seed -- $50 million at an undisclosed valuation, led by Menlo Ventures (Tech Funding News)

Comparable Nvidia deals, all per Tech Funding News:

  • Hugging Face -- agreed acquisition for about $12.9 billion in September 2026, expected to close in the first half of 2027
  • Groq -- a reported $20 billion non-exclusive licensing agreement in December 2025, structured with hires of its founder and staff rather than as a conventional acquisition

Why Nvidia wants a diffusion model team

The deal could strengthen Nvidia's Nemotron family of open models and give it more control over diffusion model work, Tech Funding News reported. The talks began before Jensen Huang's resignation as CEO was announced on November 12, 2026, according to The Information as cited by Tech Funding News.

Nvidia reported $115.8 billion of third-quarter revenue and guided to $130 billion, Pulse reported on November 17, so a $500 million-plus purchase is small against its scale.

Inception had other suitors. Reuters reported in May 2026, per Tech Funding News, that Microsoft was in discussions with the startup and that SpaceX had also courted it.

The risks in the Inception deal

The talks are not a signed deal, and the more-than-$500 million figure is reported, not confirmed. Diffusion language models are also unproven at frontier scale: Inception's speed claims are its own benchmarks, and its published pricing competes with small, cheap models rather than frontier systems. With NVentures already on the cap table, the price also sets a mark for one of Nvidia's own venture holdings. A definitive agreement, or a denial, is the next dated event.

ShareXLinkedInEmail

More from the Pulse archive: Artificial Intelligence coverage → · All topics

Key Sources

1 source

THE WIRE in your inbox— Tech, startup & VC news with The VC Read, a few times a week. Free to subscribe, no spam.

No spam, unsubscribe anytime. Sent via Substack · Privacy · or subscribe on Substack ↗