# Applied Compute revenue, valuation & funding

Markdown copy of https://realalpha.app/companies/applied-compute/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Applied Compute was valued at $3.25B in November 2026, when it raised $350M. That month it said it had reached a $100M annualized revenue run rate. It's a private company with no stock ticker, and it has not filed for an IPO.

Checked Nov 8, 2026

## At a glance

- Valuation: $3.25B, set by a $350M round led by Elad Gil in November 2026.
- Revenue: a $100M annualized run rate in November 2026, as the company said. It publishes no audited results.
- Status: private. No stock ticker. It has not filed for an IPO.
- Founded: 2025, by former OpenAI researchers Yash Patil, Rhythm Garg and Linden Li.
- CEO: [Yash Patil](https://realalpha.app/builders/yash-patil/), co-founder.
- Investors: Elad Gil, Kleiner Perkins, Lux Capital, Benchmark, Sequoia, NVIDIA and Blackstone, among others.
- Products: AC2, its Agent Cloud for training and serving open models, AC Inference and Ari, an applied research agent.
- Team: tripled since January 2026, the company said in November, with 85% researchers and engineers.
- On Real Alpha: 3 moments of its people talking, from 1 interviews

## Is Applied Compute public? Can you buy its stock?

No. Applied Compute is a private company, so its shares don't trade on any stock exchange and it has no ticker symbol. It has not filed for an IPO, according to [Forge](https://forgeglobal.com/applied-compute_ipo/).

Its executives have given no public date for a listing.

Its shares are held by its founders, staff and investors, among them Elad Gil, Kleiner Perkins, Lux Capital, Benchmark and Sequoia. No stake sizes have been disclosed. This page reports facts and gives no investment advice.

## Valuation, round by round

Its valuation went from $1.3B in April 2026 to $3.25B in November 2026, more than doubling in seven months.

- Nov 2026 · $3.25B · $350M · Elad Gil · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)
- Apr 2026 · $1.3B · $80M · Kleiner Perkins · [Applied Compute's blog](https://www.appliedcompute.com/company/fundraise)
- Oct 2025 · — · $80M · Benchmark, Sequoia, Lux (investors named; no single lead stated) · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)
- Jul 2025 · — · $20M · — · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)

The Information reported talks at a $500M valuation in September 2025, but the October 2025 round was announced with no valuation. No lead investor has been confirmed for the $20M round in July 2025.

## Revenue

Applied Compute doesn't publish results. Its annualized run rate went from $12.8M in January 2026, as reported, to $100M in November 2026, as the company said, close to eight times.

- Nov 2026 · $100M · Company · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)
- Jan 2026 · $12.8M · Reported · [Shopifreaks](https://www.shopifreaks.com/applied-compute-talks-1-3b-valuation-with-kleiner-perkins/)

A run rate is a month's revenue times twelve. It isn't audited revenue.

## How Applied Compute makes money

Applied Compute helps companies build AI models they own. Its frontier lab veterans and applied research engineers help customers train models on their own data, deploy them at scale and improve them over time, according to [its website](https://www.appliedcompute.com).

Customers pay for both sides of that work, training and serving.

- **Training.** AC2 lets customers train open models and bring what they see in production into the next run. · [Applied Compute's blog](https://www.appliedcompute.com/platform/introducing-ac2)
- **Inference.** Of its customers, 40% use Applied Compute for both training and inference, the company said in November 2026. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)

From Applied Compute's interviews

- [Post-training wins inference because the biggest workloads train first](https://realalpha.app/s/post-training-wins-inference-because-the-biggest-workloads/) · Yash Patil · Unsupervised Learning · Oct 6, 2026

## Products

- **AC2** (private beta August 25, 2026). The Applied Compute Agent Cloud, one platform to train open models, serve them at scale and feed production learnings back into training. · [Applied Compute's blog](https://www.appliedcompute.com/platform/introducing-ac2)
- **Ari.** An applied research agent that works with customer teams in the AC2 training loop. It analyzes results and uncovers failure modes. · [Applied Compute's blog](https://www.appliedcompute.com/platform/introducing-ac2)
- **AC Inference.** Its serving product, which handles more than 1 trillion tokens a day, the company said in November 2026. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)

From Applied Compute's interviews

- [Continual learning is blocked by data-efficient sparse-reward training](https://realalpha.app/s/continual-learning-is-blocked-by-data-efficient-sparse/) · Yash Patil · Unsupervised Learning · Oct 6, 2026
- [RL is a hill-climbing machine, so protect your evals](https://realalpha.app/s/rl-is-a-hill-climbing-machine-so-protect-your-evals/) · Yash Patil · Unsupervised Learning · Oct 6, 2026

## Customers

The company names Microsoft, Shopify, Mercor, Harvey, DoorDash, Cognition, Clay and Decagon as customers, in [its November 2026 post](https://www.appliedcompute.com/company/100m-run-rate-fundraise).

Customers have run more than 1,700 training runs on AC2 since July 2026, it said. AC Inference serves more than 1 trillion tokens a day.

## Leadership

The company said in November 2026 that its team had tripled since January. It said 85% of staff are researchers and engineers, in [its funding post](https://www.appliedcompute.com/company/100m-run-rate-fundraise).

- **[Yash Patil](https://realalpha.app/builders/yash-patil/),** co-founder and CEO. He worked at OpenAI, including on its Codex programming assistant. · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)
- **Rhythm Garg,** co-founder and CTO. He was a member of technical staff at OpenAI. · [Crunchbase](https://www.crunchbase.com/organization/applied-compute)
- **Linden Li,** co-founder and a former OpenAI researcher. · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)

## Who owns Applied Compute?

Its founders, staff and investors. As a private company it doesn't publish who holds what, and no stake has been disclosed.

It had raised $160M in total by April 2026, before its $350M round in November.

- **Elad Gil** led the $350M round in November 2026, at a $3.25B valuation. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)
- **Kleiner Perkins** led the $80M round in April 2026, at a $1.3B post-money valuation. Greenoaks and Neo also took part. · [Applied Compute's blog](https://www.appliedcompute.com/company/fundraise)
- **Benchmark, Sequoia and Lux** backed the $80M raise announced at its launch in October 2025, with angel investors. · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)
- **[NVIDIA](https://realalpha.app/companies/#nvidia), Blackstone, Hanabi Capital, Shopify Ventures and HubSpot Ventures** joined the November 2026 round. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)

## Timeline and latest news

- Nov 2, 2026 · Raised $350M at a $3.25B valuation, led by Elad Gil. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)
- Nov 2, 2026 · Said it had reached a $100M annualized revenue run rate. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)
- Nov 2, 2026 · Said AC Inference serves more than 1 trillion tokens a day. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)
- Aug 25, 2026 · Opened AC2, its Agent Cloud, in private beta. · [Applied Compute's blog](https://www.appliedcompute.com/platform/introducing-ac2)
- Aug 10, 2026 · The Information reported talks to raise at roughly a $3B valuation. · [AI Weekly](https://aiweekly.co/alerts/applied-compute-in-talks-for-3b-round-led-by-elad-gil)
- Apr 8, 2026 · Raised $80M at a $1.3B post-money valuation, led by Kleiner Perkins. · [Applied Compute's blog](https://www.appliedcompute.com/company/fundraise)
- Jan 21, 2026 · The Information reported a $12.8M annualized run rate. · [Shopifreaks](https://www.shopifreaks.com/applied-compute-talks-1-3b-valuation-with-kleiner-perkins/)
- Jan 21, 2026 · The Information reported talks to raise at a $1.3B valuation. · [Shopifreaks](https://www.shopifreaks.com/applied-compute-talks-1-3b-valuation-with-kleiner-perkins/)
- Oct 29, 2025 · Launched publicly with $80M from Benchmark, Sequoia, Lux and angels. · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)
- Sep 2025 · The Information reported it was raising at a $500M valuation. · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)
- Jul 2025 · Closed a $20M round, sources told Upstarts Media. · [SiliconANGLE](https://siliconangle.com/2025/10/30/former-openai-researchers-launch-applied-compute-80m-funding/)

## In their own words, and what followed

- [Post-training wins inference because the biggest workloads train first](https://realalpha.app/s/post-training-wins-inference-because-the-biggest-workloads/) · Yash Patil · Unsupervised Learning · Oct 6, 2026
  > “where you're spending the most money, is actually where you're going to want to go and post-train first”
  Reported after · In November 2026 the company said AC Inference serves more than 1 trillion tokens a day. Of its customers, 40% use it for both training and inference. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)

- [RL is a hill-climbing machine, so protect your evals](https://realalpha.app/s/rl-is-a-hill-climbing-machine-so-protect-your-evals/) · Yash Patil · Unsupervised Learning · Oct 6, 2026
  > “it is surprisingly easy to hill climb on non-verifiable domains”
  Reported after · In November 2026 the company said customers had run more than 1,700 training runs on AC2 since July. · [Applied Compute's blog](https://www.appliedcompute.com/company/100m-run-rate-fundraise)

## Questions people ask

### Is Applied Compute a public company? Can I buy Applied Compute stock?

No. Applied Compute is private and has no ticker, so its shares aren't sold on any exchange, according to [Forge](https://forgeglobal.com/applied-compute_ipo/).

### When will Applied Compute IPO?

It has not filed for an IPO, and its executives have given no public date for one.

### Who owns Applied Compute?

Its founders, staff and investors, including Elad Gil, Kleiner Perkins, Lux Capital, Benchmark, Sequoia, NVIDIA and Blackstone. Stakes haven't been disclosed.

### Who is the CEO of Applied Compute?

[Yash Patil](https://realalpha.app/builders/yash-patil/), a co-founder who worked at OpenAI, including on Codex.

### Who founded Applied Compute?

Former OpenAI researchers Yash Patil, Rhythm Garg and Linden Li, in 2025.

### How much is Applied Compute worth?

$3.25B, set by a $350M round led by Elad Gil and announced in [November 2026](https://www.appliedcompute.com/company/100m-run-rate-fundraise).

### How much revenue does Applied Compute make?

The company said in [November 2026](https://www.appliedcompute.com/company/100m-run-rate-fundraise) it had reached a $100M annualized run rate, up from a reported $12.8M in January 2026.

### Is Applied Compute profitable?

It doesn't publish results, and no figures on its profit or losses have been reported.

### How does Applied Compute make money?

It helps companies train custom models on their own data and serve them at scale, through its AC2 platform and its engineers.

### Who are Applied Compute's customers?

The company names Microsoft, Shopify, Mercor, Harvey, DoorDash, Cognition, Clay and Decagon, in [its November 2026 post](https://www.appliedcompute.com/company/100m-run-rate-fundraise).

## Every moment from Applied Compute's people

### Applied Compute CEO on the Limits of RL, the New AI Hyperscaler & Why Post-Training Wins Inference
Unsupervised Learning · Oct 6, 2026

- [Post-training wins inference because the biggest workloads train first](https://realalpha.app/s/post-training-wins-inference-because-the-biggest-workloads/) · 2 min
- [RL is a hill-climbing machine, so protect your evals](https://realalpha.app/s/rl-is-a-hill-climbing-machine-so-protect-your-evals/) · 2 min
- [Continual learning is blocked by data-efficient sparse-reward training](https://realalpha.app/s/continual-learning-is-blocked-by-data-efficient-sparse/) · 1 min

## People on Real Alpha

- [Yash Patil](https://realalpha.app/builders/yash-patil/), Co-founder & CEO, Applied Compute

## How this page is sourced

Each figure links to the filing, company statement or news report it came from, with its date. Run rates are what reporters were told, not audited results; estimates and unconfirmed reports are marked. Quotes are what each person said on the record, with the minute they said it.