Newcomer 与 Bessemer 发布首届 AI 100 榜单，Anthropic 登顶

Newcomer 新闻长文· Eric Newcomer·2026-11-13 03:28· 59 分钟前AI 评分59

The AI 100: The 100 Private Companies Defining the Age of AI

AI 导读

Newcomer 与 Bessemer Venture Partners 在 Cerebral Valley AI Summit 上发布首届 AI 100 榜单，评选 100 家最具前景的私人 AI 公司，Anthropic、OpenAI、Databricks 排名前三，Anthropic 在其预期下周 IPO 之前夺冠。

正文 · 原文

As you read these words we’re sharing the inaugural AI 100 on-stage at our Cerebral Valley AI Summit in San Francisco.

These 100 companies, identified and ranked together with our friends at Bessemer Venture Partners, represent what we judged to be the most promising private AI companies today.

It’s a group that in just a few years has come to include some of the biggest movers in the global economy. It’s only fitting that Anthropic, ahead of its expected IPO next week, took first place in the inaugural AI 100.

The AI 100 is in many ways a reinvention of Bessemer’s storied Cloud 100. Gone are names like Celonis, Tanium, and Attentive. In their place are multiple categories of AI companies, with a collective valuation sitting somewhere around $3.5 trillion.

The big researcher-heavy model companies — including Anthropic at #1, OpenAI at #2, and Thinking Machines at #21 — loom large on the list, accounting for only 9% of the names but 55% of the collective valuation. The inference juggernauts (Fireworks at #9, Baseten at #12, Fal at #16, and Modal at #20) also fare well, as do earlier-generation startup wunderkinds that have made the pivot — be it by changing the narrative or changing the business — to perceived AI star (Stripe at #4, Anduril at #7, Ramp at #8, and Canva at #12).

The AI 100 is part art, part science. The team at Bessemer, with a lot of our input, weighed quantitative factors including revenue, valuation and market share alongside less tangible criteria such as people and culture. Bessemer helped to fact-check all the figures we included, which are based on company data and PitchBook.

There will be lots of opinions about who should have been ranked higher or lower, or who was wrongly left out. But we think we’ve landed on a pretty solid who’s who of private AI companies to watch.

Top executives from a number of AI 100 companies are speaking at Cerebral Valley today, including Databricks (#3), Fireworks (#9), Applied Intuition (#13), Mercor (#15), Perplexity (#19), Vercel (#29), Factory (#73), and Flapping Airplanes (#87).

Some takeaways:

Rivalries in the inference category are intense: Baseten, three spots behind Fireworks, is gunning hard. Modal has been on the upswing.
Robotics and world model companies are on the rise, though they’re driven more by investor conviction than revenue. Applied Intuition (#17), as one example, is helping to transform legacy industries into smart robotics companies; Bloomberg reported Wednesday that it’s in talks for a $1.5 billion raise at a $30 billion valuation.
Video has presented surprisingly rich opportunities with ElevenLabs, Fal, Higgsfield, and Runway all ranking highly. Anthropic has totally abandoned video, hoping to save its training runs for its core models, and OpenAI shut down it’s flagship video initiative, Sora. That’s left room for other startups to find their opening.
Coding, of course, looms large. Anthropic is the biggest beneficiary of the cambrian explosion in AI software development, but Cognition, Factory, Lovable, and Replit are right there too.
Legal and customer success remain some of the most promising applications, led by Harvey, Legora, Sierra, and Decagon. One legal startup that we’re excited about, Crosby, is in the vanguard of new AI native services but didn’t make the cut this year; we’ll be watching.
There’s more than a little precarity around all this. The Cloud 100 felt like a safe list, full of companies with predictable recurring revenues and easily understood business models, though that turned out to be an illusion as AI undermined investor confidence in SaaS. Still, labs that don’t have public products, like Safe Superintelligence, and buzzy startups with massive valuations, like Instinct, are even riskier. So are chip and robotics companies whose products can take many years — and dollars — to develop.

Bessemer Venture Partners’ Byron Deeter and Talia Goldberg are presenting their analysis of the list on-stage at the conference as well.

In the Newcomer tradition of superlatives, here are a few standouts from the AI 100.

Newcomer AI 100 Superlatives

Most Speculative Valuation: Prometheus (#33)
- The stealthy lab set out to develop cutting edge AI for physical engineering and raised a pretty penny before revealing any kind of business model. Co-founded by Jeff Bezos, Prometheus has brought in $18.2 billion in funding so far, a good chunk of which came from the Amazon founder himself. Its $41 billion valuation runs steep for a company operating mostly in the shadows, but having Bezos on your team is enough for many investors to jump in early. When its funding news broke in June, it had no public plans for any kind of product launch.
Most Hyped Product: TypeSafe (#52)
- Who hasn’t heard about Jev at this point? TypeSafe’s model makes fast decisions within other programs, and went viral this fall as an efficient alternative to chat-based LLMs. Jev’s rapid rise led TypeSafe to raise an $870 million Series A led by a16z shortly after the model’s public launch. Instinct (#40) is the close runner-up for most hyped product.
Hottest Recent Fundraise: Fal (#16)
- Founded in 2021 by ex-Coinbase and Oracle engineer Burkay Gur and Amazon developer Gorkem Yurtseven, Fal started as a Python infrastructure company and pivoted to hosting and running AI image, video, and audio models. Fal also offers inference and infrastructure pipelines to keep them operating at a fast speed. The team closed a whopping $1.3 billion Series F at the beginning of November as investors bet that inference demand for all kinds of models is only heading upwards.
Best Pivot: Handshake (#48)
- What started out as a job board for college students to network for post-grad jobs is now a destination for humans with expertise in specific fields to find paid gigs evaluating training data for AI models.
Cerebral Valley Loyalist Award: Higgsfield (#58), Exa (#65), Glean (#75)
- Founders from these startups win a shoutout for attending our first-ever Cerebral Valley AI summit back in the Spring of 2023.
Most Unconventional AI Use Case: Halter (#76)
- Most people don’t think of livestock management as an area ready to be disrupted by AI. Yet Halter has proven it possible, with smart collars for cows used to monitor grazing patterns, track the animals’ locations, and create virtual fences for pastures.
Savviest AI Acquirer: Stripe (#4)
- Stripe’s decision to purchase OpenRouter for around $8 billion back in August landed the payments giant a trove of valuable data on which models could be most efficient for running agents to handle different transactions or process customer requests.

来源：Newcomer 新闻长文 · newcomer.co