AI news
AI model reviews
AI tools
How-to guides
Our staffAboutContact

Hottest AI startups in Silicon Valley in 2026: who's raising, and at what price

Anthropic and OpenAI lead the Bay Area by a mile. Behind them, Databricks, SSI, Thinking Machines, Sierra and Harvey are raising billions. Here's the latest funding.

Downtown San Francisco office buildings
Photo: Hardik Pandya on Unsplash

The hottest AI startups in Silicon Valley right now are the two San Francisco frontier labs, Anthropic and OpenAI, followed by a second tier of Bay Area companies raising billions: Databricks, Safe Superintelligence, Thinking Machines Lab, Sierra and Harvey. Anthropic's last round valued it at $965 billion; OpenAI is reportedly seeking around $1.4 trillion. Below that, the region's best-funded young companies sit between roughly $15 billion and $134 billion.

That's the list. It's also a slightly strange one, because two of those "startups" are now worth more than most of the S&P 500. I think the more useful question is which Bay Area companies are growing into those numbers, and which are being priced on hope. This guide sticks to companies headquartered in San Francisco and the wider Bay Area, with funding figures checked against company announcements and reputable reporting in early November 2026. If you want the worldwide picture, including Paris and Stockholm, read our guide to the hottest AI startups globally.

The short answer: who leads in Silicon Valley

Measured by capital raised and valuation, Anthropic and OpenAI lead by a distance that's hard to picture. Then come Databricks and Safe Superintelligence. Each raised $5 billion in the third quarter of 2026. After them you'll find Thinking Machines Lab, Sierra and Harvey, three younger companies valued at $15 billion or more.

Here's the snapshot, using the most recent figure each company announced or that a reputable outlet reported.

CompanyBaseWhat it doesLatest reported roundValuation
AnthropicSan FranciscoFrontier models (Claude)$65B Series H, May 2026$965B post-money
OpenAISan FranciscoFrontier models (ChatGPT)$122B committed, March 2026$852B; reportedly seeking ~$1.4T
DatabricksSan FranciscoData and AI platform$5B, Q3 2026Not confirmed in our sources
Safe SuperintelligencePalo Alto and Tel AvivFrontier research lab$5B, Q3 2026Not confirmed in our sources
Thinking Machines LabSan FranciscoOpen-weight models, custom training (Tinker)Reportedly $5B, November 2026$40B pre-money (reported)
SierraSan FranciscoCustomer-service AI agents$950M, May 2026Above $15B
HarveySan FranciscoLegal AI$550M, September 2026$15.5B

A private valuation is a negotiated price between a company and a handful of investors. It isn't a market price, so read the table as a ranking of investor conviction rather than of product quality.

The background: why the money keeps landing in the Bay Area

An AI startup, in plain terms, is a young company whose main product runs on machine learning, usually large language models. What makes this wave odd is the cost. Training a frontier model takes enormous amounts of chips, electricity and data centre space, so the top labs raise money on a scale you'd normally associate with oil companies or national telecoms, not software.

Crunchbase's numbers for the third quarter of 2026, as summarised by FourWeekMBA, put global venture funding at $159 billion. AI companies took $102 billion of it, or 64%. A record 27 companies raised rounds of $1 billion or more, and the San Francisco Bay Area alone accounted for 24% of all venture money invested worldwide that quarter.

One region. A quarter of the planet's startup capital.

The concentration goes further than geography. When Forbes published its 2026 AI 50 list in April, the 50 companies on it had raised $305.6 billion between them, and OpenAI and Anthropic accounted for $242.6 billion of that, roughly 80%, according to MarketScale's analysis of the Forbes list. So when people talk about "AI tech companies" getting funded, they mostly mean two companies a short drive apart in San Francisco.

Why there? Part of it's history: the researchers who built today's models trained at Google, Stanford and Berkeley, and many left to start companies nearby. Part of it's the money itself, since the investors writing $5 billion cheques (Sequoia, Lightspeed, Greenoaks) are mostly based on or near Sand Hill Road.

I'd guess talent matters most, though. A lab that wants to hire 50 senior researchers in a year has few other places to find them.

Real examples: the Bay Area AI startups to watch

Anthropic

Anthropic, the maker of Claude, raised $65 billion in Series H funding at a $965 billion post-money valuation, led by Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital. The company says its run-rate revenue crossed $47 billion in May.

That's the number I'd focus on, because it suggests the valuation is tied to real sales rather than pure expectation. For a side-by-side look at the two big labs, see our OpenAI vs Anthropic comparison.

OpenAI

OpenAI closed a round in March 2026 with $122 billion in committed capital at an $852 billion valuation. In late September, Bloomberg reported it was seeking at least $30 billion more at roughly $1.4 trillion, as Pulse 2 summarised, with Reuters putting its annualised revenue near $70 billion. Sam Altman has said the company won't go public in 2026. Talks are early, and terms could change. Our OpenAI review covers the products themselves.

Databricks and Safe Superintelligence

Both raised $5 billion in the third quarter, the largest rounds Crunchbase recorded that period. They couldn't be more different. Databricks sells a data and AI platform to enterprises and has years of revenue behind it; Safe Superintelligence, co-founded by former OpenAI chief scientist Ilya Sutskever, has "one goal and one product: a safe superintelligence" and offices in Palo Alto and Tel Aviv. It hasn't released a product. Investors are paying for a research team and a bet, which is a very Silicon Valley thing to do.

Thinking Machines Lab

Mira Murati, OpenAI's former CTO, founded Thinking Machines Lab in San Francisco in February 2025. It raised a $2 billion seed at a $12 billion valuation that July, and Business Insider has reported that Accel will lead a $5 billion round at a $40 billion pre-money valuation, according to Tech Funding News. The lab makes money from Tinker, a service that lets companies train models on their own data, and it released its open-weight model Inkling 2 Small on 5 November. Tech Funding News points out that annual revenue reportedly topped $100 million, which works out at up to 400 times revenue. That multiple made me wince a little.

Sierra

Bret Taylor's Sierra builds AI agents that handle customer service for large companies. It raised $950 million led by Tiger Global and GV in May, pushing its post-money valuation above $15 billion, and it claims more than 40% of the Fortune 50 as customers. Going by its own year-two update, it passed $150 million in annual recurring revenue in February 2026. Our Sierra AI review looks at the product in more detail.

Harvey

Harvey sells AI to law firms and in-house legal teams. Its latest raise was $550 million at a $15.5 billion valuation, co-led by Diffusion and Lightspeed, and the company says 80% of Am Law 100 firms use it.

I find Harvey one of the more convincing stories on this list: lawyers bill by the hour, they're expensive, and they'll pay for anything that saves them real time. Whether that justifies $15.5 billion is another matter.

Cursor: the exit that shows the other path

Cursor, the San Francisco AI code editor made by Anysphere, would've topped most "up and coming AI companies" lists a year ago. It went from zero to $2 billion in annualised revenue in about three years, The Next Web reported. Then SpaceX bought it in a deal reportedly valued at $60 billion, which closed in mid-August, and in October Cursor was bundled into The X Subscription alongside Grok, per The New Stack. It's no longer an independent startup. Our Cursor review covers what changed for users.

Benefits and limitations of the boom

The upside is easy to see. Money at this scale pays for compute, and compute is what turns research into products people use. Anthropic's and OpenAI's revenue figures suggest businesses are paying for those products at scale, so this isn't a boom built only on demos.

The limitations are just as concrete, and I don't think they should be waved away. Concentration is the first: when 80% of the Forbes AI 50's funding goes to two companies, everyone else is competing for scraps and for the same few hundred senior researchers.

Valuation risk is the second. At up to 400 times revenue, Thinking Machines has to grow very fast just to stand still on paper, and it lost three co-founders to Meta and OpenAI between October 2025 and January 2026, Tech Funding News noted.

Third, the customers' bills are climbing; Uber's CTO told a TechCrunch event that the company "blew through our [AI] budget" soon after rolling out agentic coding tools, even though he said results were starting to show.

Here's an analogy that helps me. The Bay Area right now looks a bit like a gold rush town where two mining companies own most of the river and everyone else is selling shovels, maps and insurance to them. The shovel sellers (Sierra, Harvey, Databricks) probably have the steadier business. The miners have the bigger upside, and the bigger hole if the seam runs out.

Practical takeaways for you

If you're a job seeker, these companies are hiring heavily, mostly in person in San Francisco. Pay is high, and our AI engineering salary guide breaks down typical ranges. I'd weigh equity offers carefully, though, since a stake bought at a $40 billion valuation only pays if the company later sells or lists for more.

If you're a buyer choosing tools, funding tells you who's likely to be around in three years. It doesn't tell you who's best for your team. Check pricing, data policies and lock-in. Cursor users found that out in October, when Claude and OpenAI models started costing extra under the new bundle.

If you're an investor without access to private rounds, you're mostly watching from the sidelines until an IPO. OpenAI's ruled one out for this year.

And if you're simply curious about who the top AI developers are, the honest short list is short: Anthropic and OpenAI at the frontier, plus a handful of specialists. Our piece on agentic AI explains the agent technology most of these companies are now selling.

The next six months to watch

A few things are already in motion. OpenAI's reported $30 billion bridge round could close, or slip. Thinking Machines' $5 billion round was reported as close to done in early November, and the full Inkling 2 Small weights are due under an Apache 2.0 licence shortly after the launch, once outside safety testing is finished. Consolidation seems likely to continue too, given how Cursor ended up inside SpaceX.

My guess is that 2027 will separate the Bay Area companies with fast-growing revenue (Anthropic, Sierra, Harvey) from those priced mainly on their founders' reputations.

I could be wrong. Investors have kept paying more for less revenue for two years now, and nothing in the third-quarter numbers suggests they're tired yet. For a wider view of where the technology itself is heading, see our piece on the future of AI.

FAQ

What are the biggest AI companies?

Among private companies, Anthropic and OpenAI are the biggest by a wide margin. Anthropic was valued at $965 billion after its $65 billion Series H, and OpenAI at $852 billion after its March 2026 round, with reports that it's seeking about $1.4 trillion. Both are based in San Francisco. Public giants such as Nvidia, Microsoft and Google are larger still.

Who are the top AI companies?

At the frontier, the top AI labs are Anthropic and OpenAI. In the Bay Area, the next tier includes Databricks and Safe Superintelligence (each raised $5 billion in Q3 2026), Thinking Machines Lab (reportedly raising at $40 billion), and Sierra and Harvey, both valued at $15 billion or more. I'd say Harvey and Sierra look the most grounded in revenue.

Which AI companies in San Francisco are worth following?

Anthropic, OpenAI, Databricks, Thinking Machines Lab, Sierra and Harvey are all headquartered in San Francisco. Safe Superintelligence is down the peninsula in Palo Alto. Cursor's still in the city, but SpaceX has owned it since August.

Is the AI funding boom a bubble?

Nobody knows yet, and I'm wary of anyone who claims to. The frontier labs report large and fast-growing revenue, which a classic bubble lacks. But some valuations run to hundreds of times sales, and the money is very concentrated, so I think a shakeout among second-tier companies is probably more likely than a collapse at the top.

Keep reading