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

List of AI companies based in the United States in 2026: who builds what, and where

A list of AI companies based in the United States, sorted by layer, from Nvidia's chips to Anthropic, OpenAI and the apps built on them.

San Francisco skyline across the bay
Photo: Andrew Whitmore on Unsplash

The AI companies based in the United States that matter most right now are OpenAI, Anthropic, Google, Microsoft, Meta, Amazon and Nvidia, followed by a second tier that includes SpaceXSI (the former xAI), Databricks, Scale AI, Perplexity, Cognition and Harvey. Most of them sit within an hour's drive of San Francisco. If you want a list of AI companies based in the United States that's actually useful, it helps to sort them by what they sell, because a chip maker and a legal chatbot don't have much in common beyond the label.

This week gave us a good reason to redraw the map. On November 19, Anthropic's shares started trading on Nasdaq, and earlier in the month Nvidia lost the founder who'd run it for 33 years. I'd argue the list looks more settled than it is.

A list of AI companies based in the United States, sorted by layer

Here's the core list. I've grouped each company by the layer of the industry it mostly works in, from the chips at the bottom to the apps you open on your laptop.

CompanyHeadquartersLayerKnown for
NvidiaSanta Clara, CAChipsGPUs that train most large models
AMDSanta Clara, CAChipsRival AI accelerators
MicrosoftRedmond, WACloud and appsAzure, Copilot, a large OpenAI stake
AmazonSeattle, WACloudAWS and its own AI chips
Google (Alphabet)Mountain View, CACloud, models, appsGemini, Google DeepMind
OpenAISan Francisco, CAModels and appsChatGPT, Codex, GPT models
AnthropicSan Francisco, CAModels and appsClaude
MetaMenlo Park, CAModels and appsLlama models, Meta AI
SpaceXSI (formerly xAI)Palo Alto, CA (main US office)Models and appsGrok, Cursor
Thinking Machines LabSan Francisco, CAModelsResearch and customisable models
DatabricksSan Francisco, CAData platformsData and AI platform for companies
Scale AISan Francisco, CATraining dataData labelling and model evaluation
PalantirDenver, COEnterprise softwareAI platforms for government and business
PerplexitySan Francisco, CAAppsAI search and the Comet browser
CognitionSan Francisco, CAAppsDevin, the Windsurf editor
HarveySan Francisco, CAAppsAI for law firms
Otter.aiMountain View, CAAppsMeeting transcription
WaymoMountain View, CARoboticsDriverless robotaxis

Count the headquarters and something jumps out. Of these 18 companies, 14 are in California, and 12 of those are in the Bay Area. Only Microsoft, Amazon and Palantir sit outside the state. That's a small sample I picked myself, so don't read it as a census, but I'd guess a longer list would tilt the same way.

For a wider take on who leads and why, our guide to AI companies in the United States covers the bigger picture, and our page on AI chip companies goes deeper on the hardware layer.

The background: money, chips and a lot of data centres

An "AI company" is a loose term. In practice it means a business whose main product depends on machine learning, software trained on large amounts of data to predict, write, sort or generate things.

The US doesn't just have many of these firms; it funds them on a scale nobody else matches. Stanford's 2026 AI Index counted $285.9 billion of US private AI investment in 2025, more than 23 times the $12.4 billion it measured in China. The same report counted 1,953 newly funded US AI companies that year, over ten times the next country. It also found the US hosts 5,427 data centres, more than ten times any other country.

Those numbers explain the shape of the table. Training a frontier model needs thousands of chips, buildings full of servers and investors willing to wait years for a return. Very few places can supply all three at once.

There's a caveat in the same Stanford data, and I think it's the most interesting line in it: the number of AI researchers and developers moving to the US has dropped 89% since 2017, with an 80% fall in the last year alone. Money still flows in. People, less so.

Real examples worth knowing

Anthropic and OpenAI

These two San Francisco labs are the centre of gravity. Anthropic priced its IPO at $83 a share, selling 980 million shares under the ticker ANT, which puts its fully diluted value at roughly $1.5 trillion. OpenAI is still private; it closed a $122 billion round at an $852 billion post-money valuation on March 31. Honestly, that a company founded in 2021 listed above OpenAI's last private price surprised me. You can read our OpenAI vs Anthropic comparison if you're choosing between their products rather than their shares.

Nvidia

Nvidia's chips sit underneath almost everything else on the list. Earlier this month the company announced that Jensen Huang had resigned as CEO and from the board, effective immediately. CFO Colette Kress is interim CEO while the board searches for a permanent replacement, and Harvey C. Jones became the company's first-ever board chair. It's too early to say what that means for the products. Probably very little in the short run.

SpaceXSI, the former xAI

This one confuses people, and fair enough. xAI, the Grok maker, now operates as SpaceXSI inside SpaceX, with offices in Palo Alto, Seattle, Memphis and London. Since SpaceX also bought Cursor, the AI code editor, Grok, Cursor and X Premium now come bundled on one plan. Anyone keeping an old list will need to merge three rows into one.

The less famous middle

Plenty of important US AI companies never show up in consumer news. Databricks, founded in 2013 by the creators of Apache Spark, says more than 20,000 organisations use its platform, including 70% of the Fortune 500. Scale AI sells training data and model evaluations, and lists its own valuation at $29 billion. Then there's Harvey, which counts more than 3,500 customers in over 85 countries, mostly law firms and legal teams. And Otter.ai, the Mountain View note-taker, is the kind of app most office workers have met without thinking of it as an AI company at all (our Otter AI review covers its plans and its privacy lawsuit).

Benefits and limitations of a US-centred industry

For you as a buyer, the concentration has real upsides. Most leading models are American, so English-language support, US payment options and integrations with Microsoft and Google tools tend to come first. Stanford found that in March, Anthropic's top model led the best Chinese model by just 2.7%, so the US lead is real but narrow.

The downsides are just as concrete:

  • A handful of firms control the models that thousands of smaller apps are built on, so a price change at one lab ripples outward.
  • Most of the companies depend on one chip designer, Nvidia, and Stanford notes that one foundry, TSMC in Taiwan, makes almost every leading AI chip.
  • Valuations are high and volatile: the Nasdaq 100 fell about 13% from its early October peak.
  • US data rules differ from the EU's, which matters if you handle European customers' data.

In my view, concentration is fine for users until the day it isn't. When three or four labs set the prices, you've got little leverage.

Practical takeaways for you

If you're picking a vendor, check which layer you're really buying from. An app like Otter or Harvey sits on top of someone else's model, so its quality and its pricing partly depend on deals you never see. That's not bad, but it's worth asking a sales rep which model they use and whether they can switch.

If you're job hunting, the list above is a decent starting map, though I'd look past the giants. The middle layer (data platforms, evaluation, vertical apps for law or healthcare) probably hires more people per dollar raised than a frontier lab does.

And if you're investing, this article isn't advice. Anthropic is now a public stock, OpenAI isn't, and SpaceXSI sits inside SpaceX, so the ways to get exposure to each are completely different.

For the startups behind the big names, our list of the hottest AI startups is a useful next stop.

The next twelve months

Here's what we know. Anthropic is now public, Nvidia needs a permanent CEO, and OpenAI has raised enough private money that a listing seems likely at some point (it hasn't announced one). Here's what I'm guessing. I'd expect more consolidation like SpaceX buying both xAI and Cursor, because the cost of training top models keeps rising and only a few owners can pay it. I also think the falling number of researchers moving to the US is the trend to watch; if it continues, more of the next list may be headquartered somewhere else.

So treat any list of American AI firms, this one included, as a snapshot. A year ago, three of the rows above would have looked quite different.

FAQ

Which is the biggest AI company based in the United States?

By market value, the biggest US AI-focused companies are the tech giants such as Nvidia, Microsoft and Alphabet. Among companies built mainly around AI models, Anthropic and OpenAI lead. Anthropic's IPO priced at $83 a share, valuing it at about $1.5 trillion, while OpenAI's last private round valued it at $852 billion.

Where are most US AI companies located?

Most sit in the San Francisco Bay Area. In the 18-company list above, 12 are in the Bay Area and 14 in California overall. Seattle is the other major hub, home to Microsoft (in nearby Redmond) and Amazon. Smaller clusters exist in New York, Austin and Boston, but they don't host many frontier labs.

Is xAI still a separate company?

No. xAI now runs as SpaceXSI inside SpaceX, and it still builds Grok. Older lists that show it as an independent startup are out of date.

How many AI companies are there in the US?

Nobody keeps an exact count, because the definition is loose. A useful proxy comes from Stanford's AI Index, which counted 1,953 newly funded AI companies in the US in 2025 alone, more than ten times the next country. The total number of US firms using AI in their products is far larger.

Is Anthropic an American company?

Yes. Anthropic is a public benefit corporation headquartered in San Francisco and founded in 2021. Its shares began trading on the Nasdaq Global Select Market on November 19 under the ticker ANT. It makes the Claude models, which you can read about in our Claude AI review.

Keep reading