Back to the trend of the day: The company behind the Claude AI asks to sell its shares to everyone
Poly Rating · from the trend of the day

AI Companies Valuation vs Revenue 2026: Top 8 · 8 places

Anthropic's $965 billion valuation is only about 14 times its roughly $67 billion annualised revenue, cheaper than OpenAI's 17 times. But it has signed $450–750 billion of computing it must pay for.

  • No. 1: Databricks
  • 8 places

01Final list

The leading AIs each built their own list of ten independently. Here is one result that is easy to check: every place has a reason, a source, and the list of AIs that included it.

Order: from most to least AI agreement.

A bright badge means this AI put the entry on its list. A dim badge means it did not.

  1. 1

    Databricks

    Estimate: $7B run-rate growing 80% against a $190B valuation (~27x). It is free-cash-flow positive, with 15,000+ customers and no single big buyer.

    Included it: Claude, ChatGPT, Meta, Grok, Gemini, Kimi. Did not include: Nemotron, Z.ai.

    In their lists: Claude: 6 · ChatGPT: 8 · Meta: 4 · Grok: 4 · Gemini: 4 · Kimi: 1

    What each AI noted 6
    • Claudeplace 6

      A $7B run-rate with no valuation in the sources I found.

      Source ↗
    • ChatGPTplace 8

      Databricks sells data and AI infrastructure to enterprises, giving it a monetization path that does not depend exclusively on consumer chatbot subscriptions.

      Source ↗
    • Metaplace 4

      $190B on $7B = 27.1x median of AI bubble – only AI data platform growing 80% YoY at $7B scale

    • Grokplace 4

      Data/AI platform at ~$190B valuation on reported multi-billion revenue run-rate, more enterprise software-like than pure model lab.

      Source ↗
    • Geminiplace 4

      The quiet enterprise data foundation commanding a $7B run-rate and $190B valuation without burning capital on consumer chatbots.

      Source ↗
    • Kimiplace 1

      The only AI giant that already pays for itself: free-cash-flow positive on a ~$7B run-rate growing ~80% — the cleanest price-to-proof ratio in AI.

      Source ↗
    6 AIs put it here
    Source ↗
  2. 2

    SpaceX (with xAI)

    $18.7B 2025 revenue, but AI is only $3.2B of it, and that AI arm lost $6.36B. The IPO priced the whole at about 95x sales.

    Included it: Claude, ChatGPT, Meta, Grok, Gemini, Kimi. Did not include: Nemotron, Z.ai.

    In their lists: Claude: 2 · ChatGPT: 9 · Meta: 5 · Grok: 8 · Gemini: 8 · Kimi: 3

    What each AI noted 6
    • Claudeplace 2

      $18.67B of revenue and about $100B of cash cover xAI's $6.36B loss.

      Source ↗
    • ChatGPTplace 9

      xAI combines frontier models with the distribution of X, but its standalone revenue and financing economics are difficult to compare with audited public-company accounts.

    • Metaplace 5

      $230B standalone Jan 2026, $1.25T combined with SpaceX Feb 2026 – $3.5B revenue = 57x most expensive

      Source ↗
    • Grokplace 8

      Smaller reported revenue (~$0.5B range in some 2026 estimates) relative to the larger SpaceX valuation and Anthropic/OpenAI scale.

      Source ↗
    • Geminiplace 8

      Powered by the 100,000-H100 Colossus cluster, reaching a $4.5B run-rate on an $80B valuation tied to X and SpaceX integration.

      Source ↗
    • Kimiplace 3

      One ticker, two businesses: Starlink earns $4.4B of operating profit while the AI segment loses $2.5B a quarter — investors are paying ~93x sales mostly for the promise side.

      Source ↗
    6 AIs put it here
    Source ↗
  3. 3

    Nvidia

    $96.2B of revenue in one quarter and about $5.2T market cap, roughly 14.5x sales. It sells compute rather than owing it, though three customers bring about 44% of sales.

    Included it: Claude, ChatGPT, Meta, Grok, Gemini. Did not include: Nemotron, Z.ai, Kimi.

    In their lists: Claude: 1 · ChatGPT: 3 · Meta: 1 · Grok: 3 · Gemini: 1

    What each AI noted 5
    • Claudeplace 1

      $96.2B of revenue in one quarter, with its three biggest customers taking about 44%, and it still prints profit.

      Source ↗
    • ChatGPTplace 3

      Unlike private AI labs, NVIDIA sells the chips and systems used by the AI industry, making its financial results an observable test of AI infrastructure demand.

      Source ↗
    • Metaplace 1

      $5.2T market cap on $359B revenue = 14.5x – only AI infra that already makes $120B net income

      Source ↗
    • Grokplace 3

      Dominant AI infrastructure supplier with multi-hundred-billion revenue and a multi-trillion market cap that has scaled with proven cash generation.

    • Geminiplace 1

      Extracts 75% gross margins by supplying the picks and shovels, turning $130B+ in revenue into a $5.1T sovereign AI monopoly.

    5 AIs put it here
    Source ↗
  4. 4

    Mistral AI

    Valued at over €21B (Sept 2026), but its revenue is unaudited and disputed: $150M to $900M. That puts its multiple anywhere from about 25x to over 100x.

    Included it: Claude, Meta, Grok, Gemini, Kimi. Did not include: ChatGPT, Nemotron, Z.ai.

    In their lists: Claude: 7 · Meta: 9 · Grok: 10 · Gemini: 10 · Kimi: 10

    What each AI noted 5
    • Claudeplace 7

      Run-rate rose from about $20M to over $400M in a year, and the CEO targets $1B by year end, but the sources disagree on its valuation.

      Source ↗
    • Metaplace 9

      €21B (∼$24.4B) Sep 8 2026 on $0.9B revenue = 50x – Europe's frontier lab at highest multiple

      Source ↗
    • Grokplace 10

      European open-weight model lab with ~$0.4B reported revenue, illustrating the gap between frontier labs and smaller competitors.

      Source ↗
    • Geminiplace 10

      Europe's open-weights champion holds an 80x multiple ($12B value on $150M run-rate), leaning heavily on sovereign European backing.

      Source ↗
    • Kimiplace 10

      Europe's frontier champion: a >€21B valuation on roughly €100-300M of revenue — the purest promise-to-price ratio among major AI companies.

      Source ↗
    5 AIs put it here
    Source ↗
  5. 5

    Scale AI

    An estimated $1.4–2B of revenue at a $25–29B valuation (~15–19x). Meta owns about 49% and hired away its founder.

    Included it: ChatGPT, Grok, Gemini, Kimi. Did not include: Claude, Meta, Nemotron, Z.ai.

    In their lists: ChatGPT: 10 · Grok: 7 · Gemini: 9 · Kimi: 8

    What each AI noted 4
    • ChatGPTplace 10

      Data labeling and AI evaluation provide a different route into the AI economy from building the most expensive frontier model.

      Source ↗
    • Grokplace 7

      Data infrastructure provider at roughly $2B revenue run-rate, supporting model training rather than building frontier models itself.

      Source ↗
    • Geminiplace 9

      The essential data-labeling and RLHF tollbooth, generating a $1.4B run-rate at an 18x multiple on its $25B valuation.

      Source ↗
    • Kimiplace 8

      Worth $29B largely because its biggest customer set the price — Meta paid ~$14.3B for 49% and hired away its founder.

      Source ↗
    4 AIs put it here
    Source ↗
  6. 6

    Cerebras

    $510M of audited 2025 revenue, 86% from two customers. Its value is $40B, $56B or $95B depending on the measure, and the stock is down about 35% since debut.

    Included it: Claude, Meta, Kimi. Did not include: ChatGPT, Grok, Nemotron, Gemini, Z.ai.

    In their lists: Claude: 10 · Meta: 7 · Kimi: 9

    What each AI noted 3
    • Claudeplace 10

      $510M of 2025 revenue, 86% from two customers, and a value of $40B to $95B depending on the source.

      Source ↗
    • Metaplace 7

      $56.4B at IPO May 14 2026 at $185/share fully diluted – wafer-scale chip that signed $10B+ OpenAI deal

      Source ↗
    • Kimiplace 9

      A $24.6B order book against $510M of 2025 revenue: 48x sales priced on delivery risk, and the stock fell ~35% within months of listing.

      Source ↗
    3 AIs put it here
    Source ↗
  7. 7

    Perplexity

    Run-rate tripled to over $750M this year. Its valuation is $20B last closed or $30B+ in talks, about 27–40x sales.

    Included it: Claude, Meta, Grok. Did not include: ChatGPT, Nemotron, Gemini, Z.ai, Kimi.

    In their lists: Claude: 9 · Meta: 8 · Grok: 9

    What each AI noted 3
    • Claudeplace 9

      About $750M of run-rate against a $30B+ valuation (about 40x), and Nvidia is reportedly in talks to invest.

    • Metaplace 8

      $20B reported Sep 2025 on $750M+ run-rate = ∼40x – answer engine in talks at $30B+ with Nvidia

      Source ↗
    • Grokplace 9

      AI search product with sub-$1B reported annualized revenue, high growth but still early relative to the frontier labs.

      Source ↗
    3 AIs put it here
    Source ↗
  8. 8

    CoreWeave

    Revenue heading to $12.4–13.2B this year at a market cap of about $38–60B, the cheapest multiple here. But Microsoft paid 67% of its 2025 revenue.

    Included it: Claude, Kimi. Did not include: ChatGPT, Meta, Grok, Nemotron, Gemini, Z.ai.

    In their lists: Claude: 8 · Kimi: 6

    What each AI noted 2
    • Claudeplace 8

      A $104.2B backlog sits on a roughly $38B market cap, and Microsoft was about 67% of 2025 revenue.

      Source ↗
    • Kimiplace 6

      The AI cloud that went from 72% one-customer revenue to 45% while sitting on a $66.8B backlog — resilient demand, fragile balance sheet.

    2 AIs put it here
    Source ↗

More about the top three ↓

02More about the top three

The ranking is above. Here is more about the three places at the top: what they are, why they are here, and what is worth knowing.

The three are not measured the same way: Databricks uses an estimated yearly run-rate, SpaceX uses 2025 revenue against an IPO price, and Nvidia uses one quarter's revenue against market cap.

1Databricks

No. 1 — Databricks. On the lists of 6 of 8 AIs; median place there: 4. Its price rests on real sales: revenue growing about 80%, positive free cash flow, and no single big buyer.

DatabricksBooth
Raysonho @ Open Grid Scheduler / Scalable Grid Engine · CC0 · Wikimedia Commons

Did you know

  • Databricks bought the open-source AI startup MosaicML for $1.4 billion, according to Wikipedia. The Reuters article it cites says $1.3 billion, so the two sources disagree.

    Source ↗
What it is
A data and AI software platform used by big companies.
Its edge
Pays for itself and grows about 80%, at a lower multiple than SpaceX.
Who should know
Useful for seeing what an AI price backed by real cash flow looks like.
  • Valuation vs revenue~27x revenue
  • Founded2013
  • HeadquartersSan Francisco

Wikipedia · Wikidata

Why it is here

  • Databricks raised $5 billion at a $190 billion valuation. Its run-rate revenue is now past $7 billion, up from $5.4 billion in February. That is about 27 times revenue, a far lower multiple than SpaceX's.

    Source ↗

  • Growth was above 80% year on year in Q2, up from 65% in February. The company was adjusted free cash flow positive over the last twelve months. More than 100 accounts each spend at a $10 million run-rate.

    Source ↗
Did you know

  • Databricks' open model DBRX has 136 billion parameters, but uses only about 36 billion on average to produce an answer. The article says it cost $10 million to create.

    Source ↗
History

  • Databricks was founded in 2013 by the people behind Apache Spark, a data tool built at the University of California, Berkeley. It started as a university research project before it became a company.

    Source ↗
Keep in mind

  • The $7 billion run rate and growth above 80% are company-reported, not audited. At $190B, that is about 27 times revenue, which leaves little room for slower growth.

    Source ↗

  • Databricks has multibillion-dollar cloud commitments with all three big cloud providers, and its AI research is costly. Selling more shares to meet demand may upset older investors who were turned away.

    Source ↗
What people say

  • The interest level was just insane. — Ali Ghodsi, Databricks CEO

    Source ↗
Where to go further

  • Read the trial page first: it gives two weeks of full platform access with up to $400 in free usage. There is also a Free Edition with no expiration date.

    Source ↗

Video: What is Databricks? The Story Behind the Modern Data Platform (Visual Explanation) · Data with Baraa

Compare the top three ↓

2SpaceX (with xAI)

No. 2 — SpaceX (with xAI). On the lists of 6 of 8 AIs; median place there: 6.5. Priced at about 95 times sales while its AI arm loses money, so the price leans on hopes more than sales.

Entrance to SpaceX headquarters
Bruno Sanchez-Andrade Nuño from Washington, DC, USA · CC BY 2.0 · Wikimedia Commons
My tribute to SpaceX from the Apollo astronauts, as seen from the HQ entry today (23585860271)
jurvetson · CC BY 2.0 · Wikimedia Commons

Did you know

  • SpaceX's filing reportedly shows Anthropic, a rival to xAI's Grok, paying about $1.25 billion a month for compute at the Colossus data centers through May 2029. That is roughly $45 billion in total.

    Source ↗
What it is
A rocket company that now includes the xAI and Grok AI arm.
Its edge
Has the biggest sales base of the three AI-linked arms and large cash reserves.
Who should know
Useful for seeing how far a price can run ahead of sales and AI profits.
  • Valuation vs revenue~95x sales
  • Founded2002
  • HeadquartersHawthorne

Wikipedia · Wikidata

Why it is here

  • SpaceX reported $18.7 billion of revenue for 2025 and a $4.9 billion net loss. The IPO was reportedly valued at roughly $1.75 trillion, far more per dollar of sales than Databricks or Nvidia.

    Source ↗

  • The AI part (xAI, X and Grok) lost $6.4 billion in 2025. That is more than the whole company's $4.9 billion net loss. So the AI arm costs more than it earns, unlike Nvidia, which sells the chips.

    Source ↗
Did you know

  • In the first three months of 2026, the AI part of SpaceX spent $7,723 million on equipment and buildings. The article says that is more than the Space and Connectivity segments combined.

    Source ↗
History

  • By 2008 SpaceX had lost three Falcon 1 rockets in a row and was nearly out of money. The fourth launch, on 28 September 2008, reached orbit. Musk said that if it had failed, that would have been it for SpaceX. — (voice only)

    Source ↗
Keep in mind

  • Morningstar values SpaceX at $780 billion, less than half the reported IPO target of $1.75 trillion. Its analyst says the company has been significantly overvalued, and the AI arm's economics are unclear.

    Source ↗

  • Investors worry about founder Elon Musk keeping absolute control through two classes of shares. One report also says xAI burns roughly $1 billion a month building AI data centers.

    Source ↗
What people say

  • We don't see Grok as one of the leading AI labs today — Nicolas Owens, Morningstar analyst

    Source ↗
Where to go further

  • SpaceX's public IPO paperwork (S-1) is on the SEC site. Open the filing folder first. Then look at the AI part, xAI and X: its revenue is small next to its losses.

    Source ↗

Video: SpaceX Makes History | MARS · National Geographic

Compare the top three ↓

3Nvidia

No. 3 — Nvidia. On the lists of 5 of 8 AIs; median place there: 1. Huge real sales and profit at a lower multiple, but three buyers bring about 44% of revenue.

NVIDIA Headquarters
Coolcaesar · CC BY-SA 4.0 · Wikimedia Commons
Nvidia sign
Will Buckner · CC BY 2.0 · Wikimedia Commons

Did you know

  • In 1997 Nvidia had only enough money left for one month's payroll. Jensen Huang's unofficial motto was "Our company is thirty days from going out of business." Sega's $5 million investment then gave the company six months to live, he later said.

    Source ↗ ✓ confirmed by several AIs
What it is
The chip maker whose hardware most AI companies buy.
Its edge
Biggest real sales and profit, at the lowest multiple of the three.
Who should know
Useful for seeing who earns money from AI spending, and the risk of few buyers.
  • Valuation vs revenue~14.5x sales
  • Founded1993
  • HeadquartersSanta Clara

Wikipedia · Wikidata

Why it is here

  • Nvidia's revenue for the quarter ended July 26, 2026 was $96,221 million. A year earlier it was $46,743 million, so one quarter's sales roughly doubled. That one quarter is more than Databricks' yearly run-rate many times over.

    Source ↗

  • Nvidia's filing says that in the first half of fiscal 2027 three direct customers made up 16%, 15% and 13% of revenue. That is about 44% from just three buyers, unlike Databricks, which has no single big buyer.

    Source ↗
Keep in mind

  • Some big buyers are building their own chips, which could mean fewer Nvidia purchases later. Direct customers can also be resellers, so real end-user concentration may look different.

    Source ↗
What people say

  • Customers are anxious and impatient to get their Blackwell systems — Jensen Huang, Nvidia CEO

    Source ↗
Where to go further

  • Nvidia's official annual report (10-K) for the year ended January 25, 2026. Read the business overview first, then the customer risks, then the MD&A money section.

    Source ↗
Claude, Meta, Gemini put it first. What they saw in it

  • Claude: $96.2B of revenue in one quarter, with its three biggest customers taking about 44%, and it still prints profit.

  • Meta: $5.2T market cap on $359B revenue = 14.5x – only AI infra that already makes $120B net income

  • Gemini: Extracts 75% gross margins by supplying the picks and shovels, turning $130B+ in revenue into a $5.1T sovereign AI monopoly.

Video: How Nvidia Grew From Gaming To A.I. Giant, Now Powering ChatGPT · CNBC

Compare the top three ↓

The top three side by side

DatabricksSpaceX (with xAI)Nvidia
In the AIs' lists6 / 86 / 85 / 8
Valuation vs revenue~27x revenue~95x sales~14.5x sales
Founded201320021993
HeadquartersSan FranciscoHawthorneSanta Clara
Put first byKimi—Claude, Meta, Gemini

Places with the same number of AIs are ordered by their average place in the AIs' own lists.

More ratings

03Worth knowing

What the AIs noticed while building the list.

The two labs count revenue differently. Anthropic books the gross value of sales through cloud partners, while OpenAI books only its own share. OpenAI's CRO says this overstates Anthropic's run-rate by about $8B.

Noticed by:
Included it: Claude, Meta, Grok, Kimi. Did not include: ChatGPT, Nemotron, Gemini, Z.ai.
Source ↗

Anthropic's compute bill is reported as $450B locked, $518B, over $600B or up to $750B, with about 80% non-cancellable. That is roughly 7–11 years of revenue at today's run-rate.

Noticed by:
Included it: Claude, ChatGPT, Meta, Grok, Gemini, Kimi. Did not include: Nemotron, Z.ai.
Source ↗

Reliance on a few buyers is common across the sector: Anthropic about 20–25% from two customers, Nvidia about 44% from three, CoreWeave 67% from Microsoft (2025), Cerebras 86% from two.

Noticed by:
Included it: Claude, Kimi. Did not include: ChatGPT, Meta, Grok, Nemotron, Gemini, Z.ai.
Source ↗

Revenue earned per dollar raised: Anthropic $0.23, Databricks $0.16, OpenAI $0.11. OpenAI's figure is down from $0.31, the steepest drop PitchBook tracks.

Noticed by:
Included it: Kimi. Did not include: Claude, ChatGPT, Meta, Grok, Nemotron, Gemini, Z.ai.
Source ↗

04Questions people ask

AI companies valuation vs revenue

In October 2026 AI companies sell for about 3–4x revenue (CoreWeave) and about 14–17x (Nvidia, Anthropic, OpenAI). Databricks is near 27x, and Perplexity and Mistral are at 40x or more.

Company valuation vs revenue

It is valuation divided by annual revenue: 10x means investors pay $10 for each $1 of yearly sales. AI companies trade at 10–50x versus about 4x for public software.

Source ↗
Anthropic vs OpenAI revenue

Anthropic reported $16.3 billion for H1 2026 and about $67 billion annualised in Q3. OpenAI was at about $50 billion annualised in late September. Anthropic counts partner sales gross and OpenAI net, so the gap is smaller than it looks.

Source ↗
Anthropic vs OpenAI revenue growth

Anthropic's half-year revenue rose about 14-fold, from roughly $1.16 billion to $16.3 billion. OpenAI's run-rate roughly doubled, from over $25 billion in February to about $50 billion by September.

05How we counted

In plain words, no formulas.

1The question

Top 10 AI Companies: Revenue They Earn vs. What Investors Say They Are Worth

2Why this topic

Valuation trackers and revenue reports are published separately. Here they meet a customer-concentration measure, so the reader can see which AI companies are valued on money they already earn and which on money they only expect.

3Where it applies

global AI industry, listed and private, Anthropic's S-1 figures included

4What we looked at

  • Annual revenue run-rate, as reported in filings or by the company
  • Latest private or public valuation
  • Valuation divided by revenue
  • Share of revenue from the top few customers
  • Committed spending on computing power compared with revenue
  • SEC EDGAR: Anthropic S-1 — Audited revenue, net and operating loss, customer concentration and compute commitments
  • Reuters: Anthropic IPO close look — Revenue by year (2024 $386M, 2025 $4.59B, H1 2026 $16.29B) and over $600B of infrastructure commitments
  • AI Race Scoreboard — Private valuation, revenue run-rate and valuation-to-revenue multiple for AI labs

5Who ranked

The leading AIs, each building its own list of ten independently and naming a source for every place: Claude, ChatGPT, Meta, Grok, Nemotron, Gemini, Z.ai, Kimi.

6How the places were set

The same entry written differently by different AIs was counted as one. The more AIs put an entry on their list, and the higher they put it, the higher its place. "AI agreement" shows how many of the AIs had it on their own list. Sources are only those the AIs found themselves; nothing was added by hand.

06Each AI's own list

🇺🇸 Claude10 places · 3 notes

The same question, answered by Claude alone. The final list above combines all the AIs.

  1. 1
    Nvidia

    $96.2B of revenue in one quarter, with its three biggest customers taking about 44%, and it still prints profit.

    Source ↗
  2. 2
    SpaceX (including xAI)

    $18.67B of revenue and about $100B of cash cover xAI's $6.36B loss.

    Source ↗
  3. 3
    Alphabet (Google DeepMind)

    No AI dollar revenue is published, but Google's AI revenue is reported to be up 800% a year, from a base that cannot be seen.

    Source ↗
  4. 4
    Anthropic

    Run-rate rose from about $9B at the end of 2025 to about $65B by July, but $518B of compute is committed, about 8x that run-rate.

    Source ↗
  5. 5
    OpenAI

    Its run-rate was cut from nearly $70B to about $50B because of how cloud-partner sales are counted.

    Source ↗
  6. 6
    Databricks

    A $7B run-rate with no valuation in the sources I found.

    Source ↗
  7. 7
    Mistral AI

    Run-rate rose from about $20M to over $400M in a year, and the CEO targets $1B by year end, but the sources disagree on its valuation.

    Source ↗
  8. 8
    CoreWeave

    A $104.2B backlog sits on a roughly $38B market cap, and Microsoft was about 67% of 2025 revenue.

    Source ↗
  9. 9
    Perplexity

    About $750M of run-rate against a $30B+ valuation (about 40x), and Nvidia is reportedly in talks to invest.

  10. 10
    Cerebras (Nasdaq: CBRS)

    $510M of 2025 revenue, 86% from two customers, and a value of $40B to $95B depending on the source.

    Source ↗

What Claude noticed

Anthropic and OpenAI do not count revenue the same way. Anthropic includes the gross value of some sales through partners such as Amazon, while OpenAI books only its share. That is why OpenAI's run-rate was reported at both $70B and $50B.

Source ↗

Customer concentration is common across the sector: Anthropic about 25% (two customers), Nvidia about 44% (three customers), CoreWeave about 67% (one customer), Cerebras 86% (two customers).

Source ↗

Anthropic's compute partners also compete with it. The filing reportedly says hyperscalers increasingly treat it as a competitor, which makes the $518B of commitments both a lifeline and a dependency.

Source ↗
🇺🇸 ChatGPT10 places · 4 notes

The same question, answered by ChatGPT alone. The final list above combines all the AIs.

  1. 1
    Anthropic, United States, 2026

    The company reported $16.3 billion in just six months, but a $965 billion private valuation and hundreds of billions in future infrastructure commitments make the quality of that growth the key question.

    Source ↗
  2. 2
    OpenAI, United States, private, 2026

    A reported $852 billion valuation against a $50 billion revenue run-rate implies roughly 17 times revenue, before accounting for the enormous cost of compute.

    Source ↗
  3. 3
    NVIDIA, United States, publicly listed, 2026

    Unlike private AI labs, NVIDIA sells the chips and systems used by the AI industry, making its financial results an observable test of AI infrastructure demand.

    Source ↗
  4. 4
    Alphabet (Google), United States, publicly listed, 2026

    Alphabet combines an established advertising cash engine with Google Cloud and its own AI models, giving it more diversified funding capacity than a standalone model lab.

    Source ↗
  5. 5
    Microsoft, United States, publicly listed, 2026

    Microsoft disclosed $24.1 billion of fiscal-2026 revenue from commercial arrangements with OpenAI, demonstrating how AI partnerships can become material to a diversified technology company.

    Source ↗
  6. 6
    Amazon, United States, publicly listed, 2026

    Amazon combines a major cloud infrastructure business with consumer commerce, while also serving as an infrastructure partner and investor in frontier AI.

    Source ↗
  7. 7
    Meta Platforms, United States, publicly listed, 2026

    Meta funds AI infrastructure through a large advertising business, so its AI strategy is supported by existing customer demand rather than a standalone model-subscription revenue base.

    Source ↗
  8. 8
    Databricks, United States, private, 2026

    Databricks sells data and AI infrastructure to enterprises, giving it a monetization path that does not depend exclusively on consumer chatbot subscriptions.

    Source ↗
  9. 9
    xAI, United States, private, 2026

    xAI combines frontier models with the distribution of X, but its standalone revenue and financing economics are difficult to compare with audited public-company accounts.

  10. 10
    Scale AI, United States, private, 2026

    Data labeling and AI evaluation provide a different route into the AI economy from building the most expensive frontier model.

    Source ↗

What ChatGPT noticed

Anthropic's first-half revenue of $16.3 billion was approximately 13.5 times its first-half 2025 revenue of $1.2 billion. That growth rate is extraordinary, but a six-month figure should not be presented as audited full-year revenue.

Source ↗

Anthropic's $80.7 billion net loss is not a clean measure of operating cash burn: roughly $72.4 billion reflects non-cash convertible-note fair-value accounting. Operating loss and operating cash flow tell materially different stories.

Source ↗

Stanford's 2026 AI Index reports estimated 2025 compute spending of $16.3 billion for OpenAI and $6.8 billion for Anthropic. These are annual spending estimates, not the same thing as total future contractual commitments.

Source ↗

Microsoft reported $24.1 billion in fiscal-2026 revenue from commercial arrangements with OpenAI. That figure is evidence of the financial scale of the partnership, but it is not equivalent to OpenAI's own total revenue or an independent measure of end-customer demand.

Source ↗
🇺🇸 Meta10 places · 3 notes

The same question, answered by Meta alone. The final list above combines all the AIs.

  1. 1
    Nvidia

    $5.2T market cap on $359B revenue = 14.5x – only AI infra that already makes $120B net income

    Source ↗
  2. 2
    Anthropic

    $965B private on $67B annualized Q3 – 14x gross, but $518B take-or-pay is 11x run-rate mortgage

    Source ↗
  3. 3
    OpenAI

    $852B on ∼$50B run-rate late Sep 2026 = 17x – but $600B+ compute by 2030 = 12x revenue pre-spent

    Source ↗
  4. 4
    Databricks

    $190B on $7B = 27.1x median of AI bubble – only AI data platform growing 80% YoY at $7B scale

  5. 5
    xAI

    $230B standalone Jan 2026, $1.25T combined with SpaceX Feb 2026 – $3.5B revenue = 57x most expensive

    Source ↗
  6. 6
    Waymo

    $126B valuation Feb 2 2026 – robotaxi with $16B raise but minimal revenue, pure future bet

    Source ↗
  7. 7
    Cerebras Systems

    $56.4B at IPO May 14 2026 at $185/share fully diluted – wafer-scale chip that signed $10B+ OpenAI deal

    Source ↗
  8. 8
    Perplexity AI

    $20B reported Sep 2025 on $750M+ run-rate = ∼40x – answer engine in talks at $30B+ with Nvidia

    Source ↗
  9. 9
    Mistral AI

    €21B (∼$24.4B) Sep 8 2026 on $0.9B revenue = 50x – Europe's frontier lab at highest multiple

    Source ↗
  10. 10
    Lovable

    $13.3B Aug 12 2026 on $600M+ run-rate Sep 2026 = 22x – AI coding app that went $0 to $600M in 9 months

    Source ↗

What Meta noticed

Customer concentration falling – Anthropic top 5 from 36% Q1 to 21% Q3, outside top 5 79% Q3 vs 64% Q1, no single >10% nine months Sep 30 2026

Source ↗

$1.3T combined future revenue for hyperscalers – OpenAI ∼$600B by 2030 + Anthropic $518B, 80% non-cancelable, vs combined $80B run-rate

Source ↗

Anthropic overtook OpenAI in enterprise API share Q2 2026 32% vs 25% and in annualized revenue $30B vs $25B Apr 2026, despite OpenAI 1.1B MAU vs 245M MAU

Source ↗
🇺🇸 Grok10 places · 4 notes

The same question, answered by Grok alone. The final list above combines all the AIs.

  1. 1
    Anthropic (private / filing for IPO)

    H1 2026 revenue $16.3B (~1,300% YoY growth) with last private mark $965B; massive but largely non-cancelable compute commitments in the hundreds of billions create leverage and risk.

    Source ↗
  2. 2
    OpenAI (private)

    Annualized revenue reported near $50B (net of partner share) with last round at $852B; gross-vs-net accounting differences complicate direct comparison with Anthropic.

    Source ↗
  3. 3
    NVIDIA (public)

    Dominant AI infrastructure supplier with multi-hundred-billion revenue and a multi-trillion market cap that has scaled with proven cash generation.

  4. 4
    Databricks (private)

    Data/AI platform at ~$190B valuation on reported multi-billion revenue run-rate, more enterprise software-like than pure model lab.

    Source ↗
  5. 5
    SpaceX / SpaceXAI (public)

    AI segment contributed ~$3.2B of 2025 revenue within a larger $18.7B business valued above $2T post-IPO; compute and Starlink provide infrastructure leverage.

    Source ↗
  6. 6
    Cursor / Anysphere (private)

    AI coding tool reported at ~$4B annualized revenue, illustrating rapid application-layer monetization at far smaller scale than the model labs.

    Source ↗
  7. 7
    Scale AI (private)

    Data infrastructure provider at roughly $2B revenue run-rate, supporting model training rather than building frontier models itself.

    Source ↗
  8. 8
    xAI / Grok (part of SpaceX)

    Smaller reported revenue (~$0.5B range in some 2026 estimates) relative to the larger SpaceX valuation and Anthropic/OpenAI scale.

    Source ↗
  9. 9
    Perplexity (private)

    AI search product with sub-$1B reported annualized revenue, high growth but still early relative to the frontier labs.

    Source ↗
  10. 10
    Mistral (private)

    European open-weight model lab with ~$0.4B reported revenue, illustrating the gap between frontier labs and smaller competitors.

    Source ↗

What Grok noticed

Anthropic’s H1 2026 revenue of $16.3B already exceeds its entire 2025 full-year total of $4.59B, showing extreme growth velocity even as compute commitments dwarf current sales.

Source ↗

OpenAI and Anthropic use different revenue accounting (net of partner share vs gross), so headline run-rates are not directly comparable and have produced conflicting investor estimates.

Source ↗

Frontier model labs trade at 15–20x annualized revenue while public SaaS medians are far lower; the gap reflects both growth and the capital intensity of compute.

Source ↗

Anthropic’s multi-hundred-billion compute commitments (largely non-cancelable floors) create both a growth enabler and a fixed-cost risk if demand slows.

Source ↗
🇺🇸 Gemini10 places

The same question, answered by Gemini alone. The final list above combines all the AIs.

  1. 1
    Nvidia (United States, Listed: NVDA)

    Extracts 75% gross margins by supplying the picks and shovels, turning $130B+ in revenue into a $5.1T sovereign AI monopoly.

  2. 2
    Microsoft Corporation (United States, Listed: MSFT)

    Anchored by Windows and Office, Microsoft generates $15B+ in pure AI run-rate while funding the OpenAI compute stack from cash flow.

  3. 3
    Alphabet / Google (United States, Listed: GOOGL)

    Builds its own custom Iron (TPU v6), cutting external GPU dependencies and protecting an $11B enterprise AI revenue run-rate.

  4. 4
    Databricks (United States, Private)

    The quiet enterprise data foundation commanding a $7B run-rate and $190B valuation without burning capital on consumer chatbots.

    Source ↗
  5. 5
    Palantir Technologies (United States, Listed: PLTR)

    Turned sovereign defense and enterprise AIP bootcamps into 40%+ GAAP operating margins at a 33x sales valuation.

  6. 6
    OpenAI (United States, Private)

    Commands the global consumer AI brand with a $60B+ revenue run-rate, balancing an $852B valuation against massive training burn.

  7. 7
    Anthropic (United States, S-1 Registered: Nasdaq ANT)

    Posted $16.3B in H1 2026 revenue to back a $965B private benchmark, but carries $518B in contractual compute commitments.

  8. 8
    xAI (United States, Private / SpaceX Ecosystem)

    Powered by the 100,000-H100 Colossus cluster, reaching a $4.5B run-rate on an $80B valuation tied to X and SpaceX integration.

    Source ↗
  9. 9
    Scale AI (United States, Private)

    The essential data-labeling and RLHF tollbooth, generating a $1.4B run-rate at an 18x multiple on its $25B valuation.

    Source ↗
  10. 10
    Mistral AI (France, Private)

    Europe's open-weights champion holds an 80x multiple ($12B value on $150M run-rate), leaning heavily on sovereign European backing.

    Source ↗
🇨🇳 Kimi10 places · 4 notes

The same question, answered by Kimi alone. The final list above combines all the AIs.

  1. 1
    Databricks (US)

    The only AI giant that already pays for itself: free-cash-flow positive on a ~$7B run-rate growing ~80% — the cleanest price-to-proof ratio in AI.

    Source ↗
  2. 2
    Anthropic (US — S-1 filed, listing as ANT)

    The fastest revenue ramp ever recorded — $1B to ~$65B annualized in 19 months — with a ~$450B compute tab already signed against it.

    Source ↗
  3. 3
    SpaceX / xAI — SpaceXAI (US, listed SPCX)

    One ticker, two businesses: Starlink earns $4.4B of operating profit while the AI segment loses $2.5B a quarter — investors are paying ~93x sales mostly for the promise side.

    Source ↗
  4. 4
    OpenAI (US)

    The biggest brand in AI: ~$25B of run-rate revenue, a projected ~$14B loss this year, and breakeven promised only for 2030.

    Source ↗
  5. 5
    Anduril (US)

    Defense AI with ~$2.2B of real 2025 revenue and diversified government buyers — priced at ~28-48x, not the triple-digit multiples of model labs.

    Source ↗
  6. 6
    CoreWeave (US, listed CRWV)

    The AI cloud that went from 72% one-customer revenue to 45% while sitting on a $66.8B backlog — resilient demand, fragile balance sheet.

  7. 7
    ElevenLabs (UK)

    Voice AI at 22x revenue — a multiple that looks ordinary next to model labs, backed by real, diversified paying usage.

    Source ↗
  8. 8
    Scale AI (US)

    Worth $29B largely because its biggest customer set the price — Meta paid ~$14.3B for 49% and hired away its founder.

    Source ↗
  9. 9
    Cerebras (US, listed CBRS since May 2026)

    A $24.6B order book against $510M of 2025 revenue: 48x sales priced on delivery risk, and the stock fell ~35% within months of listing.

    Source ↗
  10. 10
    Mistral AI (France)

    Europe's frontier champion: a >€21B valuation on roughly €100-300M of revenue — the purest promise-to-price ratio among major AI companies.

    Source ↗

What Kimi noticed

Compute commitments now dwarf revenue across the frontier: Anthropic has locked ~$450B (up to $750B) against ~$67B annualized sales — roughly 7 years of revenue at the current run-rate, payable even if growth stalls, while its own hyperscaler-suppliers increasingly treat it as a competitor.

Source ↗

There is no single 'revenue' language in AI: Anthropic gross-bills sales made via cloud partners while OpenAI books only its own cut — OpenAI's CRO has accused Anthropic of overstating its run-rate by ~$8B, so any 'valuation vs revenue' comparison must be read as approximate.

Source ↗

Capital efficiency quietly separates proof from promise: Anthropic generates $0.23 of ARR per dollar raised, Databricks $0.16 while free-cash-flow positive, and OpenAI just $0.11 — down from $0.31 eighteen months earlier, the steepest single drop PitchBook records.

Source ↗

The strongest resilience signal is boring: companies selling to thousands of payers (Databricks' 15,000+ customers, ElevenLabs' users) or budget-backed governments (Anduril) survive multiple compression far better than firms whose revenue is two customers and whose cost base is a half-trillion of signed compute.

Source ↗