AI has reset everything we knew about the pace, size and scale of company building. At least for the true outliers.
The first: Anthropic raised $65 billion at a $965 billion valuation and filed confidentially for an IPO. Founded in 2021. That puts it on track to cross $1 trillion in roughly five years from founding.
The second: Anthropic is doing somewhere around $47 billion in annualized revenue with about 5,000 employees. That is roughly $9.4 million in revenue per person.
Neither number makes sense against any prior decade of B2B. Put them next to the companies we used to hold up as the fastest, most valuable, most efficient businesses ever built, and the gap is not incremental. It is a different curve.
Time to $1 trillion, by founding year
Here is how long it has taken companies to reach a trillion-dollar valuation:
- Anthropic: ~5 years (founded 2021, ~$965B and climbing)
- OpenAI: ~10 years (founded 2015, ~$852B and climbing)
- Google: 21 years (founded 1998, crossed $1T in January 2020)
- SpaceX: ~24 years (founded 2002, crossed via the $1.25T xAI merger in February 2026)
- Apple: 42 years (founded 1976, crossed $1T in August 2018)
Apple needed four decades and the single most successful consumer hardware product in history. Google needed two decades and a structural monopoly on search. Anthropic is doing it in five years on the back of a model and a coding tool.
One caveat, stated plainly: Anthropic and OpenAI have not formally crossed $1T yet. They are at $965B and $852B and both have filed for IPOs, so 2026 is the likely crossing year for both. SpaceX’s crossing reflects the merged SpaceX and xAI entity, not standalone SpaceX. Even with those asterisks, the compression is the story. Nothing in the prior generation came close to this pace.
They Did It With a Fractional of the Headcount
Speed alone is not the interesting part. Plenty of companies have raised fast and flamed out. The interesting part is that Anthropic and OpenAI got here with a fraction of the headcount their predecessors needed at far smaller revenue.
Revenue per employee:
- Anthropic: ~$9.4M (~$47B / ~5,000)
- OpenAI: ~$5.3M (~$24B / ~4,500)
- Apple: ~$2.5M ($416B / ~164,000)
- Alphabet: ~$2.1M ($403B / 190,820)
- SpaceX: ~$1.1M (~$16B / ~14,000)
Apple and Alphabet are two of the best-run businesses on the planet, and they generate roughly $2 to $2.5 million per employee. Anthropic is at nearly four times that, and it is a five-year-old company.
For context on how far this breaks from the old playbook: when Google crossed $30 billion in revenue, it had around 32,000 people. When Salesforce crossed $30 billion, it took roughly 79,000. Anthropic crossed that same threshold with about 5,000.
Revenue and Headcount Are Partial Decoupled Now. Not Entirely, But Partially
The old model assumed revenue and headcount scaled together. Every dollar of new ARR required more account executives, more support, more managers to manage the managers. Revenue per employee crept up over decades as the business matured. It never stepped up 4x in five years.
Three things broke that link:
- The product is the labor. When the thing you sell is intelligence delivered through an API, you are not adding humans to serve each new customer. The marginal cost of the next million in revenue is compute, not a bigger org chart.
- The cost base is fixed and compressing. The largest input for these companies is compute, and that cost is dropping fast while revenue climbs against it. Inference costs have fallen roughly 90% year over year. Margins expand as revenue scales against a relatively fixed base rather than a growing payroll.
- Lean is now a choice, not a constraint. Anthropic is not at 5,000 people because it ran out of time to hire. It is at 5,000 because that is what it needs. OpenAI plans to roughly double headcount to 8,000 by the end of 2026, and even then it stays far more efficient than any traditional software company at the same revenue.
Even the exits are collapsing
Valuations are not the only thing compressing. Exits are too.
Cursor, the coding tool built by Anysphere, went from founding in 2022 to a $60 billion deal with SpaceX in 2026. About four years. That is the fastest path to an exit of this scale anyone has put up. And the largest tech acquisition ever, too.
Cursor got there the same way: a product that does the work, a $2 billion revenue run rate by early 2026 climbing toward a projected $6 billion by year end, and a team estimated in the low hundreds. A four-year-old company carrying a $60 billion price tag with fewer people than a mid-size B2B firm keeps in one regional sales office.
Wiz is the same pattern from the closed-deal side. Google paid $32 billion in all cash for the cloud security company in March 2026, the largest acquisition of a venture-backed startup ever recorded. Wiz was founded in 2020 and had turned down Google’s $23 billion offer the year before. Six years from a standing start to a $32 billion close, at a multiple north of 30x ARR.
The old benchmark for a fast, enormous exit was WhatsApp: five years to a $19 billion sale, and most of that was user count, not revenue. Wiz cleared $32 billion on real ARR. Cursor’s deal is set at nearly double that. One has closed, one is structured as a right to acquire that SpaceX has not yet exercised, but the prices are set and the timelines run about a third of what they used to.
Tripling Headcount Alone Isn’t The Answer Anymore
I have lived both sides of this. I built and sold two companies the “old” way, albeit in both cases very capital efficiently. We run SaaStr AI today with three humans and a stack of AI agents, and revenue has swung from down 19% to up 47% year over year on that model. The math that used to govern B2B does not govern it anymore.
If your plan assumes you need to triple headcount to triple revenue, you are planning for a company that already looks dated. The teams winning the next decade are the ones whose revenue-per-employee line bends up sharply, not the ones whose org chart grows in lockstep with the top line.
The trillion-dollar timeline collapsing from 42 years to 5 is the headline. The structural change underneath it is that you no longer need 79,000 people to build something enormous. You need a product that does the work, a cost base that compresses as you scale, and the discipline to stay small on purpose.
Watch revenue per employee. It is the clearest signal of which kind of company you are building.
