Everyone’s gotten more efficient in B2B and tech overall, and … the trend isn’t plateauing. It’s continuing, and even potentially accelerating.
A16z just released data showing ARR per employee has essentially tripled at top-performing companies since 2018. The 90th percentile is now pushing $700K ARR per FTE. Even the 75th percentile has nearly doubled to $350K.
Let that sink in.
The Old Hiring and Headcount Benchmarks Are Dead
For years, we all used roughly the same rule of thumb: $200K ARR per employee was “good.” Especially until you approached pre-IPO. Hit that and you were reasonably efficient. Below $150K and you were probably overstaffed. Above $250K and you were running lean.
That mental model is now obsolete.
The data shows top-quartile companies are now generating $350K-$700K per employee. And here’s what’s more interesting: larger companies are getting MORE efficient, not less.
Companies at $250M+ ARR are hitting nearly $500K per employee. That’s the opposite of what traditional scaling theory predicted. We always assumed you’d add layers, add overhead, add complexity as you scaled. The best companies are proving that wrong.
What’s Driving This?
Three things, and they’re compounding:
- First, AI is doing real work now. Not “AI-assisted” work. Actual work. Support tickets that used to require 50 agents now require 15. SDR teams that needed 20 people to hit quota now need 8 with AI handling research, sequencing, and initial outreach. Engineering teams shipping 3x more code with the same headcount.
- Second, the companies that figured out PLG early are now reaping the benefits at scale. When 60-70% of your new revenue comes from self-serve or product-led expansion, you don’t need the same sales army your competitors do.
- Third, remote work killed a lot of middle management. Companies that went remote-first in 2020-2021 and actually committed to it ended up with flatter orgs. Fewer people managing people who manage people.
- Fourth, we now know we can do it. Shopify has held headcount flat for 3 years — and it didn’t hurt growth. We’ve all seen it doesn’t hurt growth to concentrate in fewer, better people. Even at scale.
The Gap is Widening. And That Means The Best Are Pulling Away In Efficiency, Too.
Here’s the part that should concern you if you’re running a median company: the gap between the 50th and 90th percentile has never been wider.
In 2018, the 90th percentile was maybe 2x the median. Now it’s closer to 3.5x.
That means the efficient companies aren’t just slightly better—they’re operating in a completely different reality. They have more cash to invest in product. More margin to survive downturns. More runway to wait for the right outcome.
If you’re running at $150K ARR per employee while your competitors are at $450K, you’re not in the same fight.
The Scoreboard: Who’s Actually Doing This?
Let’s look at the numbers. First, the AI-native companies that are redefining what’s possible:
-
Cursor — $1B ARR with ~300 employees. That’s $3.3M ARR per employee. They hit $100M ARR in 12 months with basically seed capital. Even at scale, they’re running at 7x what the best public SaaS companies achieve.
-
Midjourney — $500M ARR with ~100-160 employees. That’s $3-5M ARR per employee. Zero external funding. Zero marketing spend. Built entirely on Discord with a team smaller than most Series A startups.
-
ElevenLabs — $330M ARR with ~200 employees (projecting 400 by end of year). That’s $1.6M ARR per employee currently — they went from $0 to $200M ARR in under 3 years while staying profitable.
-
Lovable — $200M ARR with ~100 employees. That’s $2M ARR per employee. They hit $100M ARR in just 8 months with only 45 people. The fastest 0-to-100M in European startup history.
-
Replit — $250M ARR with ~150-200 employees. That’s $1.3-1.7M ARR per employee. Grew from $2.8M to $150M ARR in less than a year after launching Replit Agent. A 50x revenue increase with minimal headcount growth.
-
Perplexity — ~$200M ARR with ~250 employees. That’s $800K ARR per employee — and they went from processing 3,000 queries/day in 2022 to 30 million/day in 2025.
-
Gamma — $100M ARR with ~50 employees. That’s $2M ARR per employee. Profitable for 2+ years. They hit $100M with only $23M in total funding and 70 million users generating 1M+ presentations per day. The “PowerPoint killer” that actually makes money.
-
Glean — $200M ARR with ~1,000 employees. That’s $200K ARR per employee — which sounds “normal” until you realize they doubled ARR in just 9 months, indexed 27 billion documents, and now consume 20 trillion tokens per year. Enterprise AI at scale, growing faster than almost any pure-play enterprise software company this decade.
-
Harvey AI — $100M+ ARR with ~350 employees. That’s $300K+ ARR per employee in legal AI. They hit $100M ARR in just 3 years from founding, with 42% of AmLaw 100 firms now on the platform. Usage-based expansion is driving 4x growth in weekly active users.
-
Synthesia — $100M ARR with ~500 employees. That’s $200K ARR per employee — and they hit $1.1M of new ARR in a single day. 70%+ of Fortune 100 companies use their AI video platform. They rejected a $3B acquisition offer from Adobe.
These aren’t outliers anymore. They’re the template.
Now compare to the best public B2B companies—the ones we used to hold up as efficiency benchmarks:
- Salesforce — $480K revenue per employee
- Snowflake — $480K revenue per employee
- ServiceNow — $460K revenue per employee
- Workday — $450K revenue per employee
- Datadog — $430K revenue per employee
- CrowdStrike — $410K revenue per employee
- HubSpot — $320K revenue per employee
These are excellent numbers by historical standards. ServiceNow and Salesforce are operating at elite efficiency for their scale. But Cursor is running at 7x their efficiency. Midjourney at 10x.
The gap between “great traditional SaaS” and “AI-native” is a full order of magnitude.
What This Means for Hiring in 2026/2027
Every hire needs to be scrutinized differently now. The question isn’t “can we afford this person?” It’s “will this person generate or enable at least $400K in ARR?”
For some roles, that math is obvious. A great AE with a $1M quota and 80% attainment? Easy yes.
For others, you need to think harder. That extra product manager—will they ship something that moves revenue enough to justify the slot? That marketing hire—will they generate pipeline that actually converts?
And for roles that AI can do 80% of? You probably shouldn’t be hiring a human at all.
Great is $500k-$700k in ARR Per Employee
The new baseline for “good” is $300K ARR per employee. The new target for “great” is $500K+.
If you’re not there, it doesn’t mean you’re doomed. But it does mean you need a clear plan to get there. What can AI take over? What roles can be consolidated? What layers can be removed?
This isn’t about being cheap. It’s about survival. The efficient companies will outlast and out-invest the bloated ones. That’s always been true in SaaS—it’s just that the definition of “efficient” has changed dramatically.
And it’s not going back.
