The mental model everyone gets wrong
When someone says "unicorn," the mental image is Anthropic at $965 billion. Or OpenAI at $850 billion. Or Stripe at $159 billion. These are the companies that dominate headlines, define the narrative, and shape what founders, employees, and investors think of when they think about billion-dollar startups.
They are not the median. They are the extreme tail of a distribution that includes 938 US private companies worth $1 billion or more. The median unicorn is worth $2 billion. It took its founders six years to get there. Most people have never heard of it. And its story tells you more about what building a valuable company actually looks like than any of the top-15 names put together.
This is what the actual shape of the market looks like — and what founders and employees should be calibrating against instead of the extreme tail.
What the data actually says
The data comes from Stanford GSB professor Ilya Strebulaev's Unicorn Board, an academic dataset tracking US private unicorns as of June 2026. The methodology is transparent. The data is credible. It tells a story most industry publications don't.
938 US private companies at $1 billion or more in post-money valuation. $5.25 trillion in aggregate post-money valuation across the entire dataset. Median time from founding to $1 billion status: 6 years. Median post-money valuation across all unicorns: $2 billion.
These four numbers matter because they're the base rates. Every startup story you read is written about the extreme end of the distribution. The base rates tell you what the middle looks like.
1. The distribution is extreme
The top 15 US unicorns account for a massive share of aggregate value. Anthropic alone is worth $965 billion — nearly 20% of the entire unicorn universe by itself. Add OpenAI at $850 billion and you're at 35%. Add Stripe, Databricks, and Waymo and you're past 45%. The top 15 companies represent close to half of the total value across all 938 unicorns.
The median unicorn is worth $2 billion. That's 480 times smaller than Anthropic. It's larger than most people realize, but nowhere near the top of the distribution.
This matters because mental models built on top-of-distribution companies are wrong. When founders benchmark against OpenAI, they're benchmarking against a company that would represent a 400x outlier if you plotted the full unicorn distribution. When employees imagine "unicorn outcomes," they're imagining Stripe or Databricks, not the median $2 billion fintech startup nobody has heard of.
The distribution has extreme skew. The top of the tail dominates the aggregate value. Everything else — the 900+ companies from the top 30 down to the newest unicorns — makes up the rest.
2. Six years is fast, not slow
The median 6-year timeline from founding to $1 billion is faster than most founders assume when they start their companies. But it's slower than the narrative around Anthropic and OpenAI suggests.
Anthropic was founded in 2021 and hit unicorn status in 2022 — one year from founding to $1 billion. OpenAI took 8 years. Safe Superintelligence, founded in 2024, hit unicorn status the same year. Project Prometheus, founded in 2025, was a unicorn by 2026.
These are outliers. The median unicorn spent six years being unglamorous before crossing the threshold. Six years of building products, hiring teams, raising rounds, missing quarters, adjusting positioning, changing markets. Six years is the story most founders never see in press releases — the middle of the journey between the announcement of the company and the announcement of the $1 billion round.
Six years is fast compared to almost every other industry. It's slower than the current AI cycle suggests. It's the base rate for what a successful venture-backed company looks like when you strip out the outliers.
When someone says "unicorn," picture a $2 billion company you've never heard of that took six years to build. That's what the data actually looks like.
3. The industry mix is broader than the headlines
Popular narrative in 2026 says everything is AI. The unicorn data says something more nuanced.
The industry breakdown of US unicorns: Software (801 companies), AI (355), Infrastructure (222), Fintech (179), Marketplace (165), Hardware (153), Healthcare (115), Cybersecurity (111), Life Sciences (68), GenAI (65). Companies carry multiple tags — so an AI fintech shows up in both categories — but the pattern is clear.
Software dominates. AI is huge. But so are Infrastructure and Fintech and Marketplace. Hardware unicorns (153 of them) are more common than most people realize. Healthcare and Cybersecurity are meaningful categories. Life Sciences is a real category that gets almost no attention in the general startup discourse.
The unicorn universe is not just AI. It's also fintech companies most people haven't heard of. It's cybersecurity companies serving enterprises. It's healthcare platforms in narrow verticals. It's marketplace businesses that quietly hit billion-dollar valuations.
If your mental model of "billion-dollar startup" is dominated by AI, the data is telling you that model is skewed by the current news cycle. There are 583 non-AI-tagged unicorns in this dataset. The industry mix is more diverse than the story.
What this means for calibration
Three practical implications:
For founders. "We want to be a unicorn" is a reasonable ambition but a bad plan. Most successful founders never explicitly targeted unicorn status. They built companies solving real problems and reached $1 billion as a byproduct of good product-market fit sustained over years. The 6-year median tells you the realistic timeline. The $2 billion median tells you what "success" looks like at the middle of the distribution — not the top.
The founders who benchmark against Anthropic or OpenAI are benchmarking against outliers. The founders who calibrate against the median are calibrating against reality.
For employees. If you're joining a Series C startup and hoping the equity grants you're taking will make you wealthy at exit, the base rates matter. Even if the company reaches unicorn status — which is not the base case for any given company — your 0.05% equity in a $2 billion outcome is $1 million pre-preference, meaningfully less after the liquidation preference stack.
The mental model of "join a startup, become rich when it unicorns" is dominated by outliers. The median outcome even for successful startups is more modest than the narrative suggests.
For investors. The power law in venture is real and getting more extreme. If the top 15 unicorns account for $2.4 trillion of the $5.25 trillion aggregate, missing them means missing most of the market's return. But the base rate is that any given portfolio company won't be in the top 15. The math of venture is entirely about not missing the outliers while accepting that most companies won't be outliers.
This is why fund construction matters. This is why concentration matters. This is why "we picked the right ones" is the entire venture business.
What to do about it
Three concrete actions depending on what you're building or doing:
Recalibrate your reference points. Whatever you're building or investing in or holding equity in, benchmark against medians and not against Anthropic. The median tells you what success actually looks like statistically. The tail tells you what's possible in extreme cases.
Model outcomes at the median, not the top. If you're modeling your equity value at a company, use the median unicorn valuation ($2 billion) as your base case for a "successful outcome," not the top-15 examples. Do the math against the preference stack. See what you actually own in that scenario.
Study the median unicorns, not just the top 15. The lessons from the 480x outlier at the top of the distribution are not always transferable. The lessons from the median unicorn — a $2 billion company built over six years in a category most people don't pay attention to — are more relevant to most founders.
The takeaway
The mental model of "unicorn" as $850 billion AI company is a media artifact, not a market reality. The market reality is 938 companies at $1 billion or more, with a median of $2 billion, built over a median of six years, distributed across software, AI, fintech, infrastructure, healthcare, and other categories most people don't associate with unicorn status.
That's the data. That's what the market actually looks like. Calibrate accordingly.
Want to understand the mechanics behind how startups reach unicorn status, what happens on the cap table along the way, and what your equity is actually worth at each stage? Try the Equity & Fundraising track on Gargiulo — 65 scenarios covering SAFEs, cap tables, liquidation preferences, and the math that determines what unicorn status actually means for founders and employees. Sterling has spreadsheets.
Source: Ilya Strebulaev's Unicorn Board (data-driven.vc), Stanford Graduate School of Business, updated June 2026.