Research & Tools / Hypergrowth

The odds

Does it happen? Yes, and rarely.

Among companies that make the run, it is fast. 31 companies here have dated milestones bracketing $1M and $20M ARR, and their median interval is eight months.

But that describes the winners, not the average or your chances. The broader benchmarks are more sober. Fewer than 1% of AI-native startups in ChartMogul’s dataset reach $10M ARR within 12 months. While that is eight times likelier than for non-AI peers, it is from a very small base. A separate a16z sample of hundreds of companies the firm had spent significant time with reached a median of just over $2M ARR in year one. That is impressive, but it is a venture-selected sample, not the median of all enterprise AI startups. Even AI-native companies in Bessemer’s Cloud 100, a cohort selected for success, take 5.7 years on average to reach $100M.

Combine those and you get a more defensible finding: the ceiling moved dramatically, and selected venture cohorts are moving faster too. But in the broadest dataset here, fewer than 1% reached $10M within a year.

The ceiling is what everyone sees. Slack took roughly 30 months to reach $100M ARR and was the fastest business app in history. Wiz reset that record at 18 months in 2022. At least seven later companies have reported reaching $100M faster than Wiz, and the fastest cases now cluster near eight months from launch.

So when someone quotes a growth number at you, ask which population it comes from: the ceiling, a selected winners’ cohort, or the baseline. Most bad goals are set from the ceiling while the company is staffed and structured like the median.

Four growth curves

“How fast” is the wrong question on its own, because fast means four different things. I sorted the AI-era companies into four archetypes. The first two names I borrowed from Bessemer Venture Partners; the second two are my own labels.

The first three describe viable growth paths. The fourth is a warning category: early demand accompanied by weak, disputed, or undisclosed evidence of durability. Some companies recover by changing strategy; the point is that the early curve alone does not tell you which will.

Speed is not the signal

Sorting the dataset by archetype produces the most uncomfortable number in this research. Median months from $1M to $20M ARR:

The cautionary tales were the fastest group in the set. Three companies is far too small a sample to call this a finding, and the archetype labels are mine, assigned after the outcomes were known. The narrower conclusion is still useful: in this sample, early slope alone did not distinguish durable growth from a cautionary case.

Which means your job is not to get onto the fastest curve. It is to work out which growth curve you are actually on and build the system that curve needs.

The odds in depth →Four growth curves in depth →

Conditions

When hypergrowth does happen, eight patterns recur across the winners studied here. Four concern the product and four concern go-to-market. They are heuristics, not proven predictors: the dataset contains successful companies, not a full spread. None is new to the AI era, and each has a pre-AI twin.

On the product side: self-propagating proof, so something in the product helps create the next buyer, whether it’s a public artifact like Gamma’s watermark or a measured and shared outcome or reference. Economics that expand, so revenue can grow faster than sales capacity through usage, outcomes, or tiers. A step-function opportunity, because the fastest curves often bend when a capability or market shift changes what is possible. And a margin path, because speed without a credible route to improving unit economics leaves durability unproven.

On the go-to-market side: new budget or buyer autonomy, rather than a line item you must take from an incumbent. A founder as an early channel, often including the new name you give the buyer. Proof before scale, so customer evidence precedes large investments in headcount and paid distribution. And durable revenue evidence, meaning retention you can measure alongside an ARR number you can defend.

Two of those eight behave differently from the rest, and the difference is testable.

The inverse matters too. If every deal requires a committee, if proof must be recreated account by account, if you are displacing an incumbent at renewal, or if you sell into a regulated domain, the growth curve changes. Abridge, Harvey, Legora, and EliseAI all grew quickly through that friction, but as Shooting Stars or Compounders rather than Supernovas.

Conditions in depth →

What to do

My research selected companies whose outcomes were already known, so it cannot tell you whether you will succeed. It can help you figure out what the math requires, which growth curve companies like yours were actually on, and which metrics went with it. So what follows is prescriptive about the questions but not the odds.

Many marketing leaders are being asked to produce a Supernova curve. Test whether that curve is realistic for you by getting the leadership team in a room to answer five questions together, because the necessary evidence and decisions span several functions.

1. What does winning mean here? Not the number, the growth curve. Supernova, Shooting Star, or Compounder. Ask everyone to write it down before the meeting. If the room produces three different words, that is telling, and getting on the same page drives the rest of the agenda.
2. Where will we play? Choose the buyer, use case, budget source, and buying motion before the deal-size math. Are you creating new budget or displacing an incumbent? Can the buyer self-serve, or does every deal require a committee? Then test the choice against the math: $19M in 18 months is about four $250K enterprise deals, 21 $50K mid-market deals, or 3,500 $300 self-serve conversions each month.
3. How will we win, and how will it scale? Why will the chosen buyer choose you, and what lets revenue outgrow seller headcount: usage or credit pricing, account expansion, deal size, an ecosystem, or an installed base? If the room cannot name both the advantage and the scaling mechanism, the Supernova curve is not yet supported.
4. What would have to be true? Take the eight conditions one at a time and sort them together into evidenced, assumed, and missing. This turns a debate about the target into a list the team can test against. Give each assumption to whoever is best placed to settle it, and set a date to reconvene.
5. What will we be measured on? The ARR definition, cohort measurement for activation and retention, gross margin, and net revenue retention. Metrics follow from the growth curve, so agree on them in the same room. A Compounder held to Supernova metrics fails on paper while doing well in the real world.

Then look at the five answers side by side and ask whether they reinforce each other. A Supernova aspiration built on displacement budget, committee buying, and no expansion mechanism is an inconsistent plan. Change the growth-curve assumption, where-to-play choice, or scaling mechanism rather than simply demanding more from all five. The leadership team should decide what needs to change and how.

What marketing brings to that room is specific: the target arithmetic, evidence about the buyer, and the evidence for and against each condition. Marketing can lead positioning, buyer language, brand, and proof distribution. It can co-own packaging, activation, community, and the self-serve entry point with Product and Sales, and bring customer evidence into pricing. Product owns the roadmap and instrumentation; Finance owns the revenue definition and margin model; the CEO owns the ambition and leadership trade-offs.

What to do in depth →