B2B 软件增长分化：仅 7 家超 30%，Figma 的 AI

Seven public B2B software companies are growing faster than 30%. Here they are, most recently reported quarter, straight from the earnings releases.

Annualized is that quarter’s revenue times four. It’s a rough figure, and for companies with seasonality it will be off by a few points, but it puts everything on the same footing as the run-rate numbers later in this post.

That’s the list. Not the top of the list. The list.

What This Number Used to Be

The SaaS Capital Index median growth rate peaked above 30% in 2021. Above. Half the index cleared what seven companies clear today.

In five years, the median became the 90th percentile.

That’s the whole story of this market. If you’re the CEO of a $200M ARR B2B company growing 18%, you are not underperforming your peers. You are your peers. The peer set moved, and it moved fast.

An independent read of the SaaS Capital Index data file for June 30, 2026 found the same shape across the 58 constituents reporting both growth and a multiple: 18 companies growing under 10%, 23 at 10% to 20%, 11 at 20% to 30%, and only 6 above 30%. Different universe than ours, nearly identical answer.

The Cluster Right Below the Line

The more interesting finding is who missed, and by how little.

Put the two tables side by side and one thing jumps out: scale is not what separates them. Atlassian is the second-largest company on either list at roughly $7 billion annualized, and it missed the line by two points. Samsara made the list at $1.9 billion. Snowflake cleared it at $5.6 billion while CrowdStrike missed at $5.5 billion.

These are not struggling companies. Several are among the best-run businesses in software. They are also all sitting in a band that earns a 5.5x median revenue multiple, while the sub-10% band earns 1.9x and the 10% to 20% band earns 3.1x.

So there is a dense, high-quality cluster of companies four to seven points below a line that separates one valuation regime from another. That’s the actual competitive situation in B2B right now. A handful of points of growth is worth more than it has been in a decade.

What the Seven Have in Common

Four of the seven do not primarily charge by the seat.

Palantir, Datadog, Cloudflare, and Snowflake all bill against usage in some form. When a customer runs more AI workloads, their bill goes up automatically. Nobody has to negotiate a seat expansion. Nobody has to convince a CFO that the team grew. The AI boom flows through the pricing model without a sales cycle.

Cloudflare’s CEO framed the quarter around exactly this: the shift to AI answer engines and agent-driven commerce as a rewrite of the internet for machine-to-machine traffic. That’s a company whose revenue goes up when machines do more work. Datadog said the same thing in different words, that customers are building and deploying with AI and using the platform to observe and secure it. Snowflake sits under the data those agents read.

Rubrik and Samsara are the two that don’t fit the consumption pattern cleanly. Both sell into a specific structural shift instead: Rubrik into data security and recovery for AI deployments, Samsara into digitizing physical operations. Different mechanism, same principle, which is that the thing driving spend isn’t headcount.

Figma Is the One Worth Studying

Figma is the exception on this list, and it’s the most useful company on it for anyone reading this.

Figma sells seats. It is the most seat-dependent business among the seven. Its revenue grew 48% in the quarter ended June 30, its third consecutive quarter of accelerating growth, with net dollar retention at 136%.

The mechanism is that it added a consumption layer on top of seats rather than replacing them. Q2 was the first full quarter of AI credit monetization. Customers expanded on both dimensions, seats and AI credit add-ons. The CFO noted roughly two-thirds of customers above $10,000 in ARR added full seats at renewal. One large technology customer added more than 25,000 paid seats through an AI credit add-on.

The AI product didn’t cannibalize the seat count. It pulled seats along with it. That’s the mechanism most B2B companies can actually copy.

It isn’t free. Gross margin fell five points year over year, and management was direct about why: they don’t charge for products in beta, so they carry the inference cost with no offsetting consumption revenue. Q3 guidance implies 36% growth, a step down from 48%, partly on tougher comparisons.

That’s what the transition looks like from inside. Accelerating top line, compressing gross margin, and an honest acknowledgment that the comps get harder. It is not clean. It’s still the clearest playbook on this list.

Now Put a Second Generation Next to It

Everything above is one generation of companies. Here is the other one.

Anthropic’s run rate went up sevenfold in a year. Its Q2 revenue exceeded $11.5 billion against $787 million in the same quarter a year earlier. That is more than fourteen times, at multibillion-dollar scale, with positive adjusted operating income in the quarter.

Higgsfield is the one worth staring at. The platform did not exist before March 2025. It crossed a $500M annualized run rate in June 2026 and $700M by August, when it raised $400M at a $5.4B valuation. It was cash-flow positive on the way through, with 390 of the Fortune 500 as customers.

Harvey matters for a different reason: it sells seats. Per-lawyer licenses to law firms, the most traditional B2B motion on this page. It roughly doubled ARR in seven months. Whatever is happening here, it isn’t only a pricing-model story.

The slowest company in this group grew faster over seven months than all but one of the seven public companies grew over twelve.

Databricks and Stripe: The Almost-Public Comps

The objection to the list above is that those are small companies compounding off small bases. Fine. Look at the two that are neither small nor public.

Databricks hit $6.9 billion in annualized revenue in June 2026, growing over 80% year over year, up from $5.4 billion in its fiscal Q4. It then crossed $7 billion and raised $5 billion at a $190 billion valuation. AI products alone account for roughly $1.7 billion of annual revenue.

At roughly $7 billion, Databricks is bigger than every public company above except Palantir and Atlassian, and it is growing faster than all of them. It is Atlassian’s size, growing at three times Atlassian’s rate.

CEO Ali Ghodsi was direct about the mechanism, and it’s the same one running underneath four of our seven. Consumption plus agentic AI. The agents generate far more queries, the agent platform itself generates revenue, and the result is more consumption of everything. He also volunteered that gross margin is going lower, because those queries cost money to serve.

Stripe is the counterweight, and the more instructive one for most readers. Revenue reached $6.8 billion in 2025, up 33%, its fastest growth since 2021 at a revenue base more than four times its 2021 size. Free cash flow rose 52% to $3.2 billion. Total payment volume hit $1.9 trillion, up 34%. Q1 2026 alone did $2 billion in revenue. The February 2026 tender valued it at $159 billion.

Stripe is not an AI-native company. It’s fifteen years old, and it grew 33% by sitting underneath the companies that are. It processes payments for the AI labs and for the long tail of Replit, Lovable, Vercel, Cursor, and Midjourney. It’s taxing the boom rather than competing in it.

You don’t have to be an AI-native company or rebuild into one. You can be the thing they all have to buy.

Two Clouds, and a Gap Where the Middle Used to Be

Plot both generations on one chart, revenue against growth, and the picture is not a spectrum. It’s two clouds.