Figma’s IPO: The Full Breakdown & Why Melio’s $2.5BN Acquisition is “Discouraging”
Summary
Figma’s filing supports a $20–30 billion debut, but even $20 billion is already expensive. Revenue reached $821 million, up 46% year on year, with $1.5 billion of cash, no debt, and a recent free-cash-flow margin above 40%—roughly a “rule of 80” profile tracking toward $1 billion of revenue. Adobe’s failed $20 billion offer now looks strategically sound, although accepting cash two years earlier probably would have produced the better IRR.
Venture liquidity is returning through fewer, much larger winners, not a broad reopening. Figma and Scale could return roughly $3.5 billion to Index across Wiz and Scale in a compressed period; Harry said Index and Founders Fund can produce 3x DPI on unusually large funds. Longer private holding periods mean an early winner can return $2 billion rather than $400–500 million, but “in the end, you got to be right.”
Melio’s $2.5 billion sale is both meaningful liquidity and an uncomfortable clearing price for merely excellent software companies. Xero presented Melio at $153 million of revenue growing 127%, while Jason later noted $187 million in March and suggested it might have exceeded $200 million when signed. Rory called the outcome “discouraging,” while Jason questioned whether the growth rate was sustainable. Its $4–4.5 billion private marks and $650 million preference stack also show why late investors may accept 1x while founders and early holders remain highly price-sensitive.
The AI spending boom may be strategically unavoidable without being financially rational. Harry questioned whether $300–400 billion of annual capex will earn a positive NPV over two or three years, distinguishing that test from the game-theoretic need to “show up at the party” if AGI arrives. At some point humans will overinvest; the uncertainty is whether that happened last year, happens now, or remains two years away.
For existing software companies, adding AI features is insufficient—the panel’s test is whether AI has already reaccelerated growth. Aggregate growth across roughly 300–400 B2B companies, including AI leaders, was described as almost unchanged from 12 months earlier, implying vendors are still stealing a largely fixed technology budget. Harry’s cutoff was brutal: “If Oracle can get AI native by June 30th and your portfolio startup can’t… I’d give up.”
The Scale transaction appears to have redistributed a roughly $1 billion revenue pool to competitors. With its 49% owner recruiting Scale’s best people and rival cloud providers unlikely to share confidential data, one panelist called the remaining company an “empty husk,” and Harry agreed that the business had effectively moved elsewhere. Surge AI, reportedly already near $1 billion of bootstrapped revenue and now raising $1 billion at $15 billion, became the episode’s clearest example of hidden market share suddenly becoming financeable.
Private equity will not rescue the backlog of subscale software assets until buyers’ 2x expectations meet sellers’ 5–6x hopes. Couchbase’s $1.5 billion sale—about 5.7x its $215 million of revenue despite 12% growth and no profit—looked like a thesis-specific “rifle shot,” not a market-clearing precedent. Roll-ups can turn twenty stranded $100 million businesses into a liquid platform, but “price clears all markets,” and AI is making some legacy assets look less salvageable.
The winner-take-most curve is accelerating founder exits, talent hopping, and entry-price inflation. The panel cited 2,221 CEO departures, up 24%, as 4–6-year founder journeys stretch to 12–13 years while enormous AI compensation makes even strong companies feel second-tier. That opportunity cost reached venture pricing too: Rory described losing a Series 8 at $5 million of ARR that moved from roughly $225 million to $600 million.
Deep dive
1. Figma is approaching $1 billion at a rule-of-80 profile
Rory’s late-night reading of the S-1 found $821 million of revenue, 46% year-on-year growth, $1.5 billion of cash, no debt, and free-cash-flow margins above 40% in the latest quarter. His phone-screen arithmetic: “rule of 80,” with the company clearly tracking toward $1 billion.
The valuation debate was therefore not whether Figma receives a strong reception, but whether the market chooses $20 billion, $25 billion, or $30 billion. Yet Rory stressed that even $20 billion is roughly 20x current revenue—hardly cheap beside established enterprise leaders.
Jason’s pushback on celebrating the failed Adobe deal was precise: cash two years earlier, without a lock-up, probably produces the superior IRR because “a bird in the hand” arrived sooner. But remaining independent was “a much better way to live your life” for the team and possibly the investors.
Adobe’s proposed $20 billion now looks prescient rather than profligate. Figma had expanded beyond designers into developers and nearly everyone building software products; roughly 30% of users were developers, while only 30–40% were designers. Jason’s broader M&A lesson was that capable incumbents should buy billion-dollar revenue assets they can keep growing.
2. A departed deal sponsor can leave a portfolio company financially orphaned
Harry connected the departure of Scott Belsky, who had returned to Adobe as chief product officer, to a venture problem: companies become orphaned when their original partner leaves, especially when junior partners move on and leave no champion willing to “go the last mile,” get on a plane, challenge the CEO, or assemble another round.
Rory’s test is access to allocation, not boardroom wisdom. A former partner may remain an exceptional director, but if that person cannot advocate where the firm’s money is decided, they are “to a rounding error… useless”; retain them as an independent and put a current capital advocate in the room.
Reserve governance produced real disagreement. Jason liked separating the decision from the sponsoring partner because partners repeatedly try to rescue weak companies; Rory wanted the partner’s recommendation combined with a robust group process, since a net-new decision-maker lacks the historical knowledge of how the company evolved.
3. Reserve accuracy improves after product-market fit, but trouble capital rarely makes home runs
Harry’s seed experience challenges the premise of reserves: after an 18-month deployment, none of the five companies he forecast as fund returners became one, while the eventual winners were absent from his list. Scarce seed reserves therefore chase the fastest current growers, not necessarily the most durable value creators.
Harry argued that the picture improves after product-market fit and early acceleration. If a company is on plan or within 25% of it two years after the investment, his estimated probability of a 5x rose from 30% to 70%, though he conceded he is “not sure it is 70 anymore.”
On pay-to-play rounds, Rory began with “it never works well” and then supplied the exception: an aggressive down round in FedEx near its network tipping point made investors “like bandits.” His statistical conclusion survived the anecdote—the probability of a home run in troubled situations is low, not zero.
4. The real cost of a bridge is partner time, not merely its mediocre return
Bridges can still convert a 0.5x into a 2.5x and prevent a damaging hole in a fund, but Rory would commit only finite money against a clear path to cash-flow positivity. “You tend not to lose money on the bridges. You just don’t make as much as you think.”
His non-negotiable was execution without theater: “Don’t make this too hard as well.” The opportunity cost that kills is partner time, so a founder who wants support must stop “dinking around,” execute the plan, and build realizable value.
Rory acknowledged that this is not entirely cold-blooded. After founders work six or seven years, the difference between collapse and a modest exit may be $10–30 million personally; Harry added that one company he rescued from fumes now exceeds $300 million, though its founder later did not remember the rescue occurring.
5. Fewer exits can return more cash when the winners stay private longer
Harry estimated roughly $3.5 billion returning to Index across Wiz and Scale, arguing that Index and Founders Fund can generate 3x DPI on funds of this scale. LPs exposed to Index, Kleiner, Sequoia, and similar winners are seeing “massive amounts of cash coming back.”
Rory’s formulation was “fewer, bigger winners.” There may be fewer IPOs, but companies have compounded privately for longer, so an early position that once returned $400–500 million can now return more than $2 billion.
The temptation is to extrapolate Figma’s possible $30 billion outcome into trillion-dollar venture outcomes a decade later. Rory rejected that line: public-market windows remain cyclical, and a step change in when companies list does not imply another identical step every ten years.
Larger funds do have somewhere to deploy capital because leading AI companies require extraordinary financing. But the ideal remains Figma: invest at a price premised on a $1 billion outcome and receive $30 billion, rather than bidding today as if the exceptional outcome were already guaranteed.
6. Melio exposed the clearing price for a very good, non-elite company
Rory called Xero’s $2.5 billion purchase “discouraging.” Xero’s materials put Melio at $153 million of revenue growing 127%; Jason later noted $187 million in March and reasoned it might have exceeded $200 million at signing. Jason also questioned whether the 127% growth rate was sustainable.
Rory saw accounts payable as crowded, go-to-market intensive, and strategically suited to consolidation with an ERP-adjacent buyer; roughly 13–14x forward revenue therefore looked like an industrially sensible outcome.
Melio’s last private marks were discussed as $4–4.5 billion, versus a $650 million preference stack. That is the “lovely rigged game” of late-stage investing: recover 1x on enough losers, accept a 0% IRR, and let the winners supply the fund’s upside.
Board incentives diverge sharply below the last valuation. Late investors may simply say, “Give me my money back, and move on,” while every incremental dollar accrues heavily to founders and early investors; public-company consideration also avoids the existential argument over what private buyer stock is notionally worth.
7. Secondary liquidity is rational, but asymmetric rights create resentment
The panel defended founders who sell shares while growth investors are force-feeding capital into a hot company. A founder moving from roughly 49% ownership to 42% may be making a rational decision, not abandoning the business—especially when the financing cannot be absorbed solely through primary issuance.
Rory’s requested bargain was simple: if a founder sells more than $10 million, let him sell proportionally too. Harry noted that co-sale rights are designed for this purpose; Jason wanted a “super co-sale” if more than $20 million were being sold. Larger holders can waive those rights, sometimes without smaller investors even being told.
Private-to-private acquisitions intensify the problem because both price and consideration are notional. Rory joked that repeated portfolio acquisitions left him with more Airtable stock than anyone at an $11 billion mark; with public shares, “you read The Wall Street Journal” and immediately know what the consideration is worth.
8. AI compensation may reveal motivation while destabilizing everyone left behind
Harry asked whether giving an AI researcher $100 million creates the same inefficiency as stuffing a startup with excess capital. Rory jokingly cited an almost one-to-one correlation between buying “the massive pad in Atherton” and declining revenue, while acknowledging that he could not prove causation.
Rory’s distinction: large amounts of money do not necessarily change people; they “reveal what you really want to do.” Some recipients double down, while others discover they wanted to travel and play golf—the payout liberates the underlying preference.
Zuck has previously made expensive arrangements work with Bret Taylor and Kevin Systrom, so individual demotivation may not be the central failure mode. The greater risk is internal disruption when adjacent engineers see radically different compensation and become “really pissed off.”
9. AI capex is strategically rational before it is demonstrably positive-NPV
Harry questioned whether $300–400 billion of annual AI capex, plus salaries, will look economically rational in two or three years. The discussion separated positive NPV from game theory: spending $60 billion to ensure an invitation to the AGI “party” may be strategically mandatory without producing an accountant’s return.
Overinvestment is eventually inevitable because “that’s just what humans do” when confronting a vast opportunity; Harry simply could not tell whether it happened last year, is happening now, or lies two years ahead. The spending has already transformed huge free-cash-flow machines into “cash incineration” businesses.
Menlo’s new $1.5 billion fund illustrated the fundraising bar: Chime supplies realized, “hard dollar” proof, while Anthropic supplies the forward AI narrative. LPs understand the outcome math, but past access—Uber, Chime, Anthropic—is still the strongest evidence that a manager might enter tomorrow’s defining deals.
10. Couchbase is a rifle-shot buyout, not evidence that PE will clear software inventory
Couchbase sold for $1.5 billion at roughly 5.7x its $215 million of revenue despite 12% growth and no profit. Rory viewed Haveli’s purchase as a concentrated thematic bet on Couchbase’s faster-growing new product and potential AI relevance, not a template for every slow-growing infrastructure company.
Harry’s field evidence was colder: two portfolio companies with strategic and PE characteristics recently received mediocre M&A offers, yet neither had received a single PE inquiry. Two years earlier, such outreach arrived weekly.
Roll-ups such as Visma and Constellation illustrate that twenty stranded $100 million businesses can become a $2 billion platform the market will value. But Constellation’s model is about 2x revenue, sellers now hope for 5–6x, and Harry’s proposed Grammarly-centered productivity roll-up still struck Jason as a thesis that works more reliably “over lunch and cocktails” than in transactions.
Jason’s eventual sorting rule was growth quality: some mid-teens companies have a credible route to reacceleration; others will never reaccelerate and should seek an exit. AI widens the gap by making many formerly serviceable legacy assets look “hopeless.”
11. Incumbents had time to become AI-relevant—and June 30 was the rhetorical deadline
Harry’s rule was categorical: “If you haven’t grown because of AI, you failed.” Shipping a copilot is insufficient; a company must have reaccelerated because competitors are accelerating, making flat relative market share equivalent to decline.
ICONIQ’s data across roughly 300–400 B2B companies, including prominent AI startups, showed aggregate growth almost unchanged from 12 months earlier. The panel’s reading was that vendors are reallocating a mostly fixed technology budget; the market expands materially only if AI converts labor spending into software spending.
Harry contrasted an AI-native portfolio company reaching more revenue in nine months than a Superhuman-like predecessor reached in eight years. Yet incumbents can adapt: Intercom did, and 25-year-old vLex made its legal libraries AI-relevant before Clio bought it for $1 billion.
Jason credited Howie at Airtable for making an “unbelievable shift in product strategy” quickly, while emphasizing that results remain TBD. Rory’s boardroom rule was harsher: if directors are still debating whether to do something in AI, “just stop going to the board meetings.”
12. Oracle, Scale, and the talent exodus show both sides of the winner-take-most market
Oracle’s reported $30 billion-a-year OpenAI deal became the incumbent benchmark: Larry Ellison redirected cash from buybacks into Nvidia GPUs, and that capex appears to be converting into cloud demand. The AI business barely registers inside Oracle’s revenue mix but dominates investment—an “upside-down AI business” inside a profitable legacy company.
The deal works financially if AI investment continues for another three or four years; until then, the market rewards Oracle for leaning in. Harry’s taunt to laggards: “If Oracle can get AI native by June 30th and your portfolio startup can’t… I’d give up.”
In the Scale transaction, its 49% owner recruited key employees, and rival cloud providers were unlikely to share confidential data. One panelist called the remaining company an “empty husk,” while Harry argued that roughly $1 billion of revenue had effectively moved to Turing, Surge, Mercor, and others. Surge reportedly built roughly $1 billion of revenue before seeking $1 billion at a $15 billion valuation.
That concentration reaches people and venture pricing. The panel cited 2,221 CEO departures, up 24%, while LaunchDarkly’s CEO moved to Asana and 4–6-year founder commitments stretched to 12–13 years; Rory described losing a Series 8 at $5 million of ARR that moved from roughly $225 million to $600 million because missing the sole defining winner now feels costlier than overpaying.
In quickfire, Rory and Jason both rejected an Elon Musk political party this year—Jason principally because he did not care. All three ultimately backed Cluely’s founder becoming a paper billionaire before 2029, conditional on a claimed $5 million run rate not cratering, 200–300x AI multiples persisting, and ownership remaining around 25–50%.