Figma's 250% Pop - The Greatest IPO Mispricing Ever? Meta & Microsoft Blowout Quarters: Broken Down

Summary

• The “Figma left $3bn on the table” crowd is “talking out of their ass,” per Rory O’Driscoll, while Harry says he has sat in five or six pricing rooms: the order book had bids at 38, 39, 40 — nobody at 98 — and “the 98 price only happened because the IPO happened at 38.” Bankers engineer a designed 15-20% pop; occasionally you get an inadvertent 250% one — the largest since 1999 — and mega pops may just be “a natural intermittent consequence of the IPO process,” like earthquakes. Direct listings wouldn’t have fixed it.
• The real cost of the pop is structural: institutional buyers set internal exit price targets before buying, and “none of those price targets are going to be more than 110 bucks a share” — so Figma now trades above the targets of the very long-onlies it courted, and Rory believes “most of them are selling those shares right now.” Brian Halligan pushes back: the long-onlies took a toehold and held; “I think it’s hedge funds that sold.”
• The IPO window is wide open and the cheap money has flipped to the public markets. Rory’s framing: for three years private capital was cheaper and less hassle; now public multiples exceed private ones, so “if you need to raise money in the next two years, now would be a really good freaking time.” Lemkin to Canva: “Run Forest Run.” A panel line, endorsed by Halligan: “VCs are a much bigger pain in the ass than public investors.”
• CEO comp is broken twice over, says Halligan: after a 2006 regulation change, the industry moved from ISOs to RSUs, which “creates risk-averse behavior in the CEO” versus swing-for-the-fences ISOs, and peer-percentile benchmarking pays a Dylan Field ~$20M — “3% of his net worth… it doesn’t move the needle.” He likes Figma’s ~$2bn PSU moonshot; Rory notes the catch — the price triggers (up to ~$118/share) were achieved by the pop itself, so “the performance element vanished very quickly.” Moonshot grants are now standard in many growth-round term sheets.
• On AI capex, Rory’s earnings-week takeaway: this isn’t AI working — it’s that the hyperscalers’ existing businesses throw off so much cash they can fund the buildout for as long as those businesses keep generating cash. “Real men with $70 billion of free cash flow get to spend $40 billion of that on servers.” The math: $400-600bn of annual capex enabling maybe $25-30bn of apps revenue — long-term real, but “the marginal player will get caught just like in 2001.” Brian’s bubble tell: tech selling to tech; he prefers “mere mortals” revenue — ChatGPT, Harvey, Rogo.
• SMB AI is the uncracked opportunity: prosumer works (Lovable, Gamma), enterprise works, “I haven’t seen much in between.” The blocker is training — “at Brian and Jason’s sandwich shop, there’s no AI team.” Jason Lemkin: any founder who solves self-training and collapses a “Palantir-grade deployment” from six months to 60 seconds, “I want to invest this hour, this second.”
• Cognition/Windsurf at a rumored $15bn (up from a rumored 10, on ~$170M combined revenue) bought a brand and market entry, not a team — 30% laid off, the rest offered 80-hour weeks or a 9-month package. Lemkin’s lesson from the saga: once you leave the standard cap table, “you’re basically depending on the kindness of strangers… which is always a mistake.” Meanwhile Ramp’s $500M at $22bn is partly just fintech physics — “2% dilution or get a banking license” — and a high-priced round is only a “suicide round” if you have to raise again.

Deep dive

1. The IPO price is set the night before by exhausted founders across the table from their own bankers

• Halligan’s inside account: price and allocation are decided the night before, after two weeks, 12 countries, six pitches a day — “your battery is on red.” You and Morgan Stanley are “perfectly well aligned” for the whole roadshow “until this one 1-hour meeting,” when suddenly “we’re across the table from you”: the bankers’ book arrives stuffed with “their hedge fund buddies with pretty good allocations,” while the company wants only the long-onlies — Fidelity, Wellington, Capital, T. Rowe.
• Then comes the speech Harry says he’s heard in five or six pricing rooms, verbatim every time: Morgan Stanley pushed HubSpot to price at 24 because “Fidelity has told us they’re in at 24, they’re out at 25.” HubSpot played chicken, priced at 25, and Fidelity came in anyway. Brian’s careful phrasing: “I don’t think there was collusion, but definitely Morgan Stanley and Fidelity were on the same side of the table trying to get us to sell at a lower price.”
• On whose fault the pop is: “The founders make the decision on what the price is. So Dylan decided.” And Figma’s setup was combustible — mostly secondary, very few shares, 40x oversubscribed (HubSpot was 27x), with the six big long-onlies “pissed because they’re not getting a big allocation.” Halligan’s defense of the team: “There’s no one dumb at Figma… no one massively mistakenly underpriced this deal.”
• Harry’s mental model of the closing dinner is worth the price of admission: “The bankers have just screwed you over for a buck a share and in return they buy you a very expensive dinner and they liquor you up so you forget… one day later you drive out of town and the next lamb is led into the slaughter.” Why it never gets reformed: “You’re doing this once and it’s the most important thing in your life, and these guys are doing it every day.”

2. “The 98 price only happened because the IPO happened at 38” — the $3bn critique doesn’t survive the order book

• Harry separates two problems: the designed “small explosion” (the classic 15-20% pop that makes Fidelity feel good) versus the inadvertent big one — Figma’s ~250% pop, which he calls the largest since 1999. The first is a $1-per-share debate; the second is a different phenomenon entirely.
• Rory’s demolition of the Gurley-style critique: the book had orders at 38, 39, 40 — “no one was putting in an order at 98.” So “had someone walked in and said, I know this IPO is going to price at 100 bucks to open tomorrow, let’s raise at 80 — they wouldn’t have had a book.” The $3bn “wasn’t accessible”; the people claiming otherwise are “talking out of their ass.” Maybe Figma left “a couple extra bucks” — not sixty.
• The genuinely sad consequence, per Rory: every big institution buys with an internal price target for exit, and “none of those price targets are going to be more than 110 bucks a share” — so the stock now trades above the long-term targets of the long-term holders you wanted, and “most of them are selling those shares right now.” Halligan disagrees on the facts: the long-onlies built a toehold to “hold for the long haul… I think it’s hedge funds that sold.”
• Two under-discussed constituencies: a panelist ran the directed-shares program at his first startup’s IPO — employees scraped together $50-70k to buy in and “everyone in the company basically made 100 grand that day… employees don’t even know what dilution is.” And on founder dilution, Halligan’s four words — “My net worth went from X to 100x. I was just happy to be there” — which Rory calls “why this process is so hard to change.”

3. Cap table quality compounds for a decade — the Zendesk cautionary tale

• Before the IPO, HubSpot did a non-deal roadshow: every big investor they met in person came in big; the two they skipped — T. Rowe in Baltimore, Capital Group in Southern California — took three to four years to win over post-IPO. Rory’s rule: “You always remember the people you didn’t get.”
• The counterfactual: Zendesk priced three months earlier in a shaky week, “never really got” the big long-onlies, and carried a hedge-fund-heavy register for years. When trouble came, the weaker base — at the margin — fed the activist pressure and the reluctant sale (turned down ~$20bn, sold for ~$10bn). Halligan’s honest coda: “Having a good cap table is underrated” — and HubSpot’s was “more luck than skill.”
• Would a direct listing have saved Figma? Rory did the homework — the SEC now allows raising capital in one (Amplitude did it) — but a direct listing would have cleared at 39 or 40, not 100: “mega pops are a natural intermittent consequence of the IPO process,” like earthquakes. The venture-land rhyme he’s watching now: a brand name prices a round at 120, and six weeks later “there’s a bunch of people doing it at 350. Why? Well, the brand name is in.” This is a public-market equivalent of what he calls “formal.”
• One retail myth punctured along the way: HubSpot is roughly 90% owned by big institutions — “very little… owned by mom and pop, owned by Harry Stebbings’ mom.”

4. Run Forest Run: the cheap money has moved to the public markets

• Asked what Canva should do watching this, Lemkin doesn’t hedge: “Run Forest Run. The market’s wide open. The valuations are good… I’d be lining everything up to go public.” Lemkin complicates it: the founders have pledged away most of their stock, the company’s profitable, early investors have traded at tens of billions — “this clearly isn’t a Musk empire that’s being built.”
• Rory zooms out to the mechanism: “price clears all markets.” For three years private money was cheaper and easier — “5 times revenues with a bunch of people in New York busting my balls, versus 10 times revenue and I never get to talk to these growth-stage guys except once a year.” That’s now inverted: “the cheap money is now in the public markets… if you need to raise money in the next two years, now would be a really good freaking time.”
• Against the Stripe-led “why ever go public” drumbeat, Halligan’s lived comparison: private, “we had a bunch of quirky, slightly misaligned venture capital investors who were definitely in our shorts”; public, the same species but “less in our shorts.” Public investors are “underrated” and rational if you paint a long-term picture conservatively; the Twilio/Zendesk activist horror stories are “pretty rare” — and those companies “were having some issues.” The line endorsed by Halligan for the B-roll: “VCs are a much bigger pain in the ass than public investors.”
• And the human case: IPO day is “one of the top two or three days of your life. You will cry. You will laugh. You will hug.” Halligan’s memory — a co-founder showing him the stock app at a billion-dollar cap: “Take a screenshot. We’ll never see that again.” (Practical footnotes: the first trade takes hours to settle — Figma took ~6 — and pick NYSE over NASDAQ “because you get to ring the bell.”)

5. CEO comp is broken — peg it to net worth, not to peers

• Halligan’s two-part indictment: after a 2006 regulation change, the industry moved from ISOs to RSUs, which “creates risk-averse behavior in the CEO” — options made you “swing for the fences.” And comp committees pegging to the 75th percentile of peers would pay Dylan Field ~$20M a year — “that’s like 3% of his own personal net worth. It doesn’t move the needle.” Same logic as Musk: “Mary Barra makes $29 million a year. Do you think Elon cares about $29 million?” So he likes Figma’s PSU-heavy ~$2bn moonshot.
• Harry’s dissection: PSUs have “bizarrely recreated options” that were regulated out of existence in 2006 — but Figma’s package “didn’t work,” because the stock-price triggers (running up to ~$118/share) were achieved by the pop itself. “The performance element vanished very quickly” — though Dylan still vests over seven years. He’d prefer tangible multi-year targets (revenue, op income), but disclosure plus “the ISS and all the whiny babies” push everyone back to stock-price triggers — which paid out for every 2018 package and left the 2021 vintage (Airbnb’s included) stranded.
• HubSpot’s answer — net new ARR plus a floor earnings number — drew genuine admiration from Rory: “very few committees do what Brian does… so much better than just stock price.” Halligan’s framing of all comp design: “none of this is perfect — it’s what’s the least bad you can do.”
• The trend line: Lemkin sees moonshot packages in “every growth round” in his portfolio — do the deal at $1bn, but hit 10x and the founders share another 7-10% of the company. Rory predicts CEOs who nail the business but miss inflated price triggers will demand waivers (“I can just see the movie now”); Lemkin bets “nickels to dollars” grouchy VCs won’t budge — “it’s not 2024 anymore.” Related: Halligan endorses founder secondaries — selling ~$2M to Sequoia was “a horrible financial decision” in hindsight but “it stiffens my backbone” when Salesforce came knocking — and now sells the identical share count monthly so the tape reads no signal.

6. Meta’s quarter and the capex question: the cash machines are funding the frontier

• Harry’s numbers: 38% YoY adjusted EPS growth, 22% revenue growth — and a 22% drop in free cash flow. Rory’s big aha from earnings week (having “got AWS broadly right and Microsoft broadly wrong”): this is not AI working for the hyperscalers — “all their existing businesses are working so well and kicking off so much cash that they can keep doing this for the next year.” Or as he put it: “Real men with $70 billion of free cash flow get to spend $40 billion of that on servers. It’s a great country.”
• His bubble arithmetic: total AI apps revenue is maybe $25-30bn against $400-600bn of annual capex. Fast-forward ten years and apps could be $300-400bn — so the long-term trend is “almost certainly real” — but expect a stretch where “the marginal player will get caught just like the marginal player got caught in 2001,” the overlevered get burnt, and the big guys retrench then grow into it. His refusal of both poles: “It’s not amazing AI maximalism and it’s not bubble doomerism… until you discover the frontier, you’re not investing enough. Unless you try, you just end up like Europe.”
• One panelist’s tell for froth: “I get nervous about tech companies selling to tech companies” — that’s what 1999-2000 was, and Harry adds 2021-22. He wants “mere mortals” revenue: ChatGPT, Harvey selling to lawyers, Rogo to investment bankers. Lemkin won’t let the panel off: “it has to be a bubble at some level. The capex can’t last forever… hopefully we all get out.”
• Halligan’s overlooked-winners stat: from the market bottom on August 26, 2022, Oracle and SAP stock are both up ~230% — only Shopify (~300%) has done better in SaaS — outpacing HubSpot, Salesforce, Adobe, and Atlassian. Rory on Oracle specifically: they took the cash flow, bought GPUs, “and have now made themselves relevant in cloud.”

7. CEO of the year is Jensen — but the Satya case is the interesting one

• Rory names Jensen without hesitation, and a panelist adds the reason: “he’s rethinking the CEO playbook.” A panelist’s structural version: the only two categories in the stack that didn’t exist at scale before — GPUs and models — belong to Jensen and to Sam/Dario, and both now “appear to be dominant relative to the other parts of the category.” Satya and Zuck are contenders for a different prize: “managing the cash machine brilliantly at scale — which turns out to be a pretty lucrative way to spend your adult life.”
• A panelist’s Satya-over-Zuck argument comes from his Adobe VP years: for a founder “this stuff’s easy — you just call the troops together. Zuck can do what he wants.” Adobe took three years of internal convincing just to move to the cloud; Satya invited Sam Altman in, did “this kooky deal to buy 49% of OpenAI,” and went all-in on Azure for AI without founder authority. Another panelist’s label: “He’s basically a refounder.”
• A panelist’s darker, funnier read: Satya’s genius was accepting “this large bureaucratic company can’t get it done” and cutting the convoluted deal rather than banging his head against the wall — the sound in his head being “thanks a fucking lot the rest of you guys, I had to figure this out with one BD guy while all you guys were sitting on your ass not shipping AI.” Yes, Microsoft eats 49% of OpenAI’s losses — but that’s ~3% of operating income “in return for probably a trillion dollars.”

8. SMB was counter-consensus for HubSpot; SMB AI is still uncracked

• Why Halligan bet on SMB in the first place: he’d spent his career on “the soul-crushing exercise of selling to CIOs,” believed the internet disproportionately favored the small — “your success was much more about the width of your brain than the width of your wallet” — and judged the business on CAC and LTV, not the P&L. It was deeply anti-consensus: Marketo, “more enterprise,” was the consensus bet raising at bigger valuations; the SMB winners (Shopify — “even better than HubSpot” — Block, Monday) all came from “outside of consensus land in Silicon Valley.”
• The origin story doubles as a cycle lesson: in the 2009 recession HubSpot took 20 meetings up and down Sand Hill and “every household name said no,” until Scale wrote a term sheet at 66 pre on a company doing $7-10M doubling. Rory: “The time to buy is when everyone else is not buying” — that ‘09 window produced HubSpot, Box, DocuSign, and RingCentral.
• Would the SMB play work in AI today? Lemkin’s quandary: real AI products need training and forward-deployed engineers, and SMBs have neither — HubSpot’s own survey claiming 80% of “SMBs” have AI teams conflicts with his view that “at Brian and Jason’s sandwich shop, there’s no AI team.” His open checkbook: any founder who cracks self-training — collapsing “a Palantir-grade deployment” from six months to 60 seconds — “I want to invest this hour, this second.” Rory sees a barbell: prosumer works (Lovable, Gamma, Replit), enterprise works, “I haven’t seen much in between.”
• Rory’s resolution — it “may well take a year or two longer”: big companies with money to burn define the app category first, then SMB gets it pre-trained and pre-baked — call answering, order dispatch — “just turn it on and you two can sound like a big call center.” SMBs want everything the big companies have, “packaged tightly and priced tightly… in bite-sized chunks.” Halligan’s caveat on all pattern-matching from HubSpot: much of what worked then — inbound, freemium, PLG — “just wouldn’t work today. You got to keep innovating.”

9. Cognition/Windsurf: they bought a brand, not a team — and everyone’s a gray hat

• The rumored round moved from $10bn to $15bn on ~$170M of combined revenue ($85M each side). Lemkin shrugs at the jump — it rhymes with the Anthropic rumor going from 100 to 170: “demand is high for premium assets… price is how scarce assets get allocated.” It’s the private-market version of the IPO pop.
• The post-deal mechanics say what the deal was: Cognition laid off 30% of the acquired Windsurf staff and offered the remaining ~200 a choice — 80 hours a week, six days in office, or a 9-month package, decide by August 10. Lemkin’s read: “it’s clear they weren’t buying the team.” Devin is a respected niche product — his hardest-problem portfolio CEOs use it, “but they’re not deploying it across their whole team like Claude Code” — so Cognition bought a brand, ~$80M of revenue to maintain, and 3-9 months of accelerated market entry, roughly non-dilutive at the new price.
• The saga’s moral, after it emerged the founders and investors — not Google — put up the $100M left in Windsurf after the handshake: “there’s not quite as many white hats and everyone’s a gray hat, it turns out” (Lemkin). Lemkin’s generalization: once you leave the standard cap table for improvised deal structures, “you’re basically depending on the kindness of strangers, as Blanche DuBois would say, which is always a mistake” — in a normal Delaware M&A, at least “you get what you get.”

10. Ramp’s $22bn, the myth of the suicide round, and CRV’s retreat to what it does well

• Ramp’s $500M Series E at $22bn (Iconiq leading, roughly the fifth or sixth round in 18 months) is, per Lemkin, both momentum and genuine need: issuing corporate cards means funding the float. One panelist’s back-of-envelope: ~2-2.5% interchange on the volume behind ~$700M revenue implies a $3-4bn capital requirement — “you’re replacing Amex.” Hence the punchline: “venture equity is the lowest-cost capital out there. 2% dilution or get a banking license. I’ll do the 2% dilution.”
• On Harry’s “suicide round” challenge (e.g., $100M at $3.1bn), the rule is: a too-high price “is only quote suicide if you have to raise again.” Raise $500M at 2 and even 3 or 4 years later go public at 1.5 — “tough shit for the guys who paid two, but life goes on.” The fatal version is raising 100 when you needed 400 and returning into a down round. Lemkin adds these 1-2% slivers barely register anyway — his own anchor LP once told him a small $3bn markup “doesn’t count. Don’t recognize it.”
• CRV raising $750M, shrinking the team, and dropping its late-stage select fund drew praise, not concern. One panelist: “do the thing you do well… keep the message clear” — and LP appetite isn’t monolithic: the same LP can applaud CRV’s focus, hand Founders Fund another billion, and give Elad Gil $1.5bn solo. Lemkin’s carry math from his own opportunity fund: maybe 10-15% more carry for a lot of drama — “90% of my carry will come from the main fund” — and two $750M funds beat one $1.5bn because “you want to get into carry mode faster.”
• Harry’s needle on Benchmark — did fund-size discipline cost them Miles and Victor? One panelist refuses the personal version and reframes the meta-question: can the best specialist fund compete with full-stack firms? Benchmark is “among if not the best” specialists; both strategies work if executed, but each carries its risk — specialists get “crowded out by the noise,” full-stack players overextend into “lots of good individual deals” that don’t add up to compelling returns. One panelist’s blunter market check: in the X conversation the brand list is “YC, Sequoia, Andreessen — that’s it really.”