With Harry Stebbings, Jason Lemkin, and Rory O’Driscoll
Anthropic is closing $30 billion at a $900 billion valuation. That’s nearly triple the $380 billion price from February. Andre Karpathy joined Anthropic the same week. Cerebras priced its IPO at $185, popped 68% on day one, and broke $300. SpaceX set June 12th as the date for the largest IPO in history at a suspected $1.75 trillion market cap.
Meanwhile, Marc Benioff casually mentioned on All-In that Salesforce will spend $300 million on Anthropic tokens this year, almost entirely on coding. That works out to roughly $15,000 to $20,000 per engineer per year.
And on the other end of the AI boom: Cisco cuts 4,000 jobs, LinkedIn cuts 875, Meta cuts 8,000, Intuit 16,000. Eric Schmidt got booed at a graduation speech. Three years ago, kids were applauding ChatGPT at the same kind of ceremonies. The political direction of travel is exactly one way.
This is the most consequential week in AI venture in months. Here’s what it means.
Top Takeaways
1. Anthropic at $900B Is the Best Trade in Venture Right Now
If you take Anthropic’s June run rate, $900 billion is roughly 18x revenue. Compare that to hot seed and Series A rounds in AI for $10 million ARR companies trading at 30-40-100x ARR five years away from an IPO. Anthropic is so far post-IPO scale that there’s no IPO risk, no “it will go away” risk. The only risk left is valuation risk.
“Every time the multiple on this has been significantly lower than the multiple on your median Series A or B or Series C, for a higher growth rate,” Rory said. “It’s been the best trade out there.”
The flip side: why does Dario Amodei do this if it’s such an obvious deal? Because giving away 3% of the business to derisk another year of monstrous burn is rational when you’re committing to 5-6 gigawatts of compute this year at a rule-of-thumb cost of $40-50 billion per gigawatt. You’re not spending that capital directly. You’re persuading hyperscalers to spend on your behalf. But the balance sheet war is real.
There’s also a philosophical difference between Anthropic and OpenAI on fundraising. Sam Altman pushes valuations to the absolute maximum because his thesis requires infinite capital. Dario does the opposite. The last two Anthropic rounds appear to have intentionally traded at a discount to get done low-drama in days. The Anthropic philosophy: “If I want to raise $30 billion, I should probably get a check for $30 billion and call it a day.”
Compare that to OpenAI’s last round: Amazon gives $50 billion, but $20 billion is up front and $30 billion is contingent on going public. SoftBank gives $40 billion, but $30 billion has to be borrowed. “It’s like, give me a break.”
2. The Salesforce $300M Token Bill Is Both Nothing and Everything
Mark Benioff dropped on All-In that Salesforce will spend $300 million on Anthropic tokens in 2026, almost entirely on coding. Sounds enormous. Do the math.
Salesforce has roughly 20,000 developers out of 83,000 total heads. That’s $15,000 per developer per year, or $1,200 per developer per month. Salesforce’s total engineering wage bill is around $5.8 billion. So $300 million is roughly 4% additional spend on top of fully-loaded engineering cost.
Rory and Scale ran a survey across 40 portfolio companies and external companies. The average came in at $1.2K-$1.3K per developer per month. The median was lower. Salesforce is in the strike zone of normal. It’s only big because they have so many developers.
But this is the only product Salesforce buys that comes anywhere close to this amount. If Salesforce bought a brand new ERP system from SAP, they’d maybe spend $5-10 million. Anthropic tokens are the largest single external software spend at most enterprise software companies, period. From nothing two years ago.
The real question is the trajectory. To justify Anthropic’s $900B valuation and OpenAI’s path to a trillion dollars of token revenue, the math requires something like 5-7% of every knowledge worker salary and 20% of every engineering salary getting converted to tokens. Salesforce at 4% is maybe a quarter of the way there.
One of two things has to happen: either they stay at $300 million (in which case the TAM for token businesses is overestimated and there’s a real correction coming), or they 4x their token spend from here. In two years Benioff is on All-In saying “we spend a billion dollars on tokens.” Pick a side.
3. The Klaviyo Counterpoint: Most Companies Massively Overestimate Token Costs
An underrated moment from SaaStr AI 2026 last week was Andrew Bialecki’s session. Klaviyo requires every employee anywhere close to product to commit code. Every single person has to be running AI or agents to do their job. 100% adoption. They built their own custom agentic framework to enforce it.
Backstage, Jason asked Andrew how much he thought it costs to run an AI VP of marketing. Andrew guessed $250 for both. The actual answer was $257 combined. He was the only person who got that number right, because he’s actually doing it.
His point: “A lot of this stuff is not as expensive as we think. We do not need to worry about token maxing at Klaviyo, and everything we’re doing is agentic. The most junior product person, every single person has to be doing this. But if you do this right, it’s not as expensive as people think.”
The implication: there’s a bear case where the models get better and more efficient, the orchestration gets smarter, and most companies need a half or a third or a quarter as many tokens as they think they do. The Mark Benioff number is roughly right as the bear case for companies not at the bleeding edge of token-maxing. But it’s not the floor that some people assume.
4. Public B2B Has Hit Bottom and Is Quietly Reaccelerating
The bounce off the SaaS apocalypse lows is real but uneven. Figma reaccelerating to almost 50% growth. Atlassian reaccelerating north of 30% on the back of Rovo. Datadog hitting its first billion-dollar revenue quarter at 32% growth. Twilio coming back from the dead to 20% growth.
The meta question: is there more time than we thought?
“The whole world is not in San Francisco of all the buyers, of all the users,” Rory said. “Maybe Andrew at Klaviyo and Mark Benioff and these founder-led companies have enough time. Maybe another year, if they’re just getting going on their agentic journey.”
But the bigger insight: every tech in cycle has a set of industries that, once in their life, get valued on prospects and futures. Five years ago, that was B2B SaaS. Today, it’s AI. Once you lose that veneer, you’re valued for the rest of your life on revenue, growth, and cash flow. You almost never get back to 50x ARR.
Figma reaccelerating, Datadog crushing, Atlassian inflecting. These are wins. They’re just bounded wins. Datadog at 17-18x sales is the outer edge of good. Figma at 6-10x sales is the standard for good. Wix at 1x is the floor.
5. The Figma Vibe Coding Miss Cost Them $500M+, But the Captain Obvious Point Was Bigger
The honest founder take: Figma Make is the worst vibe coding product anyone on this pod has used. It was clearly not important to senior management. They left $500 million or more on the table by not building a Replit-level competitor.
But the captain obvious point that got missed: Figma is in the business of making it easier to build software. The software explosion is happening whether or not Figma owns vibe coding. Even though they lost the vibe coding race, they benefit from the software explosion the same way Work OS and RevenueCat have exploded because there’s a software explosion going on.
What Figma still hasn’t built: the workflow where you create a design, loop in the product team to approve it, and then click one button to push it into a full production prototype. Replit and Lovable both have “upload a Figma design” as a native insertion point in their prompt UI, because they know that’s exactly what their ICP wants to do. Why Figma doesn’t have that native flow is a loss of $500 million and going up.
6. Wix at 1x Revenue Is Terminal, Not a Buying Opportunity
Wix is down 45% since the stock repurchase. It’s now a $2.2 billion market cap, trading at roughly 1x revenue. Base44 (which Wix acquired) just announced hitting $150 million in ARR. Impressive substitution revenue. The core pre-AI business is no longer growing.
The structural problem is two-vector. Vector one: anyone who isn’t tech-phobic can vibe code their own site in 10 minutes with templates and integrations already built in. Squarespace and Wix are great products for genuinely tech-fearful customers, but the addressable market just shrunk.
Vector two (the one people forget): Wix and Squarespace spent the last 5 years competing as low-end Shopify alternatives. The growth was from merchant services, payments, e-commerce. Shopify went up-market AND down-market simultaneously and destroyed every sub-Shopify in the process. WooCommerce, BigCommerce, Wix, Squarespace, all hit on the same wave.
The buyback critique was sharp. Buying back stock to “offset stock-based comp dilution” is one of the worst justifications in public company management. The only reason to buy back stock is because you think it’s way cheaper than it should be. If the stock was trading at a trillion dollars, would you buy it back at a trillion to offset Slack dilution? It’s bogus reasoning that bankers love. When the business is in trouble, things that are bad can get worse, and the ability to have another billion dollars on the balance sheet for an acquisition is worth more than juicing the stock short-term.
The real reason most distressed public companies do buybacks: to placate activist shareholders before they show up. It works until it doesn’t.
7. The Compute-Starved Universe: Everything But Traditional Software Is on Fire
Drive through the South Bay and you see it. Marvell Semiconductor, SanDisk, all the superstars of the ’90s. They’re on fire. Cisco’s back. Zscaler. Coreweave. Nebius up 684%. The only thing that isn’t inflating right now is old-school B2B software.
The setup for Nebius (and the entire compute layer) is binary. If compute stays starved, these are great businesses. If compute becomes plentiful, they’re commodity businesses and the over-extended ones go bust. End of analysis.
Gavin Baker made a useful point: the slowness of permitting and the inability to bring on data centers at the pace people want might save the industry from itself. The disaster scenario is if every data center people want to build actually gets built, AND the token revenue projections turn out half a trillion light. Then you’ve got a trillion in capex against half a trillion in revenue. But if the bureaucratic inertia of the American state means you only get half the data centers built, you’re saved by inefficiency. Compute remains scarce. People holding compute do well.
The bigger question Rory framed: does AI token spend grow faster or slower than data center capacity? With OpenAI and Anthropic sitting in the middle, collecting money from corporate America and handing it to NVIDIA, that’s the bet on the table.
8. Cerebras IPO Smashes, And SpaceX Is Next
Cerebras priced at $185, popped 68% to break $300. Original range was $110-120. They walked it up to $150, then $185, then it ripped. The biggest U.S. tech IPO since Snowflake.
The lesson: this isn’t a window opening for everyone. It’s a window opening for anything Figma-or-better. Cerebras has a $24 billion backlog. They’re an N-of-one bet on an extraordinary technology brought to fruition exactly when demand exploded, and they lined up the marquee customer (OpenAI) that was missing when they pulled the IPO two years ago.
It’s a positive tell for SpaceX. But it’s not a tell for sub-Figma companies. The market is bifurcated by quality, and Cerebras passed the bar.
SpaceX files June 12 at a $1.75 trillion valuation, raising $75 billion. The most fascinating part of the S-1: it’ll tell us everything about SpaceX as it existed December 31, 2025, before X.AI closed (February), before the Anthropic deal, before the Cursor acquisition. The S-1 will describe a company with maybe $15-18 billion in revenue, 25-30% growth, slightly profitable, with capex still TBD. Then the road show has to explain everything that’s happened since. No forward projections allowed in the S-1.
The retail dynamic will be wild. 30% retail allocation is unusual. The Robinhood and GameStop crowd has never had this kind of marquee Elon vehicle to swing at. Jason’s bet: this gets to $5 trillion market cap. Harry’s bet: $3 trillion. Rory’s bet: it’ll be lower than the IPO price within a month. (Rory does not bet on casinos.)
The Facebook IPO parallel is worth flagging. Facebook was profitable, far more attractively priced than SpaceX at IPO, and the defining company of its generation. It pushed price too far and traded down 50% over the next six months. When everyone thinks something is guaranteed, that’s when it blows up.
9. The YC + OpenAI $2M Token Deal Will Inflate Round Sizes
Sam Altman just offered $2 million in OpenAI tokens to every YC startup in the current batch in exchange for equity. Yuri Milner did the same with early YC batches with cash. Smart move by Sam, who needs to claw back mindshare and respect from Anthropic.
What it means for YC valuations: probably modest inflation, but the bigger effect is round size compression. If you’ve already raised $2 million in tokens, you don’t need to raise as much cash, which means even less ownership for traditional VCs. The average YC round ownership has already been sliced to 5-6%. This could slice it to 2-3%.
The use case matters. If you’re building a Lovable or Replit, $2 million in tokens is mostly engineering spend that can now go to product. If you’re serving customers, $2 million in tokens is marketing spend (free trials, generous limits, premium freemium tiers). For the first 12 months of a typical YC company’s life, tokens are marketing. Sam just gave 200+ startups a marketing budget the size of a Series A.
The reverse signal: if OpenAI has spare tokens to give away, they have surplus compute. Anthropic, which is rumored to be capacity-constrained, is not doing this deal. That tells you something about which lab thinks demand is currently saturating supply.
10. The Tech Lash Is Coming, and the AI Industry Is Walking Into It Eyes Open
Cisco cuts 4,000. LinkedIn cuts 875 (and specifically said it’s NOT caused by AI, which is more honesty than most). Meta cuts 8,000 to fund capex. Intuit cuts 16,000. Standard Chartered just announced 7,800 “job reductions in favor of the machines.”
Eric Schmidt got booed at a graduation speech. Three years ago, the same audience would have applauded any reference to ChatGPT because every kid in the audience had used it to write their final essay. The shift in three years is the entire story.
“As it’s becoming painfully clear now, no one in America other than us here in California likes the AI trend,” Rory said. “We have people who are brilliant scientists who politically are utter morons. And the people who are utter morons at AI but brilliant at politics are going to have us for lunch.”
The political problem: the AI industry spent three years telling everyone the technology would destroy humanity and put everyone out of jobs. Now those leaders are shocked that people don’t like them. Meanwhile, electricity bills are going up because of data center demand.
Jason’s prediction: the industry is going to need a social reflation. First we get fit, replace workflows, deploy agents. Then we have a social obligation. The Eric Schmidts can’t just keep talking about putting everyone out of work. There will need to be a deliberate hiring push, almost like a 2021 social charter, where AI companies double headcount even when they don’t strictly need to, just to avoid social unrest.
The harder version of this question: are these layoffs really because AI is replacing the work, or are companies just calling them AI-driven cover for cuts that were always coming? If the answer is the former, you need radical social policy. If the answer is the latter, you need transparency before the politicians get hold of it. Either way, “we did it, we put you out of a job” is not a winning posture.
11. The Founder-Led Comeback Nobody Sees Yet
One of the most under-discussed predictions: within the next year, one of the truly down-beaten public B2B software companies will become un-down-beaten. Not because they out-innovated startups. Because they’ve got the installed base.
The play is this: some founder-led B2B leader puts his best 50 engineers in a room and says, “Listen, I don’t even need you to innovate. I just need you to build a better version of [the best prosumer AI app]. Ship it. We’ll sell the hell out of it to our base.” HubSpot has 300,000 customers waiting to buy an AI SDR from Breeze if Breeze is actually as good as the startup alternatives. Today it isn’t. Tomorrow it might be.
This will not happen at non-founder-led companies. It requires the founder-CEO authority to override 15 layers of product committee.
The candidates are obvious. Drew at Dropbox. The HubSpot crew. Maybe Klaviyo. Maybe someone surprising like Mike at Canva with 2.0. Someone is going to pull this off, and when they do, it’ll feel sudden. From 0% growth to material reacceleration in two quarters. Hard to predict who, but it’s almost mathematically inevitable that at least one of them does.
12. The Elon vs. OpenAI Lawsuit Got Dismissed on the Statute of Limitations, But the Real Story Is Different
The jury came back fast. Two and a half hours. That’s a tell. They picked the foreman, decided the statute of limitations question, took a vote, went home. They could have black-boxed it the other way (clock starts at conversion to for-profit, not at original concern) and didn’t.
The real lesson is about Sam Altman’s complicated arrangements. The “I have no equity in OpenAI” framing has now created a problem he could have avoided. If he’d just taken 5-10% ownership when OpenAI converted to for-profit, like any normal CEO of a $300 billion private company, nobody would blink. Instead, he set up adjacent vehicles (the OpenAI Startup Fund where he kept the carry, equity in companies that sell to OpenAI), and now those structures are about to get litigated in congressional testimony.
The empathy take: he was trying to do the right thing structurally and got punished for the optics. The hard take: complex arrangements always bite you in the ass when you make an enemy of the richest man on the planet who is willing to spend $40 million in legal fees just to yank your chain.
Elon will appeal. The appeal will fail. But he’ll spawn dozens of related investigations and the cost to OpenAI in time, distraction, and management attention is enormous. That was the entire point.
What This Means for B2B Founders
A few takeaways worth internalizing:
- Token spend is now the largest external software line item at most tech companies. It’s bigger than rent at some. Plan accordingly. The number isn’t as terrifying as the headlines ($15K-20K per developer per year, roughly 4% on top of fully-loaded engineering cost), but it’s only going up.
- The bear case on token costs is real if you build the harness right. Klaviyo running two AI VPs for $257/month is the proof point. Most companies massively over-spend because they haven’t built the orchestration layer. Founders who solve the agent harness problem will have a 5-10x cost advantage.
- Public market valuation gravity is permanent. Even great quarters like Datadog’s get rewarded with normal-good multiples (17-18x sales), not bubble multiples (50x+). If your business isn’t AI-adjacent in an obvious way, accept that you’re playing a different game now and run it for cash.
- The agentic comeback for the installed-base incumbents is coming. If you’re a startup competing against HubSpot, Salesforce, Atlassian, or Klaviyo, the window where you can dance around them with better AI is shorter than you think. The 18 months they’ve had to ship is mostly wasted. The next 12 won’t be.
- The political risk is not priced in. Mass layoffs framed as “AI-driven” with no reskilling, no social reinvestment, and no transparency from AI leaders is going to become a 2026 political flashpoint. If you’re a founder shipping agents that replace humans, start thinking now about your social narrative. “We did it, we put you out of a job” is not a winning posture.
Quotable Moments
Harry Stebbings
“Is this the mother of all pops, with retail getting behind Elon in a way we haven’t seen before? It has to do better than GameStop.”
“Every public company CEO wants your advice on agents and AI. Are you more bullish on Klaviyo post seeing the inner workings of Andrew? Is he a top 1% public company CEO on AI?”
“Jason says five trillion. I say three. You’re negative. That’s the spread.”
Jason Lemkin
“Andrew was the only person who got the number right. Because he’s actually doing it.”
“Hurry up, Breeze. You have 300,000 HubSpot customers just waiting to buy an AI SDR from you. If HubSpot could actually make this as good as a startup, they’re going to sell 150,000 of them.”
“We’re going to have to reflate and hire thousands and thousands of people per tech leader to avoid social unrest. The Eric Schmidts can’t just go to graduation and say ‘F you.'”
Rory O’Driscoll
“If I want to raise $30 billion, I should probably get a check for $30 billion and call it a day. That seems to be the Anthropic philosophy.”
“At 1% of R&D spend, it’s lost in the noise. At 5%, it’s real. That’s a layoff. At 20%, which is what it takes for the math on these labs to work, that’s a fifth of your engineering payroll. Huge.”
“At least when Meta was busy destroying the world, they were smart enough to pretend it was all about bringing friends together and not destroying democracy. We’ll regret that lack of transparency. The people who are utter morons at AI but brilliant at politics are going to have us for lunch.”
This post is part of the ongoing 20VC x SaaStr collaboration.
