CNBC Closing Bell · · 48 min

Anthropic’s Record IPO, AI Selloff and Market Rout

With Adena Friedman, Binky Chadha, Bradley Tusk, Guru Chahal, Leslie Picker, Megan Cassella, Nelson Griggs, Pippa Stevens, Ray Wang

ChorusBring the market's conversation to your AI. Learn more

Summary

CNBC's Closing Bell covered the Anthropic IPO and a broader tech and AI selloff. Guru Chahal of Lightspeed argued Anthropic's valuation reflects strong fundamentals, citing about $17B of Q3 revenue, 84% gross margins and 900% net dollar retention, and said its compute constraints and small share of Fortune 500 spend leave room to grow. Bradley Tusk pushed back, arguing the roughly 30x ARR multiple is far above Nvidia, Alphabet and Apple, that about $900B of capex commitments weigh on shareholders, and that rival labs, Meta's free product and state-level AI regulation threaten Anthropic. Binky Chadha of Deutsche Bank argued the market is oversold, that earnings are still growing strongly and that the Anthropic IPO was a failed test for renewed AI appetite because gains were zero-sum across AI names. Mike Santoli added that the market is debating post-2028 data center build-out, and Nelson Griggs said companies are delaying listings until 2027 given high volatility.

Key takeaways

  • Guru Chahal (bullish on Anthropic): Anthropic's business quality, growth and margins justify its valuation; long-term bullish on the company.
  • Ray Wang (bullish on NVIDIA): Nvidia's chip demand is still playing out and will run through 2028–2029 despite recent uncertainty.
  • Nelson Griggs on IPOs, M&A & deal activity: Most companies will wait until 2027 for a calmer market, so the IPO pipeline stays limited near term.
  • Binky Chadha (uncertain on Anthropic): Anthropic's IPO, priced near the lower end of relative performance, is a positive signal, but could get caught in a tech downdraft.
  • Bradley Tusk (bearish on Anthropic): Anthropic's valuation is far above peers' ARR multiples, so he would not buy the stock.

What was said, by company and theme

20 views from 6 speakers. Every quote is checked against the episode's transcript (how).

Anthropic

Across the shows: 147 speakers in other episodes, 260 bullish and 118 bearish calls, past 30 days

Where everyone stands on Anthropic

  • Guru Chahal · Lightspeed Venture PartnersBullish

    Anthropic's business quality, growth and margins justify its valuation; long-term bullish on the company.

    Lightspeed, where Chahal is a partner, has invested in every Anthropic round since 2023. He argues the valuation reflects business quality, not just a liquidity event. Revenue reportedly hit $17B in Q3, about 12x year over year, with 84% gross margins and 900% net dollar retention, meaning customers spend about $9 for every $1 of a year earlier. He says the company has beaten its own projections in each year since 2024. He concedes growth will slow as the law of large numbers takes hold, but says Anthropic is supply-constrained rather than demand-constrained, so each dollar of compute yields about $1.7 of top-line revenue. Its share of Fortune 500 spend is under 1%, leaving room to grow. He calls himself a long-term investor and says there could be upside.

    “this business just closed a $17,000,000,000 [Q3 revenue] this year. That's 12 x year over year, which at that scale is practically unheard of. ... This is at 84% gross margins.”at 9:46
  • Guru Chahal · Lightspeed Venture PartnersBullish

    Falling per-token prices are a red herring; enterprise value and ROI from intelligence drive Anthropic's economics.

    Responding to the argument that frontier labs are just token merchants on a deflationary price curve, Chahal says per-token price misses the point. Producing frontier intelligence is hard, requires a rare combination of talent, compute and serving capability, and few firms can do it. What matters is how much value the intelligence creates in enterprise use cases. Customers spending $9 for every $1 a year earlier shows they are getting ROI above their spend. He says he is very bullish over the long term on how Anthropic develops and delivers its models.

    “Per token price is a bit of a red herring if you really think about it. What really matters is how much value is this intelligence adding in actual enterprise use cases?”at 14:56
  • Ray Wang · Constellation ResearchBullish

    Anthropic and Google have captured the enterprise market; OpenAI is prosumer-focused and Meta consumer-focused.

    Wang compares the frontier labs by customer base. Anthropic and Google have done a good job capturing the enterprise market, and Anthropic has the demand. OpenAI is mainly in prosumer and small-to-medium business, and Meta is consumer-focused. He sees enterprise capture as the differentiator among the labs.

    “on the Anthropic side and on the Google side, they've done a really good job capturing the enterprise market.”at 19:03
  • Binky Chadha · Deutsche BankUncertain

    Anthropic's IPO, priced near the lower end of relative performance, is a positive signal, but could get caught in a tech downdraft.

    Chadha says it is remarkable that a big IPO came at the low end of tech's relative performance, which he calls a positive signal for the IPO market. He flags a risk, though: IPOs pop and then often reverse, and Anthropic already had a pop and a partial reversal. If the stock gets caught in the tech downdraft, it could fall with the sector. He calls the current period the fifth episode of fear in two and a half years.

    “it's pretty remarkable that we had a big IPO at the end, at the lower end of basically relative performance. And so that is a positive signal.”at 35:47
  • Bradley Tusk · Tusk VenturesBearish

    Anthropic's valuation is far above peers' ARR multiples, so he would not buy the stock.

    Tusk compares Anthropic's valuation with peers on ARR multiples: Nvidia about 8x, Alphabet about 7x and Apple about 11x, against roughly 30x for Anthropic. He adds that about $900B of capex commitments sit senior to common shares, so they weigh on what shareholders own. He notes that Anthropic's growth already fell sharply from Q2 to Q3, even though its run-rate revenue of $81B is large. On those grounds he says he would not buy at a roughly $1.3T market cap.

    “[Anthropic is at] 30 because not only is it the valuation, but then you have to factor in the $900,000,000,000 of CapEx commitments that sit senior to common shares.”at 41:59
  • Bradley Tusk · Tusk VenturesBearish

    Anthropic faces intense rivals, and free offerings like Meta's will hurt its consumer business.

    Tusk argues Anthropic faces many direct rivals: OpenAI, Grok, Gemini and DeepSeek. He singles out Meta, which offers its product for free, so users do not need a subscription. He says this will hurt Anthropic's consumer business. He also argues that Meta and Google fund their frontier labs from core businesses that make large profits, giving them more runway than Anthropic, which has to fund its losses itself.

    “You have OpenAI, you have Meta who's offering a free product. So, people don't have to pay for a subscription. That's really going to hurt [Anthropic's] consumer business.”at 42:29
  • Bradley Tusk · Tusk VenturesBearish

    Anthropic faces heavy regulatory headwinds, with state rules and voter hostility to AI.

    Tusk says data centers are hard to build right now because voters dislike AI. He expects all 50 states to pass rules regulating frontier models this year, even though Trump has floated federal regulation. In his view Anthropic would then have to comply with 50 different frameworks, which he calls a total mess. He adds that the public uses AI but hates it, so he sees the regulatory burden as a major drag on Anthropic.

    “Data centers are really hard to build right now because the voters hate AI... All 50 states this year gonna pass something to regulate frontier models. Anthropics gonna deal with 50 different regulatory frameworks.”at 42:45
  • Bradley Tusk · Tusk VenturesBearish

    Anthropic needs to capture nearly all of the enterprise AI market to justify its valuation.

    Tusk cites an expected enterprise AI market of about $1 trillion by 2030, with Anthropic's segments, such as agentic infrastructure, around $300 billion. With peers trading at high multiples, he argues Anthropic would need to capture almost the whole market. That means rivals would have to be wiped out, and Anthropic would have to beat all of them completely. He says that does not feel likely.

    “...the entire enterprise AI market is expected to be at $1,000,000,000,000... [In] the anthropic segments like agentic infrastructure and things like that, it's about 300,000,000,000. If they're anywhere near the multiples of the other companies in that range, they have to capture almost the entire market.”at 43:38
  • Bradley Tusk · Tusk VenturesBearish

    Public-market pressure will push Anthropic to release riskier models faster, which worries him.

    Tusk argues that under public-market pressure, quarterly earnings and the need to grow the stock, Anthropic has to keep releasing new models, because that is how it makes money. Each new release carries more safety risk. He says CEO Dario Amodei, who has called the company an existential risk, will have to choose every day between the stock and safety. He presents this as a structural conflict that worries him.

    “The only way they make money is by releasing new models. And every time you release a new model, you're taking a lot more safety risk.”at 44:57

NVIDIA

Across the shows: 177 speakers in other episodes, 184 bullish and 172 bearish calls, past 30 days

Where everyone stands on NVIDIA

  • Ray Wang · Constellation ResearchBullish

    Nvidia's chip demand is still playing out and will run through 2028–2029 despite recent uncertainty.

    Wang says the uncertainty weighing on the AI trade is a mix of Nvidia's scandal, interest rates and the change of control in Congress. Underneath it, demand is real: Anthropic has enterprise demand, and Nvidia's chip demand is still playing out. He expects that demand to run through 2028–2029. The market worries about the data center buildout and China, which he says will keep weighing on sentiment.

    “You see the chip demand from NVIDIA is still playing out. That's going to be going till 2028, 2029.”at 16:52

US stock market outlook

Where everyone stands on US stock market outlook

  • Michael Santoli · CNBC

    The 2018 midterm-year pattern remains in play, so the market could keep weakening despite being oversold.

    Santoli says the 2018 scenario still applies. This is a midterm election year, and stocks sold off through the fourth quarter as the Fed was tightening, with other factors also in play. Even though the market looks oversold, he says investors should keep that pattern in mind. The seasonal tailwinds, Nvidia's results and the Anthropic IPO were supposed to be the all-clear, but they did not stop the selloff.

    “The other thing is the 2018 scenario remains in place. ... This is a midterm election year and you sold off in through that fourth quarter.”at 4:04
  • Binky Chadha · Deutsche Bank

    The US market, including tech, is oversold and washed out; the selloff has overshot fundamentals.

    Chadha says the selloff has overshot. Deutsche Bank downgraded tech on October 8 because of its relative performance and positioning, then upgraded it three weeks later as tech sat at the bottom of its relative-performance channel. Positioning and valuation across the broader market now look washed out. S&P 500 equity positioning is pricing in an earnings contraction, even though earnings grew about 34% in each of the last two quarters and bottom-up consensus is close to 30% for Q4. The market is pricing in only 1–2% growth. He concludes the market is in a very oversold place.

    “So we are in a very, very oversold place.”at 35:12
  • Binky Chadha · Deutsche Bank

    The market selloff is overdone; fundamentals do not show a true downturn.

    Chadha calls this the fifth 'fear and loathing' episode in about two and a half years, and says he is not ready to believe this is the start of a lasting downturn because he does not see it in the fundamentals. He says the market is not paying attention to good macro data and that the key is to let up on the daily negative catalysts in tech. He concedes that the timing of the bottom depends on catalysts, and that the next move in data surprises should be down from a high peak.

    “I characterize it as the fifth episode of fear and loathing. We've had five episodes in the last two and a half years... I'm not sort of ready to believe that, you know, this is it because I don't see it in the fundamentals”at 36:13

AI infrastructure & data centers

Where everyone stands on AI infrastructure & data centers

  • Michael Santoli · CNBC

    Capex spending is secure near term, but the market is unsure what happens after 2028.

    Santoli says the market accepts the near-term capex story: Anthropic has about $200B of commitments that will be spent. The market is instead debating what comes after, which he calls a 2028 story that is unknowable. That uncertainty about the post-2028 build-out is what is weighing on AI-linked stocks.

    “We all know the capex story is gonna be fine now. Anthropic has $200,000,000,000. It's gonna get spent.”at 3:51
  • Bradley Tusk · Tusk Ventures

    Anthropic's roughly $900B of compute commitments and rising rates create serious debt-cost risk.

    Tusk says the roughly $900B of capex commitments equal about eleven years of Anthropic's total revenue. Rising interest rates and inflation would raise its debt costs. He concedes that the company cannot run without the compute, and that if the commitments are scaled down to about $450B with the multiple falling to 27, the analysis still holds. He is treating the capex load as a structural drag on the equity.

    “So the 900,000,000,000 now in the bank, it's eleven years of their total revenue. And then also, as interest rates keep going up and inflation keeps going up, the debt costs go up as well.”at 45:54

US growth & recession risk

Where everyone stands on US growth & recession risk

  • Binky Chadha · Deutsche Bank

    US growth is firm and recession is far off absent a new shock; recession fears are overdone.

    Chadha says the market has started to price in a possible recession as negative catalysts pile up and risk aversion spreads. Deutsche Bank's view is that growth is very firm and a recession is very far away unless a new shock hits. Combined with earnings that are still growing strongly, he argues the selloff is driven by sentiment, not a deteriorating economy.

    “We think growth is very, very firm. Recession is very, very far away, absent some new shock.”at 35:05
  • Binky Chadha · Deutsche Bank

    Macro data has surprised positively for a year, so consensus expectations are too low.

    Chadha says macro data surprises have been positive for a year, which shows consensus expectations are low and have stayed low. He says the economy is in good shape, which contrasts with the market's focus on tech. He concedes that from a high peak the next move in surprises should be down, but argues the good data still supports a view that growth is stronger than the market prices.

    “Macro data surprises have been positive for a year now. So that's telling you that the consensus in terms of expectations is a bit low and persistently so.”at 37:14

Enterprise AI adoption & AI software

Where everyone stands on Enterprise AI adoption & AI software

  • Ray Wang · Constellation Research

    Companies building on top of frontier models, not the labs, will capture the biggest competitive advantage.

    Wang argues the ecosystems built on each frontier lab are where the real advances will show up. Companies that build on top of models and add their own data are the ones gaining a competitive advantage. He puts the edge at roughly ten to one for firms that apply these models, because the game becomes gathering more data and refining signals to improve precision. He suggests investors look at the application layer and SaaS stocks.

    “Companies building on top of these models, adding to those models. These are the ones that are getting the competitive advantage. It's a 10 to one competitive advantage for those who are actually applying these models”at 19:03

IPOs, M&A & deal activity

Where everyone stands on IPOs, M&A & deal activity

  • Nelson Griggs · Nasdaq

    Most companies will wait until 2027 for a calmer market, so the IPO pipeline stays limited near term.

    Griggs says the volatility made many companies pause and ask whether it is the right time to list. The fall pipeline is weaker than he expected three months ago. Only a few deals of size are likely in the two weeks before the holidays. Most companies want to get through 2026 and look toward 2027, when a new administration and Congress may bring predictability. The VIX is high, so companies need a more constructive market before they hit go.

    “most companies are saying, need to get through 'twenty six and look forward to 'twenty seven.”at 29:57

US jobs & labor market

Where everyone stands on US jobs & labor market

  • Binky Chadha · Deutsche Bank

    The labor market is improving, and employment and real wages should keep rising.

    Chadha says the labor market is improving. The employment recovery outside education and healthcare should keep trending up over time. He ties it to the ISM manufacturing data, which supports his view that it should continue. Real wages have been flat for a year but are up a bit in recent months. If companies keep hiring, real wages and incomes should rise, which supports consumer confidence.

    “The labor market's actually improving... the recovery in employment outside of education and healthcare would just simply trend up over time. It's... in line with, for example, the ISM manufacturing.”at 37:40

Across the shows

What other episodes Chorus covered said about the same companies in the 30 days up to this one.

  • NVIDIA177 speakers · 184 bullish, 172 bearish calls
  • Anthropic147 speakers · 260 bullish, 118 bearish calls

Ask your AI who they are and what they said

Topics

  • AI trade sentiment
  • IPO market signal
  • IPO pipeline
  • application-layer advantage
  • chip demand
  • competitive pressure
  • data center capex
  • employment recovery
  • enterprise AI market share
  • enterprise market share
  • macro data surprises
  • midterm-year seasonality
  • oversold conditions and positioning
  • perpetual futures competition
  • public-market pressure on safety
  • recession risk
  • revenue growth and margins
  • selloff outlook
  • state AI regulation
  • token pricing and commoditization
  • valuation multiples

More from CNBC Closing Bell

All CNBC Closing Bell episodes

Ask your AI about this episode

Chorus gives Claude, ChatGPT and other assistants every view from this episode and every other show Chorus follows. Ask who agrees, who changed their mind, and what else was said about the same companies.

A weekly email with the calls that mattered from the most important voices in investing. Unsubscribe any time. Privacy Policy

Chorus reports what speakers said. It isn't investment advice. How Chorus works · About Chorus