Squawk on the Street

Anthropic IPO: AI Safety, Valuation and the Pure-Play AI Trade

· Summarized

Share this episode
Anyone with this link can open it.
Share Link
Start at 0:00

Key takeaways

  • Anthropic priced its IPO at $83 a share after raising the range and enlarging the offering, with indications around $105 implying a valuation near $2 trillion.
  • The Amodeis say Anthropic’s public-benefit-corporation structure legally requires it to balance shareholder returns with the public good.
  • Anthropic presents safety and deliberate pacing as a competitive and commercial advantage, arguing that trusted systems will perform better over the long term.
  • Management and investors identify frontier model capability as the core moat because it enables high-value applications such as coding, biology, robotics, and cybersecurity as well as cheaper smaller models.
  • Compute is the company’s defining financial requirement: Anthropic disclosed about $900 billion in leases and commitments but expects roughly $250 billion in post-IPO cash.
  • Wellington’s Matt Wietler says enterprise adoption creates switching costs through security, integrated tools, and return on investment, and that Anthropic and OpenAI can both become successful public companies.
  • Melius Research’s Ben Reitzes assigns Anthropic a $155 price target, citing falling serving costs, potentially high margins, rapid model releases, and growth into global knowledge work.

Summary

Anthropic’s IPO points toward a $2 trillion valuation

Squawk on the Street opens by calling Anthropic’s IPO the first major pure-play, native AI company to reach public markets. Hosts Sarah Eisen, Carl Quintanilla, and David Faber say the brother-and-sister founders, Dario and Daniela Amodei, started the company less than six years earlier, after leaving OpenAI over safety concerns. Anthropic was valued around $1.5 trillion based on indications, with estimates rising toward $1.8 trillion or more; 18.2 billion fully diluted shares would allow viewers to calculate the market value. The hosts contrast Anthropic’s unusually stable leadership with turnover at OpenAI.

Enterprise adoption expands Anthropic’s market beyond coding

The hosts describe enterprise AI as the fastest-adopted technology in history, exceeding cloud adoption. Anthropic’s S-1 frames its total addressable market as the wages of global knowledge workers, although the hosts note that wages are not the same as eventual product spending. Coding is presented as the clearest proof that AI can replace or augment work; healthcare, legal services, accounting, defense, pharmaceuticals, cybersecurity, and drug discovery could follow. They warn that information-services job growth has fallen since ChatGPT-3 and ask whether future models will replace knowledge workers rather than merely make them more productive.

Demand pushes Anthropic’s IPO proceeds sharply higher

Leslie Picker reports that Anthropic priced its IPO at $83 a share, $5 above the marketed range, and increased the offering by 155 million shares. At an indicated $103.50, the stock would gain about 25% and imply roughly a $1.9 trillion valuation. The enlarged deal produced about 26% more proceeds than initially sought, an important advantage because Anthropic must fund compute and infrastructure. Picker says the company could have approximately $250 billion in pro forma cash, but also disclosed about $900 billion in leases and other commitments. The first trade was still at least 90 minutes away.

The Amodeis defend public-benefit status and deliberate pacing

In an interview with Andrew Ross Sorkin, Dario Amodei says scaling laws predicted increasing model capabilities, but the founders did not believe the consequences until they saw them in the world. Daniela explains that Anthropic became a public benefit corporation because it is legally required to balance shareholder interests with public benefit. Asked how growth can coexist with safety, Dario compares AI to a car that could go 200 miles per hour: Anthropic could move faster, but chooses roughly 90 miles per hour while building safeguards. He argues that trust and safety should ultimately benefit both society and investors.

Anthropic says frontier performance is its competitive moat

Dario argues that AI competition is primarily about quality rather than price because the economic value rises dramatically at the frontier: models progress from chatbots to coding, biology, and robotics. Frontier capability also lets Anthropic make cheaper, smaller models; he cites Haiku 5.5 as costing four times less than Haiku 4.5 and far less than the Opus and Fable 5 families. The hosts and executives discuss open-weight models, OpenAI, Microsoft, and token-price competition, but Anthropic’s position is that only a small number of firms can reach the frontier and then efficiently serve lower-cost use cases.

Compute commitments, safety liability, and partnerships dominate investor questions

Dario says the roughly $900 billion of compute commitments are spread across years and are justified by the historically high returns from greater model capability. Anthropic has about $250 billion in cash after the offering and hopes eventually to obtain investment-grade status and use debt. Daniela says Anthropic accepts responsibility for safe training and safeguards, learning risk factors through enterprise customers; she reiterates that safety was the company’s founding purpose. The executives say Anthropic works with NVIDIA, Amazon, Google, and Broadcom, disagrees with NVIDIA on some China-chip policy, continues to find SpaceX compute high quality, and expects cooperation with the Pentagon to continue.

The founders say the IPO changes little about their daily mission

Asked about the pressure of going public during a difficult market, the Amodeis say they are calm and focused on fundamentals, repeating their principle of under-promising and over-delivering. They decline to comment on whether Jensen Huang should return to NVIDIA and do not provide a Pentagon update. Asked about Elon Musk’s SpaceX contracts, they say they enjoy working with SpaceX but cannot predict future purchasing decisions. Daniela says the IPO is a major milestone but Anthropic remains the same company pursuing its mission; she and Dario also hope to spend more time together as siblings after the event.

Wellington investor says the frontier can support multiple AI winners

Wellington’s Matt Wietler calls Anthropic’s move from zero public-market value to a potential $1.8–$2 trillion company an exceptional moment for AI. He says public investors previously had infrastructure and hyperscaler exposure but no pure-play AI company, while Anthropic was approaching $100 billion in annualized revenue and had $250 billion available on its balance sheet. Wietler sees two defenses: frontier performance is needed for biology and complex, long-context tasks, and frontier expertise produces efficient lower-cost models. He says Anthropic and OpenAI are not mutually exclusive, enterprise customers are sticky because of security and ROI, and a temporary safety pause would be appropriate if needed.

Analyst Ben Reitzes sets a $155 target using hyperscaler economics

Melius Research analyst Ben Reitzes initiates Anthropic with a buy rating and a $155 price target. He compares the company with hyperscalers, applying six times his 2029 enterprise-value-to-revenue estimate, roughly the current forward multiple for hyperscalers, while arguing that Anthropic deserves support from faster growth and potentially higher margins. He says cost to serve fell from $2.05 per dollar of revenue two years earlier to $0.14 and could reach single-digit cents as reinforcement learning and model efficiency improve. Reitzes identifies recursive self-improvement as the central safety challenge and says Anthropic must explain how much human oversight remains.

Demand, perpetual futures, and AMD’s China exposure shape the rest of the hour

Picker says institutional orders reached about $400 billion and retail orders about $100 billion; Anthropic prioritized existing institutional relationships, while retail received roughly 15% of the offering. Perpetual futures on Hyperliquid indicated a valuation above $1.8 trillion and had $325 million in 24-hour volume, with traders implying another 20–30% rise after the $83 IPO price. The program then turns to AMD, whose shares fell about 10% after Hunterbrook alleged ties to Chinese server maker IEIT, firmware appearing in potentially blacklisted companies, and more than 20% of revenue coming from China and Hong Kong. AMD was contacted for comment, while Hunterbrook disclosed a short position.

Sponsors

  • Fidelity · Fidelity says investors can start with as little as $1, with no account fees or trade commissions on U.S. stocks and ETFs; recurring investments can also be scheduled. The ad includes risk and eligibility disclosures.
  • Chumba Casino · Tyler Reddick and Bubba Wallace promote Chumba Casino as entertainment during race rain delays, featuring social casino slots, bingo, and solitaire. No purchase necessary; 21+ and other restrictions apply.
  • VantageScore · VantageScore says its modernized credit scoring uses 400% more credit and alternative data, including trended credit, rental, and utility-payment information. It claims VantageScore 4.0 brings five million mortgage-ready borrowers into the market and is used by Rocket Mortgage and nine of the ten largest banks.
  • State Street · State Street promotes sector ETFs as a way to gain exposure to leading companies across the 11 sectors of the U.S. economy without selecting individual stocks. The ad includes standard investment-risk disclosures.
  • Half Price Books · Half Price Books advertises gifts including games, toys, comics, records, and collectibles, available in stores and online, and jokes that shoppers can buy something for themselves too.
  • Schwab Market Update · Schwab promotes its daily market-preview podcast, covering stocks, U.S. and global economic news, monetary policy, results, and statistics that may affect trading.

Show notes

November 19, 2026; 10am: Squawk on the Street covers Anthropic’s historic IPO, speaking exclusively with Dario and Daniela Amodei on safety, compute, competition, and public-market expectations. The hour also tracks investor demand, analyst targets, perpetual futures signals, and the broader market implications of pure-play AI coming to Wall Street.


Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Follow this show on AudioBrevity and new episodes get summarized for you. Create a free account

Couldn't load the audio. A blocker may be stopping the show's tracking links. Open the file