NextFin

Anthropic IPO Gives SK Telecom’s Chey Tae-won a $3.2 Billion Paper Windfall

Summarized by NextFin AI
  • Anthropic priced its IPO at $83 per share, above the marketed range, creating a public valuation reference for its previously private business.
  • SK Telecom’s approximately 38.6 million shares are valued at $3.2 billion, but lockups, taxes, restrictions, and market volatility mean this remains an unrealized paper gain.
  • The IPO is designed primarily to fund Anthropic’s compute-intensive expansion, making public valuation and financing costs increasingly important to enterprise AI growth.
  • AI infrastructure demand appears structurally strong, but valuation remains cyclical; future results will depend on revenue quality, customer retention, margins, infrastructure spending, and cash consumption.

NextFin News - Anthropic’s $83-a-share IPO has turned SK Telecom’s early investment in the Claude developer into a $3.2 billion paper asset for billionaire Chey Tae-won’s empire, but the more important market test begins after the pricing headline. The listing gives public investors a way to value the strategic link between an AI model company and Korean technology groups, while forcing Anthropic to prove that extraordinary demand can support extraordinary infrastructure spending. The windfall is visible today; the durability of the business model is not.

Anthropic priced the transaction above the $70-to-$78 range marketed in its roadshow materials. The deal, combined with a private placement, raised $107.3 billion in total for Anthropic and its seven co-founders, according to the transaction details. SK Telecom’s stake, equivalent to about 38.6 million shares after Anthropic’s 10-for-1 split in September, was valued at approximately $3.2 billion at the IPO price.

That figure is a mark, not a cash distribution. The company’s offering created a public reference price for a private holding, but it did not remove the usual constraints on monetization: lockups, strategic restrictions, taxes, market impact and the possibility that trading prices diverge from the initial allocation price. For Chey’s group, Anthropic has moved from a venture investment on a balance sheet to a liquid-looking asset that will be repriced every trading day.

A Public Price Does Not Equal a Realized Return

The simplest calculation explains the headline: roughly 38.6 million shares at $83 equals about $3.2 billion. The harder calculation is what SK Telecom can actually realize and when. The carrier’s reported Anthropic holding value had already reached 3.51 trillion won, or approximately $2.65 billion, at the end of June 2026, up from 1.38 trillion won at the end of 2025. The IPO made the value more observable, but it did not guarantee that the quoted price would remain intact once broader investors could trade the stock.

The offer price also exceeded the top of the roadshow range by $5 a share, a premium of about 6.4%. That premium signals demand for the deal, but it is not the same as proof that Anthropic’s long-term economics have been validated. IPO pricing is formed at the intersection of scarcity, allocation strategy and investor expectations. A public company must then repeatedly earn the valuation through revenue growth, margins, cash generation and credible capital allocation.

Anthropic’s roadshow materials described 825 million Class A shares, including 801.35 million shares offered by the company and 23.65 million shares offered by existing stockholders. The underwriters also received a 30-day option to purchase up to 123.75 million additional shares. The structure matters because most of the proposed primary offering was designed to put capital into Anthropic rather than create an immediate exit for existing investors.

That financing function explains why the IPO matters beyond Chey’s paper gain. Anthropic is raising capital to expand a business whose principal constraint is not only software development but the cost and availability of compute. The company’s public valuation therefore becomes a financing input: a higher equity value can lower the effective cost of future capital, while a weak aftermarket can make the next round of infrastructure spending more expensive.

“Startups and Global 5000 companies alike are deploying Claude to handle complex workflows, and in doing so, Claude is learning how businesses actually operate: the context, the processes, the judgment,” Sequoia Capital partner Alfred Lin said in Anthropic’s funding announcement.

Lin’s observation points to the operating question beneath the IPO. Enterprise adoption matters only if deployment creates recurring demand at a price that exceeds the cost of serving the workload. The market will eventually need to distinguish usage growth from profitable usage growth.

The Second-Order Effect Runs Through the Chip Supply Chain

The IPO’s first-order effect is a visible uplift in the value of SK Telecom’s Anthropic stake. The second-order effect is a potential change in how investors value the companies supplying AI infrastructure. Anthropic’s Series H announcement named Micron, Samsung and SK hynix as strategic infrastructure partners, linking the model developer’s growth plans to memory, storage and logic-chip capacity.

That relationship does not establish a fixed purchase commitment or guarantee revenue for any supplier. It does show that the AI stack is becoming financially interdependent. Model companies need compute; cloud providers finance and operate it; memory and semiconductor companies expand capacity; strategic investors seek both financial returns and commercial access. An IPO can strengthen that loop by giving all participants a common public-market benchmark.

The mechanism is straightforward. If investors accept Anthropic’s public valuation, the company may have more room to fund capacity and sign long-term supply or cloud agreements. Suppliers can then plan around a more credible demand signal. If investors reject the valuation, the immediate effect may appear first in the cost of capital rather than in chip shipments. Infrastructure projects that looked attractive at private-market marks may be delayed, resized or renegotiated when public shareholders demand evidence of returns.

This is why the structural and cyclical components should not be blended. The structural shift is the growing importance of specialized infrastructure for enterprise AI. Anthropic’s strategic partnerships, massive financing needs and public listing all support the view that AI capacity is becoming a durable part of the technology economy. The cyclical component is the price investors are willing to pay for that future at a particular moment. Capital can leave expensive AI assets even while enterprises continue adopting AI tools.

The supply-chain asymmetry is easy to miss. A strong public debut can support memory demand because it reassures investors that model developers can keep financing expansion. But a weak public-market response would not instantly end current orders; it could instead affect the next investment decision. The impact would travel through budgets, financing terms and capacity plans before it appeared in reported supplier revenue.

The IPO Tests Whether AI Growth Is Structural or Merely Re-Rated

The evidence supports a two-part judgment: AI infrastructure demand is structural, while the $3.2 billion mark on SK Telecom’s stake is cyclical. The structural case rests on a change in technology architecture and corporate spending. The cyclical case rests on a valuation transfer from private rounds to a public security and can reverse without the underlying technology disappearing.

Three recurring patterns support caution about the cyclical leg. Private technology valuations can rise faster than audited operating performance during funding waves, then face a harder test when public investors receive regular disclosures. IPO prices can exceed marketed ranges when demand is concentrated, only for aftermarket trading to separate scarcity from durable ownership. Hardware suppliers can enjoy a rapid capacity buildout and later experience inventory digestion even when the long-term application remains intact. These patterns are mean-reverting because enthusiasm, financing and orders often move ahead of verified cash returns.

The strongest counter-thesis is that Anthropic’s economics may be different from those of a conventional software issuer. Its strategic relationships with hyperscalers and infrastructure suppliers could make it an anchor customer in a market where capacity, not demand, is the binding constraint. Anthropic said its Series H raised $65 billion at a $965 billion post-money valuation and included $15 billion of previously committed hyperscaler investments, including $5 billion from Amazon. If those commitments translate into sustained enterprise deployments, the IPO could validate a financing regime in which model developers, cloud platforms and chipmakers share the upside of AI expansion.

That counter-thesis attacks the central bearish argument: public scrutiny may not compress the valuation if Anthropic’s strategic position makes future demand unusually visible. It cannot be dismissed as simple enthusiasm. Yet strategic importance does not guarantee attractive unit economics. Customers can use more models while the provider earns less per unit of computation if competition lowers prices, model capability becomes interchangeable or inference costs remain high.

The market’s conventional wisdom is that Anthropic’s IPO is good news for AI. The less obvious question is who captures the value created by the spending cycle. Suppliers may retain pricing power if memory remains scarce. Cloud providers may capture revenue through hosting and financing. Enterprise customers may capture productivity gains. Anthropic captures the most value only if its models become embedded in workflows with high switching costs and strong contribution margins.

The falsifying signal for the structural thesis would be a sustained decline in Anthropic’s infrastructure commitments combined with deteriorating enterprise retention or usage growth. For the valuation thesis, a clear warning would be ANT trading below the $70 lower end of the marketed range after the first full quarter of public reporting. Neither signal would prove that AI demand had ended, but each would show that the public-market transfer of confidence had outrun the underlying economics.

What the Mark Means for Chey Tae-won’s Group

Chey Tae-won’s group gains a public reference point for an asset that previously had to be valued through private financing marks and accounting estimates. SK Telecom can use the stake as evidence that its early technology strategy created value, while SK hynix and other affiliates can frame their AI exposure as part of a connected ecosystem rather than a stand-alone semiconductor bet.

That benefit has limits. SK Telecom is not Anthropic, and a minority holding does not automatically flow through to the carrier’s operating earnings or share price. Investors will ask whether the stake can be sold, whether proceeds would be reinvested, and whether the telecom business is improving. A high public mark can increase pressure on management if the group’s market value fails to reflect the value of its private or strategic assets.

The listing also introduces governance and control questions. Anthropic’s Class A offering sits alongside other classes of stock, and the company is a public benefit corporation. Public investors must therefore assess how mission commitments, strategic investors and voting arrangements affect ordinary shareholders. These issues are not cosmetic: control rights can determine who decides on future financing, partnerships and capital spending.

The upside scenario is a reinforcing loop. Anthropic uses IPO proceeds to expand enterprise adoption, demand for compute rises, strategic suppliers deepen their relationships, and SK Telecom’s investment gains support from operating progress rather than valuation momentum. The trigger would be revenue expansion accompanied by durable customer retention, improving gross economics and continued infrastructure commitments.

The downside scenario is a public-market reset. Anthropic could keep growing usage but spend so aggressively to preserve model leadership that investors lower the valuation multiple. SK Telecom’s mark would fall, and the strategic-investment narrative would weaken. The trigger would be a widening gap between revenue growth and cash consumption, or a cut in planned infrastructure spending.

The base case is mixed across time horizons. In the short term, the IPO keeps attention on SK Telecom and Korean semiconductor exposure to AI. In the medium term, Anthropic’s reported operating metrics will matter more than the $83 allocation price. In the long term, the partnership can create durable value if AI applications become embedded in enterprise processes, but the gains will be distributed unevenly among model developers, cloud platforms, carriers and chipmakers.

The next information that matters is not another valuation headline. It is Anthropic’s public disclosure of revenue quality, customer expansion, infrastructure commitments and cash consumption, alongside SK Telecom’s explanation of how it treats and may monetize the holding. The broader AI trade will be judged by the relationship between infrastructure spending and the revenue generated by the applications using it.

SK Telecom has won a valuable mark, but not yet a final return. Anthropic’s IPO is the moment when a private AI promise becomes a public operating obligation—and when the cost of making that promise real moves onto every balance sheet connected to the stack.

Explore more exclusive insights at nextfin.ai.

Insights

What drove Anthropic's IPO pricing?

How did SK Telecom value its stake?

Why is SK Telecom's gain unrealized?

What does the IPO fund?

How does compute shape Anthropic's costs?

What role do Micron and Samsung play?

Can AI usage become profitable?

What could weaken Anthropic's valuation?

How might scrutiny change AI spending?

What risks face SK Telecom's investment?

How does Anthropic differ from software?

What would confirm durable AI demand?

What signals could trigger a reset?

How may AI value split across the stack?

What will public investors watch next?

Search, ask or analyze anything...