Zoom's Anthropic stake puts AI marks onto balance sheets

Zoom Communications’ Anthropic stake was worth about $1.27 billion in a Thursday filing, leaving analysts with a new question about the videoconferencing group: how much of the equity story now sits in a private AI mark rather than in the slower, better-known software business. The paper gain is about $1 billion on an investment Zoom made in early 2023, according to Bloomberg. That is big enough to matter on its own. It also offers one of the clearest public examples yet of the AI boom leaking out of venture rounds and into listed-company balance sheets.

Seen through a valuation model, that changes the frame. Anthropic’s fundraising and revenue sprint are no longer just inputs into private-market gossip. They are part of how investors will read a public company’s assets, capital allocation and, eventually, the quality of any earnings line touched by revaluation. When Anthropic’s second-quarter revenue run rate is discussed at $10.9 billion and Bloomberg reported the company is on pace for its first profitable quarter, the numbers do not stop with Anthropic. They travel.

Skeptics, though, read the same disclosure differently. A rising private stake can flatter the story around a listed company without producing a dollar of fresh cash from its core business. It can also encourage investors to treat the holder as a proxy for an AI name they cannot yet buy directly. That is useful while the mark is climbing. It becomes harder if pricing power softens, model costs fall or a rival grabs the better growth narrative.

What the mark actually changes

Zoom’s gain matters precisely because it is not operating performance. The company put roughly $51 million into Anthropic in 2023, and the stake is now worth about $1.27 billion, Bloomberg reported. That is a striking return on capital, but it does not say that Zoom suddenly fixed every question around post-pandemic demand, enterprise seat expansion or the durability of collaboration software spending. Analysts who follow the stock will need to separate the software franchise from the investment portfolio, then decide how much credit to give management for making the bet early. Some will reward the foresight. Others will strip the mark out and ask what is left.

From a disclosure lens, that distinction matters even if it is the driest part of the story. Unrealized gains can move headline perception faster than they move cash generation. They can also blur the line between a clean operating read and a portfolio revaluation. For investors accustomed to treating software companies as recurring-revenue machines, the appearance of a material private-company stake complicates the template. A quarter can now contain two stories at once: what the business sold, and what the market decided a separate AI asset was worth.

Management’s early positioning matters too. Zoom did not stumble into Anthropic by accident. An early stake in one of the leading model builders suggests management understood that generative AI could alter the economics of enterprise software before much of the public market had worked through the implication. In that sense, the mark is more than an accounting curiosity. It is evidence that treasury decisions, partnership strategy and product positioning are starting to overlap. That overlap is likely to widen as more listed software groups try to secure access to models, distribution or compute by taking minority positions rather than signing ordinary vendor contracts.

Why Anthropic’s valuation can carry weight

Supporters of the mark point first to Anthropic’s operating backdrop. CNBC reported that Anthropic is set to generate $10.9 billion in revenue in the second quarter, more than its sales for all of last year, while Bloomberg said the company is on track for its first profitable quarter. Those figures do not settle the valuation debate, but they make it easier to argue that the $380 billion February valuation was not pure speculative froth. Revenue at that pace begins to look like platform scale, not just venture ambition.

Usage from the field reinforces that point. Adoption can still be noisy, but demand appears broad enough to support the top-line story. A Business Insider report this week described Claude Code as the default coding tool across a swath of startups and venture-backed engineering teams, a sign that Anthropic’s products are moving from demo culture into workflow budgets.

“Everything that’s not Claude Code.”

• Dan Lorenc, chief executive of Chainguard, via Business Insider

Such evidence is what public-market investors want to see behind a private mark. Not another pitch about future transformation, but proof that enterprise users are already picking a model family, budgeting for tokens and changing how engineers work. It also helps explain why OpenAI’s reported first-quarter revenue of about $5.7 billion is not automatically a problem for Anthropic. The contest is not only about who is bigger. It is about whether more than one AI company can reach enough scale, fast enough, to justify IPO-sized valuations.

Why the proxy trade can wobble

Still, the proxy trade can wobble. CNBC’s reporting on the AI IPO race suggests markets increasingly view OpenAI and Anthropic as the next giant venture-backed listings, while a separate CNBC analysis argued that cheaper AI could still destabilize the pricing model behind those expected valuations. That warning matters for Zoom because a proxy trade works both ways. If Anthropic keeps compounding revenue and approaches public markets with credible profitability, the mark on Zoom’s books may look conservative. If model prices compress faster than revenue scales, the stake can start to feel like a volatile sidecar attached to a mature software name.

Scarcity adds to the effect. Public investors who want exposure to elite private labs still have almost no direct route, which is why coverage of the IPO race carries more weight than it would in an ordinary software cycle. When direct exposure is thin, any listed balance sheet that owns a meaningful stake can begin to trade like an option on a future listing. That can raise the holder’s strategic allure. It also imports private-market mood swings into what used to be a plain quarterly report.

Cost pressure is already visible. In the CNBC analysis, Google chief executive Sundar Pichai argued that token spending is running hot across the industry.

“many companies are already blowing through their annual token budgets, and it’s only May”

• Sundar Pichai, Google chief executive, via CNBC

All of that cuts to the core of the debate. Anthropic’s revenue surge can support Zoom’s paper gains today, but only if those revenues hold up once customers start demanding lower prices, better orchestration and clearer returns on AI spend. Public investors do not need Anthropic to fail for Zoom’s mark to become more contentious. They only need the growth curve to flatten enough that the multiple attached to it stops expanding.

More broadly, listed companies with stakes in elite AI labs may begin to trade partly as balance-sheet access points to private-model valuations. That could attract fresh capital to a small club of public names, especially while direct exposure remains scarce. It could also make quarterly reporting messier. Investors will have to decide whether they are buying operating execution, a venture portfolio in disguise, or a bit of both. Zoom’s Anthropic gain is a neat win on paper. More importantly, it is an early signal that the AI boom is starting to redraw what a public tech balance sheet can represent.