Anthropic’s IPO and the Future of AI: Three Insights from Smith Experts

As Anthropic prepares to go public under the ticker "ANT," three faculty members at the University of Maryland’s Robert H. Smith School of Business explore what the company's landmark IPO filing reveals about the future of AI. Anthropic has filed its S-1 prospectus after a period of explosive growth, reporting $16.3 billion in revenue during the first half of 2026 while committing more than $400 billion to the computing infrastructure needed to develop and run advanced AI models.

Below, Professor of Information Systems Siva Viswanathan analyzes the market race, Dean's Professor of Decisions, Operations & Information Technologie Balaji Padmanabhan explores governance and safety, and Associate Professor of Marketing Daniel McCarthy breaks down customer economics and unit profitability. 

McCarthy is known for popularizing customer-based corporate valuation, a "bottom-up" approach to valuing firms by assessing the value of their customers. Padmanabhan also serves as Associate Dean of Strategic Initiatives and directs Smith’s Center for Artificial Intelligence in Business, and Viswanathan is area chair of Decision, Operations and Information Technologies. 

Siva Viswanathan: You Don’t Have to Outrun the Bear: The Logic Behind Anthropic’s IPO

There’s an old joke about two hikers who spot a bear charging toward them. One calmly pulls running shoes from his pack and starts lacing them up. “You can’t outrun a bear,” his friend says. “I don’t have to outrun the bear,” he replies. “I just have to outrun you.”

The race among AI foundation model companies follows the same logic. At first glance, the market resembles cloud computing: enormous fixed costs, limited network effects, and enterprise customers that deliberately spread workloads across vendors. That combination typically produces a stable oligopoly. AI is different because the bear moves much faster. Switching model providers can take an afternoon, not the years often required to migrate cloud infrastructure. Frontier advantages last months rather than years, and open-weight models quickly commoditize capabilities that fall behind the state of the art. A cloud provider can survive a slowdown with a base of locked-in customers. A foundation model company cannot.

Even more important, AI contains a tipping point that cloud computing largely lacks. If a lab’s models begin meaningfully accelerating its own research, its lead could compound faster than competitors can respond, transforming today’s oligopoly into a winner-take-all market.

No AI lab can escape these pressures, but none can afford to be the slowest hiker. For Anthropic, as for OpenAI, an IPO is a much-needed pair of running shoes: the capital required to secure compute, attract talent, and stay ahead of the bear by staying ahead of one another.

Balaji Padmanabhan: Can Anthropic Balance AI Safety and Shareholder Expectations?

Anthropic’s mission is “the responsible development and maintenance of advanced AI for the long-term benefit of humanity.” For a typical public company, that commitment could create tension. Shareholders might question decisions that prioritize societal benefits over maximizing financial returns.

Anthropic, however, is not a typical corporation. It is structured as a Delaware Public Benefit Corporation (PBC), which requires directors to consider both shareholder interests and the company’s public-benefit mission. In theory, this gives management and the board greater latitude to prioritize safety and responsible AI development, even when doing so conflicts with short-term financial goals.

The company has also built additional safeguards into its governance. A majority of its board is selected by a Long-Term Benefit Trust (LTBT), an independent entity with no financial stake in Anthropic. The trust’s members are intended to bring expertise in AI safety, policy, national security, and social impact. Meanwhile, Anthropic’s founders retain substantial voting control through a special share structure, reinforcing a leadership team that has consistently emphasized AI safety.

What makes the story more interesting is that safety is not only a principle. It is also a strategy. Anthropic is positioning itself as a leading provider of AI for enterprises, a market where reliability and risk management matter greatly. Companies deploying AI products face significant financial and reputational consequences when systems fail. As a result, a reputation for safety can be a competitive advantage as well as a mission-driven commitment.

The real test comes if competitive pressures intensify. Would Anthropic delay or limit model releases on safety grounds if rivals move faster? Its PBC structure is designed to support exactly that kind of decision. Yet the company also faces substantial financial obligations. According to its filing, Anthropic has committed to more than $400 billion in non-cancelable cloud and compute agreements with providers such as Amazon and Google.

Those commitments create obvious pressure, but the picture is nuanced. Some of Anthropic’s largest infrastructure partners are also significant investors and may have incentives to support the company’s long-term success. Other counterparties, however, may be less flexible. The greater risk may not be default but the mounting cost of delays, where every safety-driven pause comes with a visible financial tradeoff.

For now, that scenario appears unlikely. Anthropic’s revenue has been growing at an extraordinary pace, and most observers do not expect the company to face an immediate conflict between mission and performance. Still, technology markets have a way of producing unexpected outcomes.

Critics have also challenged Anthropic’s messaging around safety. Some argue that companies seeking billions in investment while reserving the right to sacrifice profits for safety are asking investors to accept conflicting goals. Others contend that firms concerned about the pace of AI development should simply slow down rather than seek broader consensus.

My view is that Anthropic’s commitment to safety is genuine. The company has emphasized this mission since its founding, and several of its actions suggest the commitment goes beyond marketing. Just as important, abandoning that mission could trigger another cost: the loss of key employees whose dedication to responsible AI development is part of Anthropic’s competitive advantage. Taken together, these governance structures, cultural commitments, and talent incentives create stronger safeguards than those found in most technology companies. They do not eliminate the tension between mission and profit, but they make it more likely that Anthropic will remain aligned with its safety goals even when doing so becomes difficult.

Daniel McCarthy: What Anthropic’s S-1 Reveals About Its Unit Economics and Path to Profitability

While much attention on Anthropic's proposed public offering has focused on its extraordinary top-line revenue growth, the S-1 filing offers us a rare look into the underlying economics of the generative AI market.

What strikes me most is not just the sheer pace of growth, but what the data reveals about Anthropic's customer business. The company appears to have built a massive customer funnel, spanning roughly 40 million lower-spending users alongside thousands of enterprise accounts spending six and seven figures per year. Because a large share of revenue comes from these enterprise clients, sales growth is predominantly driven by deep business adoption rather than consumer subscriptions alone.

This customer structure also helps address valid concerns about customer concentration. Although a handful of large enterprise clients account for a significant portion of revenue, that concentration has declined over time even as spending from those same key accounts has reached new highs. What this indicates to me is that Anthropic has built a healthy progression path, establishing a deep bench of accounts steadily moving “up the ladder” into larger relationships.

Furthermore, we see exceptionally strong revenue expansion among existing enterprise clients. Anthropic’s net revenue retention levels far exceed what we typically consider strong performance in standard SaaS markets, showing that once businesses adopt the platform, their usage scales rapidly over time.

Beyond top-line metrics, the filing offers encouraging signals regarding the company's long-term unit economics. While Anthropic reported substantial net losses, much of that total stems from noncash accounting adjustments linked to the rising value of prior financing instruments. When you isolate those accounting effects from the core operating business, the picture looks very different.

The data suggests Anthropic can generate operating profit when compute utilization is high, while ongoing reductions in infrastructure costs continuously improve unit economics. With gross margins above 85% and recent model efficiencies significantly lowering the cost of running AI workloads, the foundation for long-term profitability is clearer than it might seem on the surface.

Overall, the filing paints a picture of explosive growth, deep enterprise adoption, a well-diversified customer pipeline, and improving operational economics. For public market investors, those structural signals may ultimately prove far more important than the headline noise.