Anthropic's $350bn IPO. The real question is what investors think principles are worth
Anthropic has spent its short life cultivating an image of being the grown-up in the room. Five years after its founding, the company behind Claude is preparing for an initial public offering that could value it at 350 billion dollars. In a sector defined by scale, swagger and breakneck product cycles, Anthropic is presenting something more austere: restraint, safety and a focus on core model building.
That alone sets it apart from OpenAI, its closest rival, whose valuation of 500 billion dollars rests on 2028 revenue projections that mirror Anthropic’s own. Both businesses are racing to define the next generation of AI infrastructure. Only one of them is investing in an ecosystem that includes data centres, hardware ventures, a browser and minority stakes in other companies. Anthropic is positioning itself as the opposite, a company that makes models and little else.
The valuation history of the last generation of tech giants is the backdrop. Google went public six years after launch at about 23 billion dollars. Facebook needed eight years to reach a 100 billion dollar float. Microsoft waited 11 years for an 800 million dollar debut in 1986. Anthropic hitting a potential 350 billion dollars in five years is a vivid sign of the acceleration inside AI.
The business does generate revenue. Reports cited in the FT article note a projection of 70 billion dollars in sales by 2028. At five times that figure, Anthropic’s mooted valuation lands in the neighbourhood of Meta, Alibaba and Palantir at their respective IPO multiples. The comparison that will dominate, however, is not with legacy tech but with OpenAI. With Amazon, Google, Microsoft and Nvidia already on its cap table, Anthropic’s investor base already mirrors the power core of the sector.
Anthropic also has traction among enterprises. Menlo Ventures, an investor, estimates Claude’s share of the enterprise AI market at 32 per cent. Businesses pay, and consistency matters. Anthropic’s pitch is that its models are reliable, predictable and principled. The company was founded to build systems that are helpful, honest and harmless. External assessments cited in the FT note back that up, ranking Claude as the least likely among major models to lie or provide dangerous expert-level guidance.
Whether the market rewards that discipline or discounts it remains unknown. By the time Anthropic lists, the AI landscape could be transformed again. Investors will need to value not only projected sales but the principle that has become the company’s signature asset.
