The Birth of Anthropic: Safety as a Competitive Advantage

Anthropic was founded in 2021 by siblings Dario and Daniela Amodei after they left OpenAI. Dario Amodei, former vice president of research at OpenAI, led the departure of seven key colleagues from San Francisco. The reason was concerns about too rapid commercialization of AI without sufficient safety safeguards.

Unlike OpenAI, which is heading towards achieving artificial general intelligence (AGI) as quickly as possible, Anthropic built its strategy on the concept of Constitutional AI — a method that is supposed to ensure that models behave predictably, are interpretable, and produce a minimum of hallucinations. This safety approach became a key factor in Anthropic's growth in the enterprise software market, where companies in finance, law, and healthcare require reliability and regulatory compliance.

Comparison of Valuations and Revenues

In April 2026, Anthropic's valuation on secondary markets reached $688 billion, with primary investors offering even $800 billion. This is double compared to the value from the January Series G financing round ($380 billion). OpenAI is currently valued by investors at approximately $852 billion in the primary market and roughly $765 billion on the secondary market, meaning that the difference between the two companies has dramatically narrowed.

However, the revenue comparison is more interesting. In April 2026, Anthropic announced annual revenues of $30 billion, which is more than triple compared to $9 billion at the end of 2025. OpenAI reports revenues around $25 billion. It should be noted, however, that both companies use different accounting methods: Anthropic reports gross revenues (the full amount invoiced to customers), while OpenAI counts only its share. When compared using the same methodology, OpenAI would likely still be ahead.

Nevertheless, the trend is clear: Anthropic's share of the enterprise AI market grew from 24.4% to 30.6% in March 2026, while OpenAI's share fell from approximately 46% to 35.2%. The difference between them thus narrowed from 21.6 to just 4.6 percentage points. According to data from the Ramp platform, Anthropic currently even leads in markets with the highest AI penetration — software, finance, and professional services. In direct competition for new enterprise customers, Anthropic wins in 70% of cases.

The Road to the Stock Exchange: Who Will Be First?

Anthropic officially announced preparations for an IPO in October 2026 with the goal of raising more than $60 billion, which would make it the second largest primary stock offering in history after SpaceX. The company is intensively working on structural changes: former Microsoft CFO Chris Liddell, who led the successful $23 billion General Motors IPO, joined the board of directors, and an employee stock offering worth $5–6 billion is underway. Goldman Sachs, JPMorgan, and Morgan Stanley are preparing as the main underwriting banks.

OpenAI, on the other hand, faces internal contradictions. While CEO Sam Altman would like to carry out the IPO as early as 2026, CFO Sarah Friar publicly stated that "process and organizational work is not yet complete". In August 2025, Friar also stopped reporting directly to Altman — which is an extremely unusual arrangement in American corporations and significantly slowed down preparations for entering the stock exchange. For investors, Anthropic's path to public trading is therefore significantly clearer.

Technological Differences: Constitutional AI vs. AGI

The philosophical split between the two companies is directly reflected in their products. Claude, Anthropic's flagship model, is designed to be "honest, harmless, and helpful". This manifests in a lower rate of hallucinations and higher controllability of outputs, which is critical for legal analysis, financial reporting, or medical documentation.

OpenAI is heading faster towards general intelligence and aggressively commercializing its models. This approach brings faster innovation, but also higher risk of unpredictable system behavior. For ordinary users, this means that Claude may be a safer choice for sensitive tasks, while GPT models often offer a wider range of functions and faster integration of new capabilities.

Computing Power and Regulatory Challenges

Both giants face enormous costs for computing infrastructure. OpenAI expects to spend $121 billion on AI computing by 2028. Anthropic has signed agreements with Google and Broadcom to secure approximately 3.5 GW of TPU computing power from 2027 and plans to invest $50 billion in its own data centers in Texas and New York. In March 2026, Claude experienced five outages caused by a lack of computing capacity, showing how critical this area is.

Regulations represent another uncertainty. In February 2026, the US Department of Defense placed Anthropic on a list of risky suppliers, which prohibits federal suppliers from using its models. The company filed a lawsuit and in March obtained a preliminary injunction from a California court, but the appellate court suspended the decision in April. With oral arguments scheduled for May 19, 2026, the risk of losing lucrative government contracts hangs over Anthropic like a sword of Damocles. OpenAI, on the other hand, signed a contract with the Pentagon to use models on classified networks.

What Does This Mean for the Czech Republic and European Companies?

For Czech users and companies, this battle is relevant for several reasons. Both Claude and ChatGPT are available in the Czech Republic, with Claude offering a fully functional interface in Czech, although it primarily focuses on the English market. Both companies offer API with pay-per-use pricing, with prices per million tokens ranging from single to tens of dollars depending on the chosen model. Czech companies using APIs from Anthropic or OpenAI should monitor developments in enterprise security and compliance — especially with regard to the EU AI Act, which places high demands on the transparency and risk level of AI systems.

Anthropic's growth in the enterprise market can mean better negotiating position for Czech companies when choosing an AI tool supplier. With increasing competition, pressure on API prices and better conditions for enterprise customers can be expected. At the same time, it is likely that both companies will have to better adapt their services to European regulatory requirements, which could lead to improved personal data protection guarantees even for Czech users.

Investors should monitor three key variables: the ability to manage computing costs without degrading service quality, retaining enterprise customers, and successful navigation of the regulatory environment at the intersection of national security and commercial interests. The situation is analyzed in more detail by the TradingKey server.