OpenAI in Crisis: $122 Billion Raised, But Failures Persist

OpenAI in Crisis: $122 Billion Raised, but Failures Persist

With a record fundraising of $122 billion, OpenAI finds itself in a precarious situation. The closure of Sora, an ambitious but ultimately failed project, symbolizes the current challenges facing the company. Despite an impressive valuation of $852 billion, OpenAI must contend with notable failures and increasingly fierce competition. The launch of GPT-5 did not meet expectations, and the question of profitability remains unresolved.

Sora and GPT-5: Unfulfilled Promises

The video generator Sora, touted as the future TikTok of AI, had a promising start with 3.3 million downloads in November 2025. However, the initial excitement quickly waned, and users limited themselves to creating entertainment videos without real added value. The operational costs of Sora soared, with expenses reaching $15 million per day at launch, according to Forbes, and still $1 million daily in recent weeks. The billion-dollar partnership with Disney to use licensed characters never materialized.

The launch of GPT-5 in August 2025 was also disappointing. This model, expected to represent a significant advancement following the success of GPT-4, was criticized for its inferior performance. Preliminary versions tested by AI influencers had raised high expectations, but the final product did not deliver on its promises. Errors in performance graphs during the official presentation and the sudden removal of GPT-4o sparked user outrage. Sam Altman, the CEO, himself described this failure as a "mega flop."

Strategic Dispersal in the Face of Organized Competition

OpenAI has embarked on numerous simultaneous projects, such as Atlas, a proprietary browser, and a physical device designed with Jony Ive. This dispersal of resources contrasts with the more focused strategy of its competitors. Google, with its Gemini model, has managed to establish itself through integration into popular services like Gmail and Android. Even Apple opted for Google over OpenAI for its new Siri, in a billion-dollar annual deal.

On its part, Anthropic has chosen to concentrate on improving its Claude model, which has seen its monthly sign-ups triple since November 2025. Unlike ChatGPT, Claude does not incorporate advertisements, which was highlighted during advertising spots at the Super Bowl. Sam Altman's reaction on X has reinforced the image of a leader on the defensive.

The Pentagon Episode and Its Consequences

The Pentagon episode marked a turning point for OpenAI. While Anthropic refused to lift its ethical safeguards for autonomous weapons, OpenAI seized the opportunity to secure the contract. This decision led to a dramatic spike in ChatGPT uninstalls, increasing by 295% in a single weekend. Meanwhile, Claude took the lead in the U.S. App Store. Sam Altman had to publicly amend the terms of the contract, admitting it was an "opportunistic and hasty" agreement. Anthropic's annualized revenue reached $19 billion, compared to $24 billion for OpenAI, showing a rapidly narrowing gap.

OpenAI's Financial Challenge

Despite monthly revenues of $2 billion, OpenAI faces colossal expenses. The Stargate project, aimed at developing AI infrastructure, is estimated at $1.4 trillion. The record fundraising only covers part of this ambition. The high costs of AI models also affect other giants like Google and Anthropic, which are seeking solutions to monetize their investments.

OpenAI is attempting to refocus by recruiting the creator of OpenClaw and steering GPT-5.4 towards businesses. However, this strategy comes too late, as the competition has already gained an edge. The era of limitless spending seems to be coming to an end, and OpenAI must quickly find solutions to resolve its economic equation.