The three men most responsible for the current AI moment have never disliked each other more. The Economist's latest cover story frames the rivalry between Sam Altman, Dario Amodei, and Elon Musk as the defining commercial contest of the decade — one where billions in capital, deep ideological difference, and accumulated personal grievance are all in play simultaneously.
The backstory is dense. Musk and Altman co-founded OpenAI together before a public and bitter split. Amodei departed as VP of Research in 2021, taking a cohort of safety-focused researchers with him to start Anthropic. That departure looks different now than it did then: Claude has won serious enterprise customers, backed by tens of billions in commitments from Google and Amazon. OpenAI is mid-structural transformation — converting from a capped-profit entity to a full public benefit corporation — while still dominating consumer AI through ChatGPT and the Microsoft distribution engine behind it. xAI's Grok arrived last and youngest, but Musk brings something neither rival can easily replicate: a captive audience of hundreds of millions on X and, since January, a degree of political proximity to the White House that has yet to be fully cashed in.
The 'fighting dirty' characterisation is not just headline copy. Musk's ongoing lawsuit against OpenAI — alleging breach of its nonprofit founding mission — functions as a reputational grenade as much as a legal instrument. Talent poaching, non-compete litigation, and Musk's use of X as both product and attack platform have become routine. Anthropic, positioned as the safety-conscious alternative, has nonetheless matched its rivals on frontier capability; the whole logic of responsible scaling is that you forfeit the ability to set norms the moment you fall behind on capability.
The race is sharpest in agents — autonomous systems that act, book, execute, and integrate rather than simply respond to prompts. This is where enterprise revenue actually lives, and all three companies are pressing hardest here because agents represent the conversion of raw model capability into recurring software spend. OpenAI's Operator product, Anthropic's computer-use work, and xAI's enterprise push are bets on the same underlying thesis: own the agent layer, own the budget. Google DeepMind and Meta AI are running the same play, as are open-source alternatives that have closed the capability gap faster than most insiders anticipated.
What makes The Economist's timing notable is that the contest is tightening precisely as the decisions that matter most are being locked in. Which platforms enterprise developers standardise on, which infrastructure governments quietly come to depend on — those choices are being made in procurement conversations happening right now, and they will be far harder to reverse than anything that has come before. The next twelve months of contract wins and developer adoption numbers are the scoreboard that counts.