There are decisions that at the time they are made look like corporate suicides. Then the market responds, and suddenly those same decisions seem like acts of strategic genius. This is exactly what is happening at Anthropic these days, and we at Blockworld anticipated it when we analysed the clash between the Pentagon and Anthropic: what might have seemed like a shot in the arm at the time is turning out to be one of the most effective brand moves in recent Silicon Valley history.
To recap: the Pentagon had asked Anthropic to remove the ethical constraints preventing Claude from being used for mass surveillance and fully autonomous weapon systems. CEO Dario Amodei responded with an unequivocal public statement:
we cannot in good conscience accept their request
and held the line. Defence Secretary Pete Hegseth labelled Anthropic a ‘risk to the national security supply chain’, President Donald Trump ordered all federal agencies to immediately cease using Anthropic technology, and the company found itself officially banned from the US government.
On paper, a catastrophe. In reality, an extraordinary catalyst.
The story has already unfolded in the 24 hours since our piece was published. According to yesterday’s BBC report, OpenAI announced changes to the contract it had signed on Friday with the Pentagon, the so-called ‘Department of War’ as the Trump administration has renamed it, after the deal had triggered an avalanche of criticism from the public and even internal employees. CEO Sam Altman returned to X admitting that the announcement had been handled in an ‘opportunistic and perfunctory’ manner, and added new explicit constraints: OpenAI systems cannot be used for mass surveillance of US citizens, and intelligence agencies such as the NSA cannot access the technology without a separate amendment to the contract.
The further twist, also reported by the BBC, is the Sensor Tower data: since the day OpenAI announced its agreement with the Pentagon, ChatGPT’s uninstallation rate has increased by 200% on a daily average. In other words, while Anthropic was being banned by Trump and Claude’s servers were crashing due to too much demand, the competitor that had chosen the opposite path, sign first, fix later, was suffering exactly the reputational damage that Anthropic had avoided by refusing to bend.
The overall reading therefore becomes even sharper than we wrote yesterday: Anthropic held the line on values, lost the government contract, gained the trust of users. OpenAI signed quickly, took the criticism, and now runs for cover by renegotiating. Two opposing strategies, and the market, with its brutal and immediate verdict, is already saying which one worked better.
The response from users was immediate and unequivocal. Claude, which at the end of January was just outside the top 100 in the US App Store, climbed the charts with impressive speed: sixth on Wednesday, fourth on Thursday, first on Saturday. On 1 March 2026, Claude knocked ChatGPT off the throne that OpenAI had held for years, leaving it in second place and Google Gemini in fourth. Anthropic confirmed that daily subscriptions have broken all historical records every single day of the week, free users have grown by more than 60 per cent since January, and paying subscribers have more than doubled since the beginning of the year. Demand has been so explosive that on 2 March Claude’s servers crashed due to ‘unprecedented demand’, the kind of problem any tech company would welcome.
There is one detail that makes the picture even more complex: despite the official ban, CBS News reported that Claude continues to be used in military operations related to the US-Israel-Iran conflict. The Pentagon declined to comment. This means that in practice the ‘ban’ is more symbolic than operational, but symbolism does count, and Anthropic has won on that very ground.
Those who thought that refusing to build autonomous weapons was a luxury a tech company cannot afford have just received a brutal market lesson: sometimes luxury is not having principles, not having them. Dario Amodei bet on reputation at a time when it seemed an unsustainable gamble. Today Claude is first on the App Store, the servers are crashing from too much demand, and ChatGPT looks down on him… but from second place.
