SpaceX boss Elon Musk’s Grok is losing steam while SpaceX rents vital AI compute to one of its faster rivals.
The deal, signed in early May, gives Anthropic, maker of Claude, all the computing capacity at one of Musk’s main data centres. Anthropic and rival OpenAI have been grabbing every scrap of compute they can find as demand for their models goes bonkers.
Grok has reached millions through its integration with Musk’s social network X and a few grubby attention-seeking features. Those included a sexualised AI companion, because apparently nothing says “truth-seeking” like chatbot thirst traps.
Fresh data suggests the growth story has gone a bit flat. AppMagic said Grok downloads fell to about 8.3 million in April, down from more than 20 million in January.
Research firm Recon Analytics found that paid Grok use barely moved in its survey of more than 260,000 US consumers and workers who use AI. The share saying they paid for Grok was 0.174 per cent in the second quarter of 2026, compared with 0.173 per cent a year earlier.
More than six per cent said they paid for ChatGPT, which is a rather awkward gap for a product that was supposed to shake up the AI priesthood.
Los Angeles engineer and tech investor Ben Pouladian told the Wall Street journal: “OpenAI is Coke, Anthropic is Pepsi and Grok is RC Cola. I never really saw people drinking it.”
Pouladian is hardly allergic to Muskware. He drives a Tesla and is active on X, but after downloading Grok in late 2023 and poking around, he never became a serious user. He said he prefers Anthropic’s Claude, OpenAI’s ChatGPT and sometimes Google’s Gemini.
In court in late April for his suit against OpenAI, Musk played down the size and importance of xAI, the AI company he recently merged into SpaceX. He described it as “pretty small… the smallest of the AI companies.”
OpenAI’s launch of ChatGPT in 2022 was many punters’ first proper brush with AI. By mid-2025, more than three-quarters of respondents to Recon Analytics had heard of ChatGPT.
Musk released Grok in late 2023. He said it would become the most popular AI in the world, while being “maximally truth-seeking” and less “woke” than its rivals. Musk spent much of the summer of 2025 camped at his AI startup, trying to catch up in the arms race. He personally oversaw the design of a racy chatbot.
Grok offered settings that let users create suggestive and sexualised content, which former employees said boosted engagement. That might be a metric win, if your metric is how quickly regulators start reading your terms of service.
The January peak in Grok downloads followed an update that let users virtually undress people in photos. Widespread use of the feature on images of minors drew scrutiny from regulators and lawmakers, and the company limited access.
The hottest fight among the big labs is now coding assistants, where corporate adoption is driving rapid revenue growth.
Enterprise Technology Research chief strategist and research director Erik Bradley said Grok is barely growing inside enterprise organisations, while the use of Claude and Gemini is soaring.
In an ETR survey of about 500 people, 48 per cent of respondents in March said their company was using Claude and planned to continue using it. That was up from 21 per cent the prior year. Forty per cent said their company was using and planned to keep using Gemini, up from 27 per cent a year earlier. Grok managed seven per cent, up from four per cent.
Musk faces pressure to prove his companies can make money before SpaceX’s expected initial public offering this year. Analysts said the Anthropic compute deal at the Colossus 1 data centre near Memphis, Tennessee, could bring in a few billion dollars a year.
Research firm IDC vice-president of software development Arnal Dayaratna said the deal shows Musk is turning Colossus into an external computing platform for major AI companies. That is a shift from using the facility only for internal model development.
Musk’s willingness to deal with Anthropic marks a shift in his stance towards the company. In February, Musk described its AI as “misanthropic and evil” in a post on X.
Pouladian said Musk’s new warmth towards Anthropic could come from its hostility to OpenAI, which Musk is fighting in court. “The enemy of my enemy is my friend, and it’s also my compute partner,” he said.
