Despite its initial hype upon release, the product, which Musk personally oversaw throughout last summer, has fallen out of favor.

Elon Musk’s Grok artificial intelligence model is being left far behind by formidable competitors in the fierce growth race. Notably, the agreement signed by parent company SpaceX to lease massive computing power to Anthropic is raising doubts about Grok’s ability to catch up.

The agreement, signed in early May, will allow AI model maker Claude and its namesake chatbot to utilize the full computing power at one of Musk’s key data centers. Amidst surging demand challenging operational capabilities, both Anthropic and its rival OpenAI are aggressively acquiring every available computing resource.

Since its launch two years ago, Grok has reached millions of users thanks to its direct integration with Musk’s X social network, along with controversial features such as a sexually suggestive “AI companion.” However, the latest data suggests that the tool’s growth momentum appears to have reached saturation.

According to analytics firm AppMagic, Grok’s downloads fell to 8.3 million in April, a sharp decline from its peak of over 20 million in January.

Data from research firm Recon Analytics, based on a survey of over 260,000 consumers and employees using AI in the US, paints a similar picture. The percentage of respondents who paid for Grok remained stagnant at 0.174% in the second quarter of 2026, compared to 0.173% a year earlier. For comparison, over 6% of survey participants said they were paying for ChatGPT.

“RC Cola” of the tech world

“If OpenAI is Coca-Cola, and Anthropic is Pepsi, then Grok is just like RC Cola,” said Ben Pouladian, a Los Angeles-based engineer and tech investor. “I’ve never really seen people use it widely.”

Pouladian himself is also influenced by Musk’s ecosystem: he drives a Tesla and is actively involved in the X. When Grok was released in late 2023, he downloaded and tried it, but never became a “loyal user.” He shared that he prefers Anthropic’s Claude, OpenAI’s ChatGPT, and occasionally Google’s Gemini.

Neither Musk nor Grok’s parent company, SpaceX, have commented on the matter. However, in public statements, Musk himself has described Grok as somewhat weaker in the AI ​​race.

At the court hearing related to the OpenAI lawsuit in late April, Musk downplayed the size and importance of xAI – the AI ​​company he recently merged with SpaceX. He described it as “quite small,” “very small,” and “the smallest AI company” among the big names.

Controversial strategy and business challenges

The launch of ChatGPT in 2022 paved the way for AI to reach the mass consumer. By mid-2025, more than three-quarters of respondents in Recon Analytics’ survey were aware of ChatGPT.

When releasing Grok in late 2023, Musk aimed to make it the world’s most widely used AI , promising “maximum truth-seeking” and less “woke” (excessive radicalism) than its competitors.

For much of the summer of 2025, Musk spent time at his AI startup striving to catch up in the technological arms race. He directly oversaw the design of a sexually suggestive chatbot. Grok also provided settings that allowed users to create suggestive and sexually explicit content – ​​which former employees say was intended to boost engagement rates.

The peak in downloads in January followed an update that allowed users to virtually “undress” people in photos. The misuse of this feature on images of minors came under intense scrutiny from regulators and lawmakers, forcing the company to restrict access.

Currently, the most intense battleground among major AI labs is programming support tools, where enterprise adoption is driving rapid revenue growth.

But even in this area, Grok is lagging behind. Erik Bradley, chief strategist at market research firm Enterprise Technology Research (ETR), says Grok has barely seen growth in the enterprise sector. In contrast, the use of Claude and Gemini is skyrocketing.

A March survey of 500 people by ETR showed that 48% of respondents confirmed their company was using and planned to continue using Claude (up from 21% the previous year). 40% said they were using Gemini (up from 27%). Meanwhile, only 7% said their company was using Grok (a slight increase from 4%).

Shifting focus to infrastructure services

Musk is facing pressure to demonstrate the profitability of his companies ahead of SpaceX’s planned initial public offering (IPO) this year. Analysts believe the deal for Anthropic to lease computing power at the Colossus 1 data center near Memphis could bring Musk several billion dollars annually.

Arnal Dayaratna, Vice President of Software Development at IDC, believes the deal shows Musk is beginning to turn Colossus into an outsourced computing platform for large AI companies, rather than just using it as a base for internal model development.

However, Guillermo Rauch, CEO of Vercel, cautioned against prematurely ruling Musk out of the race. He is optimistic that Musk’s recent restructuring of his AI unit will enhance its competitiveness.

“Once Elon focuses – and that’s what’s happening – we often see him perform very well,” Rauch said. He also noted that customer behavior suggests developers often switch between models very quickly. Engineers will likely flock to Grok if the company releases a new model with superior performance.

Musk’s decision to partner with Anthropic marks a major shift in his attitude. Just back in February, in an X-listing post, Musk had described the company’s AI as “hateful and toxic.”

Investor Pouladian suggests this new partnership may stem from Anthropic’s rivalry with OpenAI – the opponent Musk is battling in court: “The enemy of my enemy is my friend, and in this case, that includes my computing infrastructure partner.”