Elon Musk's artificial intelligence (AI) ambitions have taken a sharp turn: unable to beat Anthropic in the race for AI dominance, he's now profiting from their success. After Grok's monthly downloads collapsed from over 20 million in January to 8.3 million in April, Musk made a strategic move last week, leasing his full Memphis Colossus 1 supercomputer to Anthropic, the AI company he publicly criticized just months earlier.

This pivot reveals a fundamental shift in how Musk is approaching the competitive AI landscape. Rather than doubling down on xAI's struggling chatbot, he's monetizing the computing infrastructure he built to power it. The move highlights a reality that even Musk's most optimistic supporters must now confront: Grok is losing the AI race, and the economics of cloud computing may be more valuable than winning it.

Why Is Grok Falling So Far Behind?

The numbers tell a stark story. Grok posted a 14.43% month-over-month web traffic decline in April, marking the worst performance of any major AI site, while Anthropic's Claude achieved 34.18% growth during the same period. Recent survey data shows an even wider gap in enterprise adoption: 48% of companies plan to use Claude compared to just 7% for Grok.

The decline accelerated after Grok's January peak, which followed a controversial update that allowed users to virtually undress people in photos. The feature drew immediate regulatory scrutiny over potential misuse involving minors, forcing xAI to restrict access and damaging the product's reputation. Prediction markets have rendered a harsh verdict: Polymarket gives Anthropic a 67% probability of holding the top AI model position by the end of June, while xAI sits at just 2%.

"OpenAI is Coke, Anthropic is Pepsi and Grok is RC Cola. I never really saw people drinking it," said Ben Pouladian, a tech investor, to the Wall Street Journal.

Ben Pouladian, Tech Investor

Musk himself acknowledged the competitive gap in court last month, calling xAI "the smallest of the AI companies". The admission underscores how far behind Grok has fallen in a market increasingly dominated by OpenAI and Anthropic.

How Is Musk Turning Compute Into Revenue?

Rather than accept defeat, Musk identified an opportunity in the infrastructure itself. Anthropic faced a critical constraint: its growing revenue potential was capped by insufficient computing power to train and run its AI models. Grok, meanwhile, generated less than $1 billion in revenue despite the massive investment in the Memphis Colossus 1 supercomputer.

The solution was elegant: Musk leased the entire Memphis facility to Anthropic, transforming a sunk cost into a revenue stream. The All-In podcast dubbed this new venture "Elon Web Services," and Altimeter Capital's Brad Gerstner estimated the opportunity conservatively.

• Revenue Potential: Gerstler estimated the compute leasing business could add $4 to $5 billion in annual revenue to Musk's portfolio
• Valuation Multiple: At a 40 to 50 times revenue multiple, the cloud compute business could significantly boost the valuation of xAI or SpaceX
• Strategic Advantage: The deal solves Anthropic's compute bottleneck while monetizing infrastructure that was underutilized by Grok

This arrangement represents a pragmatic acknowledgment that Musk cannot win the AI model race through Grok alone. By pivoting to infrastructure, he's following a playbook that has worked in other tech sectors: if you can't dominate the application layer, profit from the platform layer.

What Changed Musk's Mind About Anthropic?

The partnership is particularly striking given Musk's public hostility toward Anthropic just months earlier. In February, he called the company "misanthropic and evil" on X, reflecting his frustration with their success and his own product's struggles. The rhetoric was sharp and personal, suggesting an unbridgeable divide.

Yet after spending time with Anthropic's team last week, Musk's tone shifted dramatically. He wrote that "no one set off my evil detector," though he reserved the right to "reclaim the compute" if Anthropic's AI "engages in actions that harm humanity". The reversal illustrates how business pragmatism can override public feuds when financial incentives align.

"The enemy of my enemy is my friend, and it's also my compute partner," noted Ben Pouladian, referencing Musk's ongoing federal lawsuit against Sam Altman of OpenAI.

Ben Pouladian, Tech Investor

The "enemy" in this case is Sam Altman, whom Musk is currently suing in federal court over OpenAI's governance and direction. By powering Anthropic's bid to outcompete OpenAI, Musk is indirectly supporting the company best positioned to challenge his rival. It's a calculated move that transforms a competitive loss into a strategic advantage against a more pressing adversary.

What Does This Mean for Musk's AI Ambitions?

The compute leasing deal doesn't eliminate xAI or Grok; it reframes them within a broader portfolio strategy. Musk still controls the infrastructure, the talent, and the potential to improve Grok's product. However, the arrangement signals that he's hedging his bets on the AI model race itself.

Some observers believe Musk shouldn't be counted out entirely. Vercel CEO Guillermo Rauch warned against underestimating Musk's ability to execute at scale, stating that "once Elon focuses, we see him perform very very well". The implication is that Musk's current pivot to compute monetization may be temporary, a way to generate revenue while he rebuilds Grok's product and market position.

Guillermo Rauch

For now, however, the numbers suggest a clear hierarchy: Anthropic's Claude is winning the AI model competition, and Musk is profiting from that reality rather than fighting it. The Memphis Colossus 1 supercomputer, built to power Grok's ascent, is now fueling Anthropic's dominance. It's a humbling outcome for Musk, but a financially rational one.