Tech • AI • Robotics • Game
Why Is Elon Musk Bursting the AI Bubble?
9.3/10TL;DR
Elon Musk’s warnings about neocloud financing, combined with tighter credit, market volatility and power constraints, have intensified concerns that the AI infrastructure boom is entering a critical stress test.
KEY POINTS
Musk targets neocloud funding
In an October 25 post, Elon Musk argued that more than 200 neoclouds could fail because debt financing has dried up. His central point was that companies relying mainly on future compute-rental contracts have little chance of raising fresh capital, directly challenging the business models of firms such as CoreWeave and Nebius.
Broader bubble warnings spread
Concern is no longer limited to one entrepreneur. Ken Griffin has warned of bubble conditions, while Ben Horowitz has said Andreessen Horowitz avoids neocloud investments because of the risk. Nvidia chief Jensen Huang has also indicated he would not rescue such companies in a default scenario, reinforcing fears across Wall Street.
Market selloff hits AI-linked assets
AI-exposed stocks fell sharply as the anxiety spread. On November 5, Nebius dropped 13% and CoreWeave fell 8%. The same session saw the Nasdaq 100 lose 6.4%, its worst day since 2020, while Alphabet sank 12.6% after missing expectations and South Korea’s Kospi plunged more than 8% intraday before closing down about 6%.
Gas turbines become a strategic choke point
A second Musk message focused on SoftBank and alleged defaults tied to gas turbines, which have become crucial for rapidly powering new data centers. Developers using the standard grid can face waits of up to two years, while turbine-based setups can be deployed much faster. At manufacturers such as Mitsubishi Power, some deliveries are already being discussed for 2031 and later, underscoring how constrained this market has become.
SoftBank pressure raises questions for OpenAI
SoftBank, a major backer of OpenAI and lead financier for the planned $500 billion Stargate data center effort, is under intense scrutiny. Its shares are described as down 44% in a month, reviving memories of founder Masayoshi Son’s huge losses after the dot-com crash. If SoftBank’s balance sheet weakens further, financing for OpenAI-linked infrastructure could become harder.
Capital is concentrating in hyperscalers
The report describes a funding environment in which Google, SpaceX and other hyperscalers have absorbed enormous pools of capital, leaving less for smaller AI infrastructure players. Google reportedly raised $28 billion in debt, holds about 4% of SpaceX and 8% of Anthropic, and recently bought a Wyoming data center site from Crusoe, signaling that well-funded giants may acquire distressed assets rather than build everything organically.
OpenAI and Anthropic face financing strains
OpenAI reportedly sought $30 billion, then $20 billion, and ultimately secured $11 billion in a first tranche, with uncertainty around further funding. Anthropic’s expected IPO valuation has also reportedly been cut from $2 trillion talk to around $1.1 trillion-$1.3 trillion. At the same time, Anthropic’s quarterly growth was said to have slowed from about 144% to 47%, raising concerns about deceleration.
Politics and power access complicate expansion
Data center growth is also being shaped by politics and permitting. JPMorgan has warned that up to 12 gigawatts of computing capacity could be at risk amid local opposition. Projects tied to Texas, Ohio, Georgia and New Mexico face greater uncertainty, while Musk’s own Colossus facilities are located in states seen as more supportive of rapid development.
Large customers explore self-reliance
Some major AI customers are no longer relying entirely on external model providers. Revolut, valued at $115 billion with 80 million customers, has begun using open models internally for part of its workload. Latham & Watkins, which generated $8.3 billion in revenue in 2025, reportedly bought Nvidia GPUs and trained models on its own data, suggesting that premium enterprise clients may increasingly build in-house capabilities.
Demand remains real despite the stress
Even with the selloff, the need for compute has not disappeared. Google Cloud was described as generating $30 billion in quarterly revenue, up 103% year on year, with $11 billion in operating income. The central debate is no longer whether AI demand exists, but whether current valuations, funding structures and infrastructure timelines can survive a period of tighter capital and political friction.
CONCLUSION
The AI buildout is facing a severe test as financing, electricity supply and customer behavior all come under pressure at once. Whether this becomes a full bubble burst or a temporary shakeout may depend on how much capital the biggest tech groups are willing to keep deploying.
Ask a question