Michael Burry, the acclaimed investor renowned for foreseeing the 2008 housing collapse, is now casting a critical eye on artificial intelligence, one of the most celebrated themes in today’s market.
Burry recently removed his hedge fund firm, Scion Asset Management, from regular regulatory reporting by deregistering it. Despite this move, he remains actively engaged in investing, identifying what he perceives as significant market mispricing.
Phil Clifton, a former associate portfolio manager at Scion, plays a central role in informing this perspective. Clifton's analysis, which questions the financial viability of AI's expansive infrastructure, underlies Burry's skeptical stance. Although the adoption of generative AI is rapidly increasing, Clifton contends that the economic benefits do not yet justify the substantial costs involved.
In a farewell letter to Scion investors at the end of October, Burry hailed Clifton as “the most prodigious thinker” he has known. CNBC acquired several of Clifton’s research notes, predating the inception of his own firm, Pomerium Capital, that articulate Scion's bearish outlook on AI.
Clifton argues that the economic expectations from this technology are overstated. He writes, “Just because a technology is good for society or revolutionizes the world doesn't mean that it's a good business proposition.”
AI usage seems widespread, with over 60% of U.S. adults interacting with AI regularly, according to Pew Research Center. However, Clifton points out that the economic impacts of this demand are unexpectedly minor. Although OpenAI is expected to generate over $20 billion in annual revenue, this is relatively minuscule against the backdrop of an AI infrastructure spending surge. Capital expenditures from hyperscalers have multiplied fourfold to nearly $400 billion annually, with predictions that this will reach $3 trillion in the next five years, as per Man Group.
“We assume other generative AI services in aggregate are insufficient to justify the sums being spent on infrastructure,” Clifton stated.
Scion draws parallels to the early 2000s telecom boom, where investment significantly surpassed usage, leading to a plummet in capacity utilization and a drastic drop in telecom pricing. Clifton sees cloud giants in a similar scenario now, ramping up AI infrastructure hoping that future demand will balance the scale. If widespread AI adoption lags, the massive data center investments could become economically unviable.
Signs of hesitation are starting to appear among big tech firms, as noted by Clifton. Microsoft has halted data center projects meant to use 2 gigawatts of electricity across the U.S. and Europe due to overcapacity. Alibaba's chairman has also cautioned about a potential AI infrastructure bubble.
Nvidia, a leading beneficiary of AI-related spending, has witnessed a stock surge fueled by record GPU orders from cloud entities. However, Scion questions these customers' ability to achieve financial success from these investments.