TradingKey - On August 9, Eastern Time, Elon Musk expressed strong support on social media platform X for a prediction by network infrastructure giant Cloudflare ( NET ). According to the projected data, internet traffic generated by AI bots will surpass that of humans for the first time in May 2026, a full 18 months ahead of the original forecast of late 2027.
Cloudflare projects that within five years, bot traffic will reach 1,000 times that of humans, making human internet activity almost akin to a 'rounding error'.
Musk wrote on X: 'AI agent internet traffic is obviously going to dwarf human usage, and there is zero doubt about that.' He added that the only system capable of supporting the ultra-high bandwidth growth required by AI is Starlink.
[Source: X]
In contrast to Musk's high-profile optimism is Michael Burry, the 'Big Short' famous for accurately predicting the 2008 subprime mortgage crisis. Burry is betting against the AI bubble with unprecedented intensity.
According to Burry's portfolio activity disclosed on Substack, he is maintaining short positions on Nvidia ( NVDA ), Micron Technology ( MU ), Palantir ( PLTR ), Tesla ( TSLA ), Caterpillar ( CAT ), Applied Materials ( AMAT ), and the iShares Semiconductor ETF ( SOXX ) . Among these seven short bets, all except Nvidia are currently profitable.
Last week, Burry added two new short targets. He shorted Oracle ( ORCL) at $144.63, while also shorting Nebius Group ( NBIS) at $211.77. The latter has surged as much as 244% over the past year. Burry stated that because the implied volatility of Nebius's put options exceeded 100% and was overpriced, he chose to short the stock directly.
[Source: Substack]
Burry's assessment is that the current massive demand for AI chip giants is not driven by actual end customers, but rather circularly driven by opaque off-balance-sheet financing arrangements. He warned that a vast number of cloud computing and AI infrastructure companies are saddled with massive long-term lease obligations, and once the supply-demand cycle reverses, these fixed costs will crush them.
In addition, he pointed out that AI infrastructure companies may be under-depreciating their GPUs, making book profits appear higher than they actually are. Taking Oracle as an example, Burry estimates that the company's earnings could be overstated by approximately 26% to 27% by 2028.
Burry also warned that after the S&P 500 Index hit a record high, the market might be approaching a 'major top', facing the risk of a sharp correction similar to that of 1987.
On one side, Musk strongly supports the view that AI traffic will vastly exceed human traffic, while on the other side, Burry continues to increase his short positions on AI leaders. Musk's 'zero doubt' clashes head-on with Burry's shorting actions, and this long-short battle over the future of AI is still some way from being resolved.
Find out more
Comments