Tech Veteran Dan Niles Sees 30% AI Stock Correction in 2027

Veteran technology investor Dan Niles is forecasting a major downturn for the artificial intelligence sector, predicting that AI-related stocks could fall by 30% to 50% beginning in early 2027. The founder of Niles Investment Management draws a parallel between the current market and the 1997-1998 period, suggesting the AI boom is in its final stages but not yet at its peak.

"I think, these stocks from wherever they are at that point in time, could be down 30% to 50%," Niles said in a May 11 interview on The Master Investor Podcast, referencing the potential peak in early 2027. "Be Nimble. Have strong opinions, but loosely held."

The bearish forecast is supported by a recent J.P. Morgan survey of 56 global institutional investors, in which 54% said they expect a stock market correction of over 30% during 2026 or 2027. A full 45% of those respondents specifically pointed to 2027 as the most likely year for the pullback, aligning directly with Niles's timeline.

At stake is the dizzying rally that has propelled companies like Nvidia, which holds a 92% share of the data center GPU market. Niles warns that the very architecture of next-generation AI could shift the hardware landscape, and his advice for investors is to begin preparing for the cycle's end by holding more cash.

Agentic AI and the High-Base Effect

The core of Niles's short-term optimism and long-term concern is "Agentic AI." Unlike conversational AI, which responds to single queries, agentic systems can perform complex, multi-step tasks autonomously. Niles argues this requires 10 to 100 times more computing resources, or tokens, driving a massive, final build-out phase for AI infrastructure that will continue to fuel the market through 2026.

However, this explosive growth creates a precarious "high-base effect." The boom began in earnest with the launch of more advanced agentic platforms in early 2026. By early 2027, year-over-year growth comparisons will become exceedingly difficult to beat, leading to a sharp deceleration that he believes will trigger the market correction.

A Pivot to CPUs Threatens GPU Dominance

A key part of Niles's thesis is a structural shift in hardware demand. While training large models is a repetitive task well-suited for GPUs, agentic AI requires a processor to orchestrate and coordinate multiple applications simultaneously—a traditional strength of CPUs. He predicts the typical ratio of eight GPUs to one CPU in a server could move closer to 1-to-1.

This shift stands to benefit long-time CPU makers like Intel and AMD at the expense of GPU titan Nvidia. "On the margin, it will affect the stock performance," Niles said, suggesting that even as the overall pie grows, Nvidia's near-monopoly on AI excitement could face headwinds. This comes as Nvidia's own CEO, Jensen Huang, has stated the company has visibility on over $1 trillion in revenue for its upcoming Blackwell and Rubin platforms through 2027.

OpenAI and IPOs Create Structural Headwinds

Beyond the high-base effect, Niles points to two structural risks. The first is the financial stability of AI leaders like OpenAI, which along with Anthropic, accounts for about half of the cloud infrastructure order backlog. He questions OpenAI's ability to fund its massive capital expenditure commitments, which he estimates at $1.4 trillion over eight years against revenues of just $24 billion. "If OpenAI has a problem, that could accelerate the process dramatically," he warned.

The second risk is a coming liquidity crunch from a wave of mega-IPOs, including potential offerings from SpaceX and Anthropic. With valuations likely exceeding $1 trillion each, fund managers will be forced to sell existing holdings to make room, pulling capital out of the broader market and established tech giants.

For now, Niles believes the market is analogous to 1998—still climbing despite underlying risks. But he cautions that the combination of high stock prices, elevated oil prices, and high bond yields presents a trio of conflicting signals that cannot hold. His advice is clear: stay flexible, but start building a cash position.

This article is for informational purposes only and does not constitute investment advice.