The September 14 selloff treated slower frontier AI development as bad news for chips. The more interesting question is whether it changes which chips get bought. Reuters Breakingviews argued that a moderation in giant model-training runs could redirect part of this year’s roughly $1 trillion AI investment toward inference, where existing models answer queries and run agents.
That creates a different contest for Intel Corporation (NASDAQ:INTC) and NVIDIA Corporation (NASDAQ:NVDA).
Inference does not erase GPUs, but it broadens the bill of materials
Intel’s opportunity is the server CPU. Production AI applications need orchestration, memory, networking and general-purpose compute around accelerators, while inference and agentic deployments can require substantially more general-purpose compute alongside accelerators. Intel Corporation also reported $6.3 billion of Data Center and AI revenue in its latest quarter, up 59% year over year. A shift toward deployed agents could therefore reinforce its core franchise before Intel proves that its foundry turnaround can earn attractive returns.
Execution is where the thesis can break. Intel is spending heavily to rebuild manufacturing leadership and still generated negative adjusted free cash flow in its latest quarter. Inference growth helps only if Intel keeps enough share and margin against AMD, Arm-based CPUs and specialized accelerators.
NVIDIA Corporation remains the strongest counterargument to a simple rotation thesis. Its GPUs run both training and inference, and CUDA plus newer inference-focused systems give Nvidia a path to monetize deployed AI even if frontier training slows. For Nvidia, the downside is relative rather than existential: fewer giant training clusters could reduce demand for the most expensive accelerators or weaken pricing at the margin.
Hedge funds were adding both before the debate changed
Insider Monkey’s database counted 138 hedge funds holding Intel in Q2 2026, up from 112 in Q1. AQR Capital Management owned 10,742,567 shares after trimming its position 7%. Nvidia rose to 285 funds from 275, while Fisher Asset Management increased its stake 3% to 90,935,947 shares. Those filings predate the September 14 safety-driven market reaction.
Intel had 152,248,752 shares sold short on August 31, equal to 3.02% of float, with 1.69 days to cover.
A frontier slowdown would not automatically make Intel a winner or Nvidia a loser. It could, however, change the mix of spending from maximum-scale training toward serving billions of useful requests. If that happens, the AI trade becomes less concentrated in one type of accelerator and more dependent on the entire data-center stack. Intel finally has a credible reason to care about that shift.
READ NEXT:
The Trillion-Dollar AI Capex Gap: Why Too Much Hardware Could Be Nvidia’s Trap and Microsoft’s Opportunity and Redditors Are Skeptical of SpaceX’s Orbital Data Centers. Are They Right?
