What factors are driving sustained high demand and value for GPUs in AI-related markets?

All answers cite surging demand for compute—driven by AI model advances, application usefulness, and supply-demand imbalances—as the core reason for GPUs’ sustained high value.

9 answers

Answers

  • Jason Goodison

    Cofounder & CTO · General Compute

    Low Latency

    Jason Goodison says latency, the end-to-end generation time, is the most obvious factor, noting that AI coding agents want answers as soon as possible and that Nvidia hardware struggles with the decode part of inference on very large models.

    The answer starts at 2:24.

    MTS · Listen on Apple Podcasts
  • Jon Gray

    President & Chief Operating Officer · Blackstone

    CUDA Ecosystem & Workload Demand

    Jon Gray says there is just so much demand for chips, and credits NVIDIA's extraordinary position in its chips and CUDA software and how it integrates things.

    Blackstone · Watch on YouTube
  • Stacy Rasgon

    Managing Director & Senior Analyst, U.S. Semiconductors and Semiconductor Capital Equipment · AB Bernstein

    Performance Outpacing Price

    Stacy Rasgon says Nvidia's GPU prices rise each generation but performance rises more than the price, so customers get more value than the added cost, which he calls very NPV positive.

    Chipstrat · Chipstrat · Read
  • Don Wilson

    Founder and Chief Executive Officer · Drw

    Inference Demand from New Models

    Don Wilson says demand for inference surged after Anthropic's newest model came out, a step up in usefulness that created a ton of inference demand, which he cites as driving GPU prices up.

    The answer starts at 9:53.

    Odd Lots · Listen on Apple Podcasts
  • Jensen Huang

    Founder and CEO · Nvidia

    AI Applications Becoming Useful

    Jensen Huang says compute demand is high right now because AI applications are going through an inflection and becoming useful, which brings in $500 billion of venture funding and thousands of startups that all need compute.

    The answer starts at 1:31:48.

    Hard Fork · Listen on Apple Podcasts
  • Mike Maniscalco

    Chief Executive Officer · QumulusAI

    Supply-Demand Balance

    Mike Maniscalco says the price stability on older chips is largely "just a supply, demand, balance thing," pointing to extended contracts and prediction markets as evidence.

    The answer starts at 23:43.

    The Drill Down · Listen on Apple Podcasts
  • Harry Stebbings

    Founder · 20vc

    Intrinsic Compute Capability

    Harry Stebbings says GPUs hold intrinsic value because they can be used for valuable work anywhere, and he ties that value to how many teraflops per second a device can do.

    20VC with Harry Stebbings · Watch on YouTube
  • Matthew Frankel, CFP®

    Stock Analyst · The Motley Fool

    Supply-Demand Imbalance

    Rachel says the roughly 75% gross margins Nvidia commands come down to a supply-demand imbalance in high-end computing, with hyperscalers ordering chips faster than they can be produced.

    Motley Fool Money · Motley Fool Money · Listen at 4:29
  • Dave B. Blundin

    Managing Partner · Link Ventures

    Ever-Growing Compute Demand

    Dave B. Blundin says demand for compute is going to near infinity, which he says makes these fundamentally good investments.

    Moonshots with Peter Diamandis · Moonshots with Peter Diamandis · Listen at 1:00:17