Quantum Computing: Where We Are and Where We’re Headed | NVIDIA GTC 2025 Fireside Chat

Quick Overview

NVIDIA CEO Jensen Huang hosted the first Quantum Day at GTC 2025, inviting leaders from competing quantum computing modalities—including trapped ions, neutral atoms, superconducting qubits, and photonics—to discuss the state-of-the-art, clarify his previous controversial remarks about quantum utility timelines, and announce NVIDIA's new quantum research lab in Boston partnering initially with Quantum Machines and Quantinuum.

Key Points: Jensen Huang initiated Quantum Day at GTC 2025 to learn directly from quantum leaders after his statement on quantum utility caused industry stocks to drop significantly, leading him to state, "the world's got this wrong." NVIDIA clarified its role: "NVIDIA doesn't make quantum computers, but we dedicate ourselves to creating accelerated computing stacks to enable quantum computers," citing tools like CUDAQ, cuQuantum libraries, and DGX Quantum for error correction. NVIDIA announced the start of a new, highly advanced accelerated computing, hybrid quantum computing research lab in Boston, intending to partner initially with Quantum Machines and Quantinuum. Different modalities presented their status: QuEra (neutral atoms) emphasizes identical, well-isolated qubits and evolving connectivity; Rigetti (superconducting) highlights scalability and improved gate fidelity now reaching 99% to 99.5%; Quantinuum (trapped ion) claims the industry's highest fidelities, projecting 100 logical qubits in 18 months. D-Wave, using superconducting annealing, reported a useful computation of magnetic material properties that would take nearly a million years classically, and introduced a distributed quantum application for blockchain proof-of-work. The discussion emphasized that quantum machines should be seen as specialized, complementary "quantum processors" or "precision instruments" for hard quantum problems, not replacements for classical computers, which is a necessary reframing to manage expectations. Future outlooks included goals for next year: Alan Baratz hoped for better model training/inference with lower power consumption; Peter Chapman predicted the first prototypes of a new kind of AGI based on quantum learning.