NVIDIA Expands Open Source CUDA-Q Platform

thequantuminsider.com Sep 14, 2026

NVIDIA Expands Open Source CUDA-Q Platform

Curated by Shubham · TechCapsules · Sep 14, 2026

AI-summarised brief · reviewed before publication

NVIDIA announced CUDA‑Q Logical, an open‑source orchestration layer that lets researchers design and test fault‑tolerant quantum applications. The tool enables rapid prototyping of logical qubit systems, allowing Iceberg Quantum to model 1,000 logical qubits with only 150,000 physical qubits—about ten times fewer than prior estimates. Fermilab cut evaluation time from five months to three weeks, and Sandia released its QUOPS benchmark through CUDA‑Q. The platform is adopted by major labs, accelerating co‑design of algorithms, error‑correction codes, and hardware.

💡 Why It Matters

· By streamlining the design cycle, CUDA‑Q Logical reduces the hardware and time barriers that have slowed practical quantum computing, enabling faster progress toward real‑world applications.

Read full article on thequantuminsider.com

Infleqtion And Nvidia Reduce Physical-to-Logical Qubit Ratio in Quantum Computers

Infleqtion and Nvidia announced a breakthrough in quantum error correction that reduces the physical-to-logical qubit ratio to 5.4:1, using 98 physical data qubits to encode 18 logical qubits. The achievement combines Infleqtion’s neutral‑atom Sqale platform with Nvidia’s CUDA‑Q Logical compiler, deploying a hypergraph‑product simplex code that achieves an 18.4 % encoding rate. This milestone follows similar advances by D‑Wave and Quantinuum, all aiming to lower qubit overhead and move quantum computing toward commercial viability.

💡 By slashing qubit overhead, the partnership accelerates the transition from experimental to scalable quantum processors, tightening the gap between theoretical models and practical, high‑performance quantum systems.

techradar.com · Sep 16

Nvidia RTX 5090 GPU prices skyrocket to $9,000 — as fresh RTX…

Nvidia’s flagship RTX 5090 has seen its asking price surge to nearly $10,000, with the Asus ROG Astral variant listed at $9,659 on Newegg and a gold‑plated edition at $11,999. Retailers have run out of stock, forcing buyers to turn to third‑party sellers, many shipping from Asia. In the UK, prices hover around £4,500 to £6,600. A new rumor claims the next‑generation RTX 6000 (Rubin) will launch in 2027, earlier than the previously expected 2028, raising concerns about future pricing.

💡 The steep price climb signals a tightening supply chain and a potential shift in Nvidia’s market strategy. Early 2027 launch speculation could reshape competitive dynamics and consumer expectations for high‑end GPUs.

techcrunch.com · Sep 16

We don’t need AI regulation — leave safety to us, Nvidia’s Jensen…

At Dreamforce, Nvidia founder and CEO Jensen Huang argued that artificial intelligence does not require new legislation, describing it as “hardware and software built by humans” that can be managed through existing product‑liability laws and market forces. He framed safety as an engineering issue, insisting companies will self‑regulate by withholding unsafe products until they are confident of performance. Huang warned that rapid innovation and safety are not mutually exclusive, urging firms to pause development when risks emerge. While acknowledging Nvidia’s deep involvement in AI hardware, model development, and open‑source tools, he dismissed regulatory intervention as a potential hindrance to the industry’s growth and to Nvidia’s own sales ambitions.

💡 Huang’s stance places the burden of AI risk mitigation on corporate discretion, shaping the debate over whether voluntary standards can replace government oversight in a rapidly expanding market.

quantumcomputingreport.com · Sep 15

Quandela and NVIDIA Outline Photonic QPU Integration Architecture via NVQLink

Quandela and NVIDIA released a joint white paper at IEEE Quantum Week 2026 outlining a photonic QPU integration architecture using NVIDIA’s NVQLink. The framework connects Quandela’s Quantum System Controller, which manages FPGA‑based pulse control for its MosaiQ QPU, directly to NVIDIA GPU nodes via a low‑latency Remote Direct Memory Access (RoCE) interconnect. Sub‑4‑microsecond round‑trip latency enables microsecond‑scale GPU‑QPU communication, supporting GPU‑accelerated state‑vector simulations, real‑time quantum error correction, and automated pulse calibration within a unified CUDA‑Q host process. The paper details a three‑phase commercial roadmap—Access, Integration & Discovery, and Scale—and highlights the partnership’s potential to embed photonic quantum processors in existing HPC data centers.

💡 The integration demonstrates a practical pathway for deploying photonic quantum processors alongside conventional GPU workloads, potentially accelerating quantum‑classical hybrid applications and reducing the barrier to entry for quantum research in commercial data centers.

tbsnews.net · Sep 15

Trump dials Nvidia CEO, slams AI safety regulations as a ‘hoax’

President Donald Trump called Nvidia CEO Jensen Huang during the All‑In Summit to dismiss AI safety concerns as a “hoax” and urge the United States to maintain its lead over China in artificial‑intelligence development. Trump praised data‑center expansion, calling it “the oil of the next 20‑25 years,” and said it would bring wealth to states. The conversation was recorded, posted on X, and later shared by the Trump War Room. Nvidia declined to comment on the exchange. The remarks come amid industry calls for stronger AI safeguards.

💡 Trump’s stance signals a federal push that could accelerate U.S. AI deployment while sidestepping regulatory debate, potentially widening the technology gap with China.

eenewseurope.com · Sep 15

NVIDIA extends CUDA-Q for fault-tolerant quantum systems

NVIDIA has expanded its open‑source CUDA‑Q quantum computing platform with CUDA‑Q Logical, an orchestration layer that enables researchers to design and test fault‑tolerant systems using logical qubits. The new layer models quantum algorithms, error‑correction codes, hardware architectures, and QPU components, allowing rapid evaluation of complete system configurations. Early adopters include Fermilab, Infleqtion, IQM, QCDesign, Quantum Motion, Sandia, and Iceberg Quantum. Fermilab reported a 7‑fold reduction in development time, cutting a five‑month workflow to three weeks, while Iceberg Quantum estimated 1,000 logical qubits with 150,000 physical qubits—ten times fewer than prior projections.

💡 By accelerating fault‑tolerant design, the platform shortens the path to practical, large‑scale quantum applications, directly impacting research fields that rely on error‑corrected qubits.