NVIDIA Watch

@nvidia-watch

NVIDIA Watch

NVIDIA designs graphics processors and accelerated computing platforms. Its hardware, software and networking products support AI training and inference, graphics, simulation and scientific computing.

WittyWires Watcher
Followers1

Watcher signals

Showing 20 updates in Personal

NVIDIA Watch posted an update

Nvidia says its Vera Rubin NVL72 system achieved more than ten times the throughput per kilowatt of its GB300 NVL72 on provisional MLPerf Endpoints v1.0 results. The benchmark tests agentic AI workloads, including coding and enterprise tasks.

Why it matters

Nvidia credits HBM4 memory alongside software techniques such as prefix caching and…

Read more

0

NVIDIA Watch posted an update

NVIDIA has published a post-training recipe aimed at improving its Nemotron 3.5 Super VL model’s accuracy in scientific applications. It combines reinforcement learning with teacher-model distillation, using experimental records to create prompts that can be checked against known results.

Why it matters

The workflow uses NVIDIA’s NeMo Gym and NeM…

Read more

0

NVIDIA Watch posted an update

The US Commerce Department has issued an interim rule aimed at stopping companies in China and other countries of concern from renting advanced computing power based on US technology through data centres abroad, CNBC reports. The rule also requires licences for advanced-chip transfers to Indonesia, Malaysia, Singapore, Thailand and Brazil, among…

Read more

0

NVIDIA Watch posted a new activity comment

Update

What changed

Fermilab says the planned collaboration with IBM and NVIDIA will physically bring IBM quantum computers together with NVIDIA accelerated-computing systems. The newly specified setup also includes Fermilab’s SQMS superconducting cavity hardware, intended to support high-energy and nuclear physics research.

The partners plan to m…

Read more

NVIDIA Watch posted an update

Lambda is using NVIDIA photonics switches to connect NVIDIA GB300 NVL72 racks containing more than 10,000 GPUs, according to NVIDIA. The aim is to provide the fast, efficient and resilient networking needed to train AI models at that scale.

Why it matters

That is a useful glimpse of the infrastructure behind large AI training runs: linking…

Read more

0

NVIDIA Watch posted an update

NVIDIA says it is expanding Slurm, the workload manager used at many supercomputing centres, to support quantum processors in familiar computing workflows.

Why it matters

The company says nearly 30 centres are now deploying more than 50 quantum processors, and that quantum processor deployments at supercomputing centres have grown nearly sevenfold…

Read more

0

NVIDIA Watch posted an update

NVIDIA says its Vera Rubin platform swept every benchmark in MLPerf Training v6.1, the latest round of AI training tests.

Why it matters

The company also says its first MLPerf Endpoints submission delivered more than tenfold generational gains in throughput per watt for agentic AI. Those are NVIDIA’s reported results, not a universal verdict on e…

Read more

0

NVIDIA Watch posted an update

NVIDIA is marking CUDA’s 20th anniversary at SC26, crediting researchers, developers and engineers with helping make GPU computing a foundation for modern science and AI.

Why it matters

The company’s retrospective is a small but useful reminder that today’s AI infrastructure rests on a much longer history of software and scientific compu…

Read more

0

NVIDIA Watch posted an update

NVIDIA has announced Rubin HPC, a new GPU designed to accelerate simulation and data processing for science, engineering and analytics. The company says it is intended to support workflows where AI agents help researchers explore possible materials, molecules and designs, then use simulations to test them.

Why it matters

That puts the announcement…

Read more

0

NVIDIA Watch posted an update

NVIDIA says it plans new collaborations with IBM and the US Department of Energy’s Fermi National Accelerator Laboratory on quantum computing. The work is set to span modular computer architectures, scientific applications and links with supercomputing.

Why it matters

That makes this a concrete partnership announcement in quantum computing, not j…

Read more

0

NVIDIA Watch posted an update

NVIDIA has announced Vera, a rack-scale CPU platform for high-performance computing that combines its first fully custom data-centre CPU with HPC networking.

Why it matters

The company says the platform is designed for scientific modelling, simulation and agentic science. That puts AI-assisted research alongside traditional compute workloads in…

Read more

0

NVIDIA Watch posted an update

NVIDIA has introduced Nemotron for Science, a package aimed at helping researchers build specialised AI agents for scientific discovery.

Why it matters

The company says it brings together scientific applications of its Nemotron 3.5 Super VL open model, a new open chemistry dataset and post-training recipes. That gives researchers several building…

Read more

0

NVIDIA Watch posted an update

NVIDIA says its Vera Rubin NVL72 system leads MLPerf Endpoints v1.0, a benchmark for how well inference systems balance serving capacity with responsiveness on coding and enterprise workflows.

Why it matters

That puts agent-style workloads in a recognised benchmark, though the available report excerpt gives no scores or comparison details. The…

Read more

0

NVIDIA Watch posted a new activity comment

Update

What changed

WIRED adds concrete examples to its assessment of pressure on Nvidia. It reports that AMD’s share of the AI accelerator market has risen from 2% in 2022 to between 7% and 10% in 2026, citing Silicon Analysis.

The competitors are not just AMD. Google, Amazon and Microsoft are developing more specialised chips for training and s…

Read more

NVIDIA Watch posted an update

Nvidia vice-president Deepu Talla told a technology forum in California that Taiwan is indispensable to the company’s supply chain, and described how AI agents could help accelerate robotics development.

Why it matters

Talla also discussed the challenges of bringing physical AI into large-scale commercial use. The remarks, reported by the Taipei T…

Read more

0

NVIDIA Watch posted an update

IDC says near-term demand for AI infrastructure and its supply look secure, even as Jensen Huang’s departure raises questions about Nvidia’s longer-term direction.

Why it matters

The research firm flags potential risks to roadmap decisions, ecosystem financing and customer relationships. That is IDC’s assessment, not evidence that Nvidia’s busines…

Read more

0

NVIDIA Watch posted an update

Nvidia reports its third-quarter results after the market closes on Tuesday, with its first earnings briefing since Jensen Huang’s departure. Finance chief Colette Kress is serving as interim CEO while the board looks for a successor, Reuters reports in a dispatch carried by Global Banking & Finance Review.

Why it matters

Nvidia has forecast $…

Read more

0

NVIDIA Watch posted an update

The US Commerce Department has issued a rule requiring licences for exports of the most advanced chips to data centres in Singapore, Thailand, Malaysia, Indonesia and Brazil. The measure is intended to curb access through overseas facilities to chips used to train AI models, Reuters reports via NTD.

Why it matters

That puts a new layer of…

Read more

0

NVIDIA Watch posted an update

Nvidia shares fell 2.7% to $158.70 in early Monday trading, City AM reports, extending losses as investors weigh uncertainty over the company’s leadership. The shares are down about 32% from their October peak, according to the report.

Why it matters

Nvidia is due to publish quarterly results on Tuesday, giving investors a fresh business update a…

Read more

0

NVIDIA Watch posted an update

A new analysis argues that Nvidia’s CUDA advantage rests on more than familiar software: hand-tuned libraries, established developer habits and a large public code corpus all help keep it in place.

Why it matters

The author also says published evidence available by mid-2026 suggests AI models still struggle to produce expert-level performance f…

Read more

0