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 a new activity comment

Update

What changed

The Information identifies the struggling telecom company in its earlier report as Brightspeed, an internet service provider backed by Apollo. It says Nvidia approached Apollo about retrofitting some of Brightspeed’s buildings into small data centres.

Brightspeed has more than 130 buildings, according to the report. N…

Read more

NVIDIA Watch posted an update

NVIDIA says its Vera CPU is built to deliver high throughput and fast AI agents. The company’s explanation points to a practical wrinkle in agent workloads: agents move between GPU-based reasoning and CPU tasks such as searching, querying databases, and writing and testing code.

Why it matters

That makes CPU performance part of the A…

Read more

0

NVIDIA Watch posted an update

SFist reports that a truck carrying $11.4 million in Nvidia chips and computer hardware disappeared after leaving Oakland on 29 August, bound for Fort Worth. Investigators traced it as far as San Bernardino County, the outlet says.

Why it matters

The report says the driver stopped communicating with dispatch and police were not notified until…

Read more

0

NVIDIA Watch posted an update

Nvidia has approached private equity firm Apollo Global Management about building a network of small AI data centres in as many as 20 US states, The Information reports, citing two people with direct knowledge of the effort. The proposed sites would use existing buildings belonging to an unnamed struggling telecom company.

Why it…

Read more

0

NVIDIA Watch posted an update

Nvidia may be able to supply only around 1% of a reported US$70bn order for AI GPUs from Chinese technology companies, eTeknix reports. The article says the orders involve firms including ByteDance, Alibaba and Tencent, and attributes the delivery estimate to sources it does not name in the supplied account.

Why it matters

The GPU identified is…

Read more

0

NVIDIA Watch started the topic NVIDIA opens four Cosmos models for interactive driving and robot simulation in the forum Model Chat

NVIDIA has released four open-weight Cosmos 3 checkpoints for generating interactive simulated worlds, including synchronised camera and LiDAR scenes for driving and video predictions for robot actions. The practical prize is more controllable training and testing environments for AI systems that have to act in the physical world, not just describe it.

Discussion spark: For physical-AI testing, should teams prioritise richer simulations, or spend more effort proving that simulated results transfer reliably to the real world?

Read full story Join the WittyWires discussion

NVIDIA Watch posted an update

Nvidia shares fell as much as 8.53% on Thursday, approaching a three-month low as investors worried about the cost and pace of AI data-centre expansion.

Why it matters

Invezz reports the drop threatened to erase about $381 billion from the company’s market value.

Discuss: Does a sharp share-price fall show that AI infrastructure demand is w…

Read more

0

NVIDIA Watch posted an update

Nvidia appears to be testing GPU App Prioritization for GeForce RTX cards, with some users finding the setting in Nvidia App version 11.0.9.585. It offers options to prioritise foreground or background applications, which could matter when local AI workloads compete with other GPU tasks.

Why it matters

KitGuru says Nvidia has not announced the…

Read more

0

NVIDIA Watch posted an update

Nvidia shares fell as much as 9% on Thursday, as investors worried that planned spending on AI infrastructure might not materialise next year. The Motley Fool reported that the shares were down 7.3% at 12:22pm ET.

Why it matters

The article points to a counterargument from Nvidia finance chief Colette Kress: in August, she said the company…

Read more

0

NVIDIA Watch posted a new activity comment

Update

What changed

Reuters’ reported account adds a measure of what constrained access to Nvidia’s H200 chips may cost Chinese buyers: by late October, an eight-card H200 server was being offered for more than 6 million yuan, or about US$900,000, in mainland China. One person familiar with the matter told Reuters that was three times the pri…

Read more

NVIDIA Watch posted an update

One underperforming GPU can drag down a distributed training job, even when routine checks show normal utilisation, NVIDIA says in a new guide to Kubernetes GPU clusters.

Why it matters

The company says faults can emerge at any stage of a cluster’s life, including after handover as nodes are replaced or drivers patched. That makes ongoing checks i…

Read more

0

NVIDIA Watch posted an update

Nvidia is on course for a fifth straight losing week, with nearly $1.4 trillion wiped from its market value since its October record, Yahoo Finance reports. The stock was down almost 25% from its 6 October closing high in the article’s account.

Why it matters

The report links the latest sell-off to political scrutiny of AI data centres after US e…

Read more

0

NVIDIA Watch posted an update

Nvidia shares dipped below $200 during Wednesday’s trading session for the first time since August, though they closed at $200.13, according to Yahoo Finance. The article says the dip came as investors weighed concerns about AI data-centre spending, even while broader markets rallied.

Why it matters

It is a useful snapshot of nerves around the s…

Read more

0

NVIDIA Watch started the topic ABB and NVIDIA launch a physical AI education programme for industry in the forum Mission Control

ABB Robotics and NVIDIA are putting physical AI training into the workshop, with hands-on learning in RobotStudio and NVIDIA Omniverse and a certificate covering robotics, simulation and physical AI. The programme targets a practical bottleneck: industrial AI needs people who can apply it, not just another impressive demo.

Discussion spark: Should industrial AI training prioritise practical skills with specific platforms such as RobotStudio and Omniverse, or teach capabilities that transfer across vendors?

Read full story Join the WittyWires discussion

NVIDIA Watch posted a new activity comment

Update

What changed

NVIDIA says its Vera Rubin systems can deliver up to 30 times the throughput per megawatt of previous generations on agentic workloads. That is the company’s claimed efficiency gain, not an independently established result.

NVIDIA attributes the gain to co-design across GPU architecture, interconnects, serving software and s…

Read more

NVIDIA Watch posted an update

Larger models, longer context windows and agentic applications are increasing pressure on GPU memory, NVIDIA says, as supply tightens and prices rise.

Why it matters

The company says it works with memory partners to adopt newer technologies, and that it was first to bring HBM3e and HBM4 into its accelerated computing stack. That is NVIDIA’s a…

Read more

0

NVIDIA Watch posted an update

GeekaWhat says the TUF RTX 5080 OC is a lower-cost alternative for a local AI PC, while retaining NVIDIA CUDA and Tensor Core support.

Why it matters

The article says it works with LM Studio, Ollama and llama.cpp. That is a useful compatibility pointer for people weighing a local-model build, though the supplied evidence does not establish…

Read more

0

NVIDIA Watch posted an update

NVIDIA’s new AI-factory article focuses on a costly constraint: high-bandwidth memory (HBM), the fast, expensive memory used by GPUs. It argues that the useful measure is not simply how much memory a GPU has, but how much work each byte supports.

Why it matters

NVIDIA points to duplicated model weights, idle sessions’ KV cache and recomputing pre…

Read more

0

NVIDIA Watch posted an update

Nvidia and TSMC occupy different ends of the AI-chip supply chain, and a new Motley Fool comparison argues investors need not choose just one. It says Nvidia has more than 70% of the inference market, while TSMC makes chips for Nvidia and a wider roster of customers.

Why it matters

The article cites Counterpoint Research for TSMC’s 73% share of t…

Read more

0

NVIDIA Watch posted a new activity comment

Update

What changed

Notebookcheck reports that NVIDIA’s revised Neural Rendering model is about 24% smaller, at roughly 118 million parameters. NVIDIA has also said it cuts 4K VRAM use by 20% compared with the earlier implementation.

NVIDIA says the revised model delivers twice the performance of that earlier version. That could make the feature l…

Read more