NVIDIA is using Palantir Foundry and cuOpt to automate its hardware supply chain allocation decisions across global manufacturing sites.
The company measures operational delivery from wafer-out to first token. This window splits into time-to-rack (the transit from fab output to an assembled data centre system) and time-to-token (which covers power, cooling, networking, and day-one software readiness.)
Hardware scaling has magnified supply constraints. An NVIDIA Grace Blackwell NVL72 rack contains 18 compute trays, with each tray requiring two Grace CPUs, four Blackwell GPUs, and 32 HBM3e memory packages sourced across thousands of suppliers, OEMs, and contract design partners.
The upcoming supply chain constructed for NVIDIA’s Vera Rubin architecture is twice as large as the network supporting Grace Blackwell.
Assembly cannot proceed until parts arrive from three designated channels: direct inventory, consignment stock, and external suppliers. Early shipments must wait on delayed components, extending the metric NVIDIA terms ‘Time of Ownership’ (the duration from when a facility receives materials to when finished sub-assemblies depart.)
Factory allocations are reworked weekly over rolling two-quarter horizons to resolve part availability, throughput limits, and customer fulfilment schedules.
To coordinate these dependencies, the NVIDIA operations team built the ‘Digital Supply Chain Intelligence’ command centre using Palantir Foundry. Foundry’s Ontology models facilities, supplier commits, component stocks, and production targets as interconnected objects and links.
NVIDIA cuOpt, an open-source library for GPU-accelerated decision optimisation, reads this operational layer directly. Formulating distribution as a mixed-integer linear program designed to minimise TOO, the solver evaluates parts constraints across every tier of the bill of materials.
Beyond outputting weekly delivery schedules, cuOpt identifies active factory limits, such as regional assembly capacity caps versus raw memory availability.
Mathematical optimisation alone failed to capture unstructured operational variables observed by human planners, including supplier call transcripts, regional weather forecasts, partner email exchanges, and geopolitical events.
NVIDIA addressed this by post-training Nemotron 3.5 Lightning, an open-weight mixture-of-experts model featuring 30 billion total parameters and approximately three billion active parameters per forward pass.
The engineering pipeline processes historical records through NeMo Anonymizer to redact sensitive operational fields, NeMo Data Designer to balance training examples with synthetic capacity disruption scenarios, and NeMo AutoModel to apply low-rank adaptation (LoRA) parameters while keeping base model weights frozen. Palantir Autopilot manages data lineage, model tracking, and recommendation delivery.
Evaluated on historical allocation records, the post-trained Nemotron 3.5 Lightning model achieved 86.7 percent decision accuracy, compared to 55.5 percent for the larger Nemotron 3 Ultra model and 17.5 percent for the un-tuned Lightning base model.
The post-trained model achieved a 58.6 percent balanced accuracy and a 57.5 percent macro-F1 score, outperforming Nemotron 3 Ultra’s 42 percent balanced accuracy and 39.5 percent macro-F1 score.
Fine-tuning completed on two NVIDIA B200 GPUs within minutes. Domain fine-tuning improved allocation decisions, though production risk forecasting further into the future remained difficult.
Operational choices, planner revisions, overrides, and observed factory outputs are continuously written back to the Palantir Ontology.
NVIDIA confirmed this dataset will form preference pairs for reinforcement learning routines – scoring recommendations on allocation precision, policy compliance, and evidence grounding – with production models remaining strictly isolated from live and unmonitored retraining.
See also: Supply chains detect fast, act slow: How AI agents fix it
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the Cyber Security & Cloud Expo. Click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post Palantir Foundry and cuOpt drive NVIDIA supply chain allocation appeared first on AI News.
OneRail has launched an AI-powered delivery platform that uses Nvidia technology to help retailers, wholesalers, and distributors decide how individual orders should be delivered.
Called OmniSTAR, the system evaluates options including owned fleets, couriers, parcel carriers, and other delivery modes, then selects the lowest-cost option that meets the required service level, according to OneRail.
The platform combines Nvidia’s cuOpt decision optimisation engine and cuDF data processing software with OneRail’s delivery pricing and performance data. Nvidia accelerated computing infrastructure is used to process the routing and delivery-mode calculations.
OneRail said the system can reduce computation times by as much as 10 times. A calculation that previously took 20 minutes can be completed in under two minutes, while a calculation taking a week can be reduced to about two days, according to the company.
OneRail said the shorter processing time allows the optimisation to run within live delivery operations, where multiple fulfilment options can be evaluated before an order is assigned.
“If you don’t have the ability to make lightning-fast decisions, you’re giving up margin,” Catania said in an interview with CNBC. “Last-mile fulfilment is expensive.”
OneRail’s broader AI systems use prediction and optimisation for different parts of the delivery process. The company said its machine-learning models estimate factors including service time, lateness risk, the probability of first-attempt delivery success, and expected price ranges.
OneRail said those predictions feed into optimisation systems that determine how an order should be executed. Separately, the company said OmniSTAR compares different fulfilment modes before selecting an option based on cost and service requirements.
Research on dynamic vehicle routing makes a similar distinction between predicting changing conditions and recalculating operational decisions as new information becomes available. A 2024 review in the European Journal of Operational Research identified travel-time prediction and real-time re-optimisation as separate areas of time-dependent routing research.
Nvidia describes cuOpt as an open-source, GPU-accelerated optimisation library for vehicle routing and other mathematical optimisation problems.
Nvidia’s documentation shows that cuOpt can account for vehicle costs, capacities, travel times, operating windows, starting locations, and other restrictions when calculating routes. Its cost models can also use distance, time, monetary cost, or a weighted combination of those measures.
OmniSTAR applies cuOpt to both routing and delivery-mode selection. OneRail said this allows the system to compare available fulfilment options for an order and identify the lowest-cost option that still meets its service requirements.
OneRail said many retailers still rely on static rules or manual planning when making these decisions, and that OmniSTAR is designed to evaluate more delivery combinations within shorter operational timeframes.
Nvidia said cuOpt does not exhaustively test every possible route. Instead, the solver generates candidate solutions and iteratively improves them using GPU-accelerated heuristics to produce high-quality results within a set computation time.
The platform also uses Nvidia cuDF, a GPU-accelerated library for tabular data processing, including filtering, joining, and aggregating datasets.
OneRail combines those capabilities with its own delivery data and operational models. Its dataset is based on millions of deliveries across a network that the company said includes more than 12 million drivers and over 1,000 logistics partners.
The data covers pricing and delivery performance across different transportation modes. OneRail said OmniSTAR can use the information to identify delivery rules that increase costs and assess how delivery choices affect item-level profitability.
The architecture disclosed for OmniSTAR centres on GPU-accelerated data processing and mathematical optimisation. Nvidia describes cuOpt as the optimisation component used for problems including vehicle routing.
Because cuOpt is stateless, changes in operating conditions require the optimisation problem to be modelled and submitted again. Nvidia cites vehicle breakdowns, driver absences, road blockages, traffic, and new high-priority orders as examples of changes that can prompt this type of dynamic reoptimisation.
OneRail said OmniSTAR can rerun delivery scenarios as variables including fuel costs, weather, and shipping conditions change. The company has separately said its use of cuOpt allows it to evaluate more routing scenarios and recalculate routes faster than its previous approach.
OmniSTAR is already deployed with selected enterprise customers.
At US Foods, OneRail said the system identified delivery configurations that were reducing margins, including low-margin products being transported long distances using higher-cost equipment. US Foods subsequently used the findings to adjust pricing and restructure some delivery patterns, according to OneRail.
OneRail also told CNBC that an unnamed large tire distributor using the platform achieved $40 million in run-rate savings over three years. The customer was not identified, and the savings figure was provided by OneRail. The company also told CNBC that it expects OmniSTAR to exceed $6 billion in gross merchandise volume during the fourth quarter of 2026.
CNBC reported that OneRail and Nvidia had worked on the project for three years before its launch. OneRail said the collaboration included direct engagement with Nvidia’s cuOpt engineering team on last-mile delivery and large-scale logistics optimisation, alongside its participation in the Nvidia Inception programme.
In March this year, FedEx launched FedEx SameDay Local in collaboration with OneRail, connecting customers to a national network of more than 1,000 delivery providers.
(Photo by Brecht Corbeel)
See also: A quarter of Nvidia’s business next year comes from labs it is financing
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post OneRail uses Nvidia AI for real-time last-mile delivery optimisation appeared first on AI News.
NVIDIA has agreed to acquire Hugging Face for $12.93 billion to scale the open-source model repository’s platform and infrastructure.
The transaction targets platform growth and infrastructure investment, aiming to expand AI access for enterprise developers, software engineers, and research institutions globally.
Built over the past decade by Clem Delangue, Julien Chaumond, Thomas Wolf, and their engineering team, Hugging Face serves as the primary home for the open model developer community.
Platform metrics show more than 18 million developers, researchers, and creators share more than three million models, 500,000 datasets, and one million applications. Commercial adoption includes more than 200,000 companies using the environment to discover, evaluate, customise, and deploy AI models.
NVIDIA stated that Hugging Face will remain an open platform for the entire AI sector. Developers will retain full control over their selection of models, software frameworks, cloud providers, inference services, and computing hardware.
NVIDIA hardware will not be mandatory to build on or deploy software through the platform. The service will maintain operational support for alternative accelerators, multi-cloud architectures, and open-weight models from all third-party builders.
Jensen Huang, Founder and CEO of NVIDIA, said: “Hugging Face will remain an open platform for the entire AI ecosystem. Developers will choose the models they want, the frameworks they want, the clouds and inference service providers they want, and the computing platforms they want.
“NVIDIA compute will not be required to build on or deploy through Hugging Face.”
Huang noted a recent open letter he co-authored regarding the role of open weights in the AI economy. The position argued that open weights broaden AI access and ensure technical leadership remains distributed across companies, academic institutions, and developer communities.
Under this operational model, commercial businesses, startups, universities, and public bodies can build on advanced capabilities without the expense of training baseline models from scratch.
The approach allows organisations to match specific models to operational tasks across factories, hospitals, farms, classrooms, and commercial businesses, while addressing cybersecurity and data sovereignty requirements.
Julien Chaumond, Co-Founder and CEO of Hugging Face, commented: “AI is at an inflection point. Open-source AI can become less relevant in the coming years if the big closed labs run away with it, or it can become the foundational fabric of the next phase of human civilisation.
“Those are vastly different outcomes, and we need the critical mass to ensure we give our collective best shot to the second outcome. Given Jensen Huang’s stance on open source AI and how he stepped up to defend it when it was under threat earlier in the summer, NVIDIA was the only partner we truly considered.”
NVIDIA stands as the largest contributor of open models and data to Hugging Face, with a portfolio of more than 500 open models and more than 250 open datasets. The company builds its libraries, tools, and models openly to allow external engineers to inspect, modify, and build atop the software.
Technical integration will focus on applying NVIDIA infrastructure and engineering resources to improve repository reliability, safety controls, model evaluation tooling, inference execution, and deployment pipelines.
“I am honored that Clem came to me as he considered the next chapter of Hugging Face and believed NVIDIA would be a great home for the company, its community and the future of open models,” Huang stated.
Hugging Face will retain its independent brand identity following the completion of the transaction, with the existing team continuing operations across multi-cloud and multi-accelerator environments.
“This gives fuel to our long-term vision and mission of unlocking the community’s progress to ensure that AI, which is the greatest breakthrough of our lifetime, is accessible to as many people as possible,” Chaumond concludes.
See also: Motional and MIT AI explains self-driving car decisions
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the Cyber Security & Cloud Expo. Click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post NVIDIA to acquire Hugging Face for $12.93B appeared first on AI News.
The Nvidia Vera chip is rarely the headline when earnings beat estimates, but it should be. When Nvidia reported Q1 revenue of US$81.62 billion on Wednesday, beating analyst estimates of US$78.86 billion, and guided Q2 at US$91 billion–well above Wall Street’s US$86.84 billion forecast–the numbers did what Nvidia numbers always do: dominate the room.
But buried in CEO Jensen Huang’s conference call with analysts was something more strategically interesting than another quarterly beat. Huang told analysts that Nvidia’s new Vera central processors unlock access to a US$200 billion market, one that sits entirely outside the US$1 trillion the company has already forecast from its Blackwell and Rubin AI GPU lineup between 2025 and 2027.
He expects Vera chip revenue to hit US$20 billion by the end of this fiscal year. “I expect (Vera) to be the second largest” sales contributor, Huang said during the call.
That’s not a footnote. That’s a second front.
The reason Nvidia needs a second front is straightforward: its biggest customers are building their own. Google, Amazon, and Microsoft–collectively expected to pour more than US$700 billion into AI infrastructure this year, up sharply from around US$400 billion in 2025, are simultaneously pouring funds into custom silicon to run AI models. Intel and AMD are also touting CPUs as a credible play for inference workloads.
The narrative in the chip industry has shifted from who can train the biggest model to who can serve it cheapest and fastest. Inference is where Nvidia’s GPU dominance is most exposed. Training large models is still firmly Nvidia territory, but inference, generating answers at scale, in real time, is increasingly where custom chips from Google’s TPU line, Amazon’s Trainium and others are making their case.
Nvidia’s answer is Vera. The chip, developed in part using technology from Groq, a startup specialising in inference that Nvidia licensed in a deal reportedly worth around US$17 billion, targets exactly this workload. The full Vera Rubin platform, which combines the Vera CPU with Rubin GPUs, is set to launch later this year.
Huang was candid about one problem: supply. “My sense is that we’ll be supply-constrained through the entire life of Vera Rubin,” he said on the call. It’s a telling admission for a product Nvidia is positioning as a major growth pillar. To get ahead of disruptions, Nvidia is spending heavily on the supply chain. The company disclosed that its supply commitments rose to US$119 billion in Q1, up from US$95.2 billion the previous quarter, a significant jump that reflects both confidence in demand and anxiety about a global memory chip crunch.
Nvidia also announced a US$80 billion share repurchase programme and raised its quarterly cash dividend to 25 cents per share, from 1 cent, moves that signal financial confidence even as Huang warned of tightening supply.
Despite the beats, Nvidia shares fell 1.6% in extended trading after the results. eMarketer analyst Jacob Bourne captured the mood: “Nvidia delivered another beat, but at this point that’s essentially priced in as it keeps beating quarter after quarter. The lingering question is whether it can convince investors the AI buildout has durability into 2027 and 2028, especially as the narrative shifts toward inference workloads and competing silicon from Google, Amazon, AMD, and Intel.”
Huang pushed back with numbers of his own. He pointed to a growing sub-segment of AI-specific cloud customers whose spend is now roughly equal to the hyperscalers, but growing faster quarter-over-quarter. “We should be growing faster than hyperscale capex,” he said.
The Vera chip is central to that argument. Whether the supply chain cooperates is a different question entirely.
(Image source: Nvidia’s Newsroom)
See Also: The Nvidia H200 China deal survived the Trump-Xi summit–just not in the way anyone expected
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and co-located with other leading technology events. Click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post Nvidia’s Vera chip is the US$200 billion bet Jensen Huang doesn’t want you to overlook appeared first on AI News.
President Trump flew to Beijing, brought Jensen Huang along at the last minute, and left two days later, telling reporters that “something could happen” on chip exports. Nothing did. Not a single Nvidia H200 has shipped to China since Trump first authorised the sales in December 2025, and US Trade Representative Jamieson Greer told Bloomberg that semiconductor controls were not even on the bilateral agenda.
The summit theatre obscured a more interesting development underneath it. The H200 isn’t stuck because Washington won’t allow it. Washington already has allowed it. Roughly 10 Chinese firms, including Alibaba, Tencent, ByteDance, and JD.com, hold approved US export licences for up to 75,000 units each, with Lenovo and Foxconn authorised as distributors. The chips aren’t moving because Beijing won’t let its own companies take delivery.
The mechanics of the stalemate are worth understanding clearly. US rules require that all H200 chips ordered by Chinese clients be used only in China. Beijing, meanwhile, has instructed Chinese tech companies to limit their use of Nvidia chips to overseas operations while supporting domestic manufacturing. The two requirements are mutually exclusive.
Chips cleared for export cannot legally be deployed where Beijing wants to deploy them, and Beijing won’t authorise the domestic use the US licences require, according to Implicator.
Commerce Secretary Howard Lutnick stated at a Senate hearing last month that Chinese firms are trying to keep their investment focused on domestic suppliers, including Huawei. Beijing’s State Council has also ordered a supply-chain security review aimed at cutting dependence on US semiconductors.
The policy contradiction is not accidental. That is the point.
The days around the summit produced several data points that matter more for the long term than Trump’s parting comment. DeepSeek confirmed its latest model had been optimised to run on Huawei processors. Tencent’s chief strategy officer said Chinese GPU supply would increase progressively through 2026, and an Alibaba executive said its T-Head proprietary GPUs had achieved scaled mass production.
This follows the April launch of DeepSeek V4, which adapted the model for Huawei’s Ascend chips – the first major Chinese frontier model to do so in training, not just inference. What the summit week confirmed is that the shift is no longer experimental. It is now a supply-chain policy. Nvidia’s China revenue has fallen to roughly 5% in recent quarters, down from above 20% before export controls tightened. The company’s own guidance for the current quarter assumes zero revenue from China.
Huang’s last-minute inclusion in the delegation – Trump called him directly after seeing media coverage that he had not been invited – suggested urgency. The outcome suggested the limits of what CEO diplomacy can achieve when the obstruction is structural, not procedural.
The stalemate matters beyond bilateral optics. Chinese AI platforms are now operating under a domestic mandate to build on Huawei’s compute stack. The question of which AI hardware architecture becomes dominant in the world’s second-largest AI market is being answered not by technical benchmarks but by government directive.
Beijing steering platforms toward Huawei Ascend chips rather than Nvidia H200S is not just a trade posture. It is a structural bet that the performance gap will close fast enough that being locked into the domestic stack is manageable. DeepSeek V4’s results suggest it may be right, at least for inference workloads.
Trump said something could happen. Greer said the decision is sovereign for China. Both are true, and neither changes the current position: the H200 deal is approved, licensed, and frozen, with Huawei filling the space it leaves behind.
(Image source: The White House)
See Also: Can China’s chip stacking strategy really challenge Nvidia’s AI dominance?
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and co-located with other leading technology events. Click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post The Nvidia H200 China deal survived the Trump-Xi summit–just not in the way anyone expected appeared first on AI News.
Anthropic’s most capable AI model has already found thousands of AI cybersecurity vulnerabilities across every major operating system and web browser. The company’s response was not to release it, but to quietly hand it to the organisations responsible for keeping the internet running.
That model is Claude Mythos Preview, and the initiative is called Project Glasswing.
The launch partners include Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, Nvidia, and Palo Alto Networks.
Beyond that core group, Anthropic has extended access to over 40 additional organisations that build or maintain critical software infrastructure. Anthropic is committing up to US$100 million in usage credits for Mythos Preview across the effort, along with US$4 million in direct donations to open-source security organisations.
Mythos Preview was not specifically trained for cybersecurity work. Anthropic said the capabilities “emerged as a downstream consequence of general improvements in code, reasoning, and autonomy”, and that the same improvements making the model better at patching vulnerabilities also make it better at exploiting them.
That last part matters. Mythos Preview has improved to the extent that it mostly saturates existing security benchmarks, forcing Anthropic to shift its focus to novel real-world tasks–specifically, zero-day vulnerabilities. These flaws were previously unknown to the software’s developers.
Among the findings: a 27-year-old bug in OpenBSD, an operating system known for its strong security posture. In another case, the model fully autonomously identified and exploited a 17-year-old remote code execution vulnerability in FreeBSD–CVE-2026-4747–that allows an unauthenticated user anywhere on the internet to obtain complete control of a server running NFS. No human was involved in the discovery or exploitation after the initial prompt to find the bug.
Nicholas Carlini from Anthropic’s research team described the model’s ability to chain together vulnerabilities: “This model can create exploits out of three, four, or sometimes five vulnerabilities that in sequence give you some kind of very sophisticated end outcome. I’ve found more bugs in the last couple of weeks than I found in the rest of my life combined.”
“We do not plan to make Claude Mythos Preview generally available due to its cybersecurity capabilities,” Newton Cheng, Frontier Red Team Cyber Lead at Anthropic, said. “Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. The fallout–for economies, public safety, and national security–could be severe.”
This is not hypothetical. Anthropic had previously disclosed what it described as the first documented case of a cyberattack largely executed by AI–a Chinese state-sponsored group that used AI agents to autonomously infiltrate roughly 30 global targets, with AI handling the majority of tactical operations independently.
The company has also privately briefed senior US government officials on Mythos Preview’s full capabilities. The intelligence community is now actively weighing how the model could reshape both offensive and defensive hacking operations.
One dimension of Project Glasswing that goes beyond the headline coalition: open-source software. Jim Zemlin, CEO of the Linux Foundation, put it plainly: “In the past, security expertise has been a luxury reserved for organisations with large security teams. Open-source maintainers, whose software underpins much of the world’s critical infrastructure, have historically been left to figure out security on their own.”
Anthropic has donated US$2.5 million to Alpha-Omega and OpenSSF through the Linux Foundation, and US$1.5 million to the Apache Software Foundation–giving maintainers of critical open-source codebases access to AI cybersecurity vulnerability scanning at a scale that was previously out of reach.
Anthropic says its eventual goal is to deploy Mythos-class models at scale, but only when new safeguards are in place. The company plans to launch new safeguards with an upcoming Claude Opus model first, allowing it to refine them with a model that does not pose the same level of risk as Mythos Preview.
The competitive picture is already shifting around it. When OpenAI released GPT-5.3-Codex in February, the company called it the first model it had classified as high-capability for cybersecurity tasks under its Preparedness Framework. Anthropic’s move with Glasswing signals that the frontier labs see controlled deployment–not open release–as the emerging standard for models at this capability level.
Whether that standard holds as these capabilities spread further is, at this point, an open question that no single initiative can answer.
See Also: Anthropic’s refusal to arm AI is exactly why the UK wants it
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the Cyber Security & Cloud Expo. Click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post Anthropic keeps new AI model private after it finds thousands of external vulnerabilities appeared first on AI News.
Login