Search “NVIDIA”

SWI Compute emerges from stealth and announces plans for European NVIDIA AI factories

On October 26, 2026, SWI Compute emerged from stealth and announced that it will build and operate NVIDIA AI factories in Europe. According to Unite.AI, the plan is backed by 3.6GW of power capacity and approximately $1 billion in NVIDIA system orders. The first factory is located in Cambridge, UK, with an initial scale of 30MW, deploying nearly 7,000 Rubin GPUs and 3,500 Vera CPUs, and is expected to come online by the end of 2027, with plans to expand to 2.3GW by 2029. The company says that starting in 2027, new sites will adopt the NVIDIA DSX AI Factory platform and the Vera Rubin reference design, which can increase GPU capacity by up to 40% within the same power budget. There are currently no subsequent developments.

View story:SWI Compute Emerges from Stealth and Unveils European NVIDIA AI Factory Plans

Nvidia deploys industrial AI systems at Siemens' Erlangen factory

On October 24, 2026, Nvidia deployed systems including digital twins, AI-generated training data and AI agents at Siemens' electronics factory in Erlangen, Germany, to accelerate production and reduce costs. The two companies said that using digital twins to generate realistic images of PCB assemblies to pretrain inspection systems can cut engineering effort by up to 85% and shorten time to market by up to 40%. The factory already uses more than 100 AI algorithms for production, logistics, quality inspection and maintenance, and Siemens plans to invest about $559 million in upgrading the plant by 2030. Analysts noted that these savings figures come from Siemens' own high-mix factory, and whether they can be replicated at existing multi-vendor sites remains to be verified. There are currently no subsequent developments.

gpu-unlock 1.0.0 provides FP/BF16 unlock for Nvidia 3000 series

The open-source project gpu-unlock 1.0.0 provides an FP/BF16 instruction issue rate unlock for Nvidia 3000 series graphics cards. According to a report dated October 24, 2026, this Linux kernel module can complete the unlock before the NVIDIA driver binds to the graphics card, doubling the FP/BF16 FMA instruction issue rate of GA102, GA103, GA104, and GA106 graphics cards to bring them to professional-card levels, thus eliminating the need to manually patch after every driver update. In measured tests, the RTX 3090 Ti's BF16 rose from 87.50 TFLOPS to 157.41 TFLOPS, FP16 from 87.46 to 118.39 TFLOPS, and TF32 from 42.58 to 82.61 TFLOPS. Subsequent reports stated that the tool lifts the restriction by rewriting the speed select register, with no need to flash the VBIOS or modify the driver; the RTX 3090 Ti's BF16 improves by about 80%, ComfyUI image generation throughput improves by 56% to 59%, and the RTX 3090's prompt prefill throughput under vLLM improves by about 33%, but the Decode stage of per-token generation sees almost no improvement due to VRAM bandwidth limits. This modification is not officially supported by Nvidia, and you must bear the power consumption, cooling, and warranty risks yourself.

The report says Amazon purchased Nvidia racks in advance but lacks power for deployment.

According to internal emails seen by Business Insider, Amazon purchased approximately $5 billion worth of Nvidia GB300 server racks in advance, but some U.S. AWS sites lack sufficient power, so the hardware can only be temporarily stored in transit warehouses. The report says AWS alone is expected to receive about $5 billion worth of GB300 racks in August, with roughly another $3 billion added each month thereafter. This phenomenon, known as dark GPUs, is putting financial pressure on data centers; Amazon's second-quarter free cash flow turned negative for the first time since 2023, and capital expenditures this year are expected to be about $220 billion. There are currently no subsequent developments.

Report: Amazon Procured Nvidia Racks Early but Lacks Power to Deploy Them

According to internal emails seen by Business Insider, Amazon purchased approximately $5 billion worth of Nvidia GB300 server racks in advance, but some U.S. AWS sites lack sufficient power, so the hardware can only be temporarily stored in transit warehouses. The report says AWS alone is expected to receive about $5 billion worth of GB300 racks in August, with roughly another $3 billion added each month thereafter. This phenomenon, known as dark GPUs, is putting financial pressure on data centers; Amazon's second-quarter free cash flow turned negative for the first time since 2023, and capital expenditures this year are expected to be about $220 billion. There are currently no subsequent developments.

View story:The report says Amazon purchased Nvidia racks in advance but lacks power for deployment.

Nvidia deploys industrial AI systems at Siemens Erlangen plant

On October 24, 2026, Nvidia deployed systems including digital twins, AI-generated training data and AI agents at Siemens' electronics factory in Erlangen, Germany, to accelerate production and reduce costs. The two companies said that using digital twins to generate realistic images of PCB assemblies to pretrain inspection systems can cut engineering effort by up to 85% and shorten time to market by up to 40%. The factory already uses more than 100 AI algorithms for production, logistics, quality inspection and maintenance, and Siemens plans to invest about $559 million in upgrading the plant by 2030. Analysts noted that these savings figures come from Siemens' own high-mix factory, and whether they can be replicated at existing multi-vendor sites remains to be verified. There are currently no subsequent developments.

View story:Nvidia deploys industrial AI systems at Siemens' Erlangen factory

gpu-unlock 1.0.0 provides FP/BF16 unlock for Nvidia 3000 series

The open-source project gpu-unlock 1.0.0 provides an FP/BF16 instruction issue rate unlock for Nvidia 3000 series graphics cards. According to a report dated October 24, 2026, this Linux kernel module can complete the unlock before the NVIDIA driver binds to the graphics card, doubling the FP/BF16 FMA instruction issue rate of GA102, GA103, GA104, and GA106 graphics cards to bring them to professional-card levels, thus eliminating the need to manually patch after every driver update. In measured tests, the RTX 3090 Ti's BF16 rose from 87.50 TFLOPS to 157.41 TFLOPS, FP16 from 87.46 to 118.39 TFLOPS, and TF32 from 42.58 to 82.61 TFLOPS. Subsequent reports stated that the tool lifts the restriction by rewriting the speed select register, with no need to flash the VBIOS or modify the driver; the RTX 3090 Ti's BF16 improves by about 80%, ComfyUI image generation throughput improves by 56% to 59%, and the RTX 3090's prompt prefill throughput under vLLM improves by about 33%, but the Decode stage of per-token generation sees almost no improvement due to VRAM bandwidth limits. This modification is not officially supported by Nvidia, and you must bear the power consumption, cooling, and warranty risks yourself.

View story:gpu-unlock 1.0.0 provides FP/BF16 unlock for Nvidia 3000 series

NVIDIA releases on-device small model Nemotron-3.5-Bolt-E4B

NVIDIA releases the on-device small model Nemotron-3.5-Bolt-E4B. A report on October 23, 2026 states that the model can run locally on devices with 6GB of memory, targeting on-device devices such as phones; it is trained based on Gemma 4, supports vision, and Unsloth simultaneously provides the GGUF file and a guide link. NVIDIA says it targets scenarios such as local assistants, interactive characters, and robots that can converse and call tools. A follow-up report on October 24 states that the model has 8B total parameters, 4B effective parameters, and a 64K context, and is post-trained using only reinforcement learning based on Google Gemma 4 E4B; the Q4_K_M version occupies 3.1 to 4.3 GB of VRAM on RTX and can run on GeForce RTX, Jetson, and DGX Spark. The report also states that in NVIDIA's own testing, it leads Gemma 4 E4B, Qwen3.5-4B, and Nemotron 3 Nano 4B in 15 of 21 benchmarks, but Tau2-Bench is 55.1, trailing Qwen3.5-4B's 75.1.

NVIDIA launches DSX POD modular AI factory reference design

On October 21, 2026, NVIDIA unveiled the DSX POD modular AI factory reference design at the GTC conference in Berlin, targeting Vera Rubin AI factories. The design is integrated and validated by partners inside modular containers before delivery, moving installation work such as compute, networking, cabling, and liquid cooling forward into the manufacturing stage. NVIDIA's press release states that the solution can shorten deployment time at a ready site from six months to as little as 14 days; a same-day report by Zhidongxi mentioned that Jensen Huang said adopting the solution can compress the time from compute arriving on site to going into production from 4 to 5 months to about 1 month. Partners include Dell, HPE, Lenovo, Foxconn (Ingrasys), QCT, and others, with global deliveries planned for early 2027. There are currently no subsequent developments.

NVIDIA launches DSX POD modular AI factory reference design

On October 21, 2026, NVIDIA unveiled the DSX POD modular AI factory reference design at the GTC conference in Berlin, targeting Vera Rubin AI factories. The design is integrated and validated by partners inside modular containers before delivery, moving installation work such as compute, networking, cabling, and liquid cooling forward into the manufacturing stage. NVIDIA's press release states that the solution can shorten deployment time at a ready site from six months to as little as 14 days; a same-day report by Zhidongxi mentioned that Jensen Huang said adopting the solution can compress the time from compute arriving on site to going into production from 4 to 5 months to about 1 month. Partners include Dell, HPE, Lenovo, Foxconn (Ingrasys), QCT, and others, with global deliveries planned for early 2027. There are currently no subsequent developments.

View story:NVIDIA launches DSX POD modular AI factory reference design

NVIDIA showcases full lineup of local AI hardware products at GTC Berlin

On October 21, 2026, NVIDIA showcased three local AI hardware lines—RTX Spark, DGX Spark, and DGX Station—at GTC Berlin, and announced multiple software updates. Official information shows that DGX Spark 64GB goes on sale on October 23; NVIDIA Sync will support clusters of up to eight DGX Sparks and will add Model Launcher this month; DGX Station for Windows is expected to launch in December; the Recursive Flint local coding model is released with open-source Apache 2.0 weights. A follow-up report on October 23 stated that DGX Spark officially supports eight-node clusters, with unified memory expandable to 1TB, and can locally run frontier models such as the full Z.ai GLM 5.3 (753B); official benchmarks show that Qwen3.8-27B (NVFP4) decoding throughput increases from about 12 tokens/s on a single node to 52 tokens/s on eight nodes, with eight nodes achieving a 4.43x decoding speedup, and nanochat pretraining achieving a 7.79x speedup with 97.4% scaling efficiency. NVIDIA Sync's eight-node cluster support will be available before the end of the year.

NVIDIA showcases its full line of local AI hardware products at GTC Berlin

On October 21, 2026, NVIDIA showcased three local AI hardware lines—RTX Spark, DGX Spark, and DGX Station—at GTC Berlin, and announced multiple software updates. Official information shows that DGX Spark 64GB goes on sale on October 23; NVIDIA Sync will support clusters of up to eight DGX Sparks and will add Model Launcher this month; DGX Station for Windows is expected to launch in December; the Recursive Flint local coding model is released with open-source Apache 2.0 weights. A follow-up report on October 23 stated that DGX Spark officially supports eight-node clusters, with unified memory expandable to 1TB, and can locally run frontier models such as the full Z.ai GLM 5.3 (753B); official benchmarks show that Qwen3.8-27B (NVFP4) decoding throughput increases from about 12 tokens/s on a single node to 52 tokens/s on eight nodes, with eight nodes achieving a 4.43x decoding speedup, and nanochat pretraining achieving a 7.79x speedup with 97.4% scaling efficiency. NVIDIA Sync's eight-node cluster support will be available before the end of the year.

View story:NVIDIA showcases full lineup of local AI hardware products at GTC Berlin

NVIDIA releases OpenShell reference architecture

On October 21, 2026, NVIDIA released the OpenShell reference architecture for building, isolating, observing, and enforcing secure agents on enterprise infrastructure and edge systems; the architecture is now available. The architecture uses Vera Rubin to run model inference, Vera CPU to host the agent harness, and BlueField-4 DPU to run Sentry for chip-level policy enforcement; Sentry can isolate out-of-bounds agents at millisecond latency. On the same day, NVIDIA released Sentry, a security software framework that monitors AI agents and enforces access policies within a hardware-isolated trust domain on the BlueField-4 DPU, without relying on host software; the framework is part of the OpenShell reference architecture announced at GTC Berlin and is planned to be opened early next year. Subsequently, Red Hat and Nvidia expanded their collaboration, planning to integrate the OpenShell reference architecture as a native capability into its AI platform, and to provide runtime security controls, model access governance, and hardware support for autonomous AI agents through Red Hat AI Factory with Nvidia.

NVIDIA releases OpenShell reference architecture

On October 21, 2026, NVIDIA released the OpenShell reference architecture for building, isolating, observing, and enforcing secure agents on enterprise infrastructure and edge systems; the architecture is now available. The architecture uses Vera Rubin to run model inference, Vera CPU to host the agent harness, and BlueField-4 DPU to run Sentry for chip-level policy enforcement; Sentry can isolate out-of-bounds agents at millisecond latency. On the same day, NVIDIA released Sentry, a security software framework that monitors AI agents and enforces access policies within a hardware-isolated trust domain on the BlueField-4 DPU, without relying on host software; the framework is part of the OpenShell reference architecture announced at GTC Berlin and is planned to be opened early next year. Subsequently, Red Hat and Nvidia expanded their collaboration, planning to integrate the OpenShell reference architecture as a native capability into its AI platform, and to provide runtime security controls, model access governance, and hardware support for autonomous AI agents through Red Hat AI Factory with Nvidia.

View story:NVIDIA releases OpenShell reference architecture

NVIDIA releases on-device small model Nemotron-3.5-Bolt-E4B

NVIDIA releases the on-device small model Nemotron-3.5-Bolt-E4B. A report on October 23, 2026 states that the model can run locally on devices with 6GB of memory, targeting on-device devices such as phones; it is trained based on Gemma 4, supports vision, and Unsloth simultaneously provides the GGUF file and a guide link. NVIDIA says it targets scenarios such as local assistants, interactive characters, and robots that can converse and call tools. A follow-up report on October 24 states that the model has 8B total parameters, 4B effective parameters, and a 64K context, and is post-trained using only reinforcement learning based on Google Gemma 4 E4B; the Q4_K_M version occupies 3.1 to 4.3 GB of VRAM on RTX and can run on GeForce RTX, Jetson, and DGX Spark. The report also states that in NVIDIA's own testing, it leads Gemma 4 E4B, Qwen3.5-4B, and Nemotron 3 Nano 4B in 15 of 21 benchmarks, but Tau2-Bench is 55.1, trailing Qwen3.5-4B's 75.1.

View story:NVIDIA releases on-device small model Nemotron-3.5-Bolt-E4B

MiniMax thanks NVIDIA NeMo for day-0 fine-tuning support for M3.1-Flash

On October 23, 2026, MiniMax publicly thanked the NVIDIA NeMo team, expressing gratitude for providing day-0 fine-tuning support for MiniMax-M3.1-Flash. According to information officially released by MiniMax, the cited post by Huiying Li states that the model had already processed 200T+ tokens through Space Bunny before its public release, and that the weights are now open. The official statement also mentions that NeMo-AutoModel's day-0 fine-tuning support covers FSDP2, expert parallelism, pipeline parallelism, and context parallelism, can scale to 128 GPUs, and can directly load NVFP4 checkpoints. Currently, this event has only this one official fact, with no subsequent developments.

MiniMax thanks NVIDIA NeMo for day-0 fine-tuning support for M3.1-Flash

On October 23, 2026, MiniMax publicly thanked the NVIDIA NeMo team, expressing gratitude for providing day-0 fine-tuning support for MiniMax-M3.1-Flash. According to information officially released by MiniMax, the cited post by Huiying Li states that the model had already processed 200T+ tokens through Space Bunny before its public release, and that the weights are now open. The official statement also mentions that NeMo-AutoModel's day-0 fine-tuning support covers FSDP2, expert parallelism, pipeline parallelism, and context parallelism, can scale to 128 GPUs, and can directly load NVFP4 checkpoints. Currently, this event has only this one official fact, with no subsequent developments.

View story:MiniMax thanks NVIDIA NeMo for day-0 fine-tuning support for M3.1-Flash

NVIDIA releases AIOps platform Minerva

On October 22, 2026, NVIDIA released Minerva, an AIOps platform for AI factories. According to NVIDIA's official blog, the platform collects telemetry data from infrastructure and AI workloads, then uses dedicated AI models to turn the data into actionable insights; it builds a dynamic knowledge graph that maps GPUs, SuperNICs, switches, cables, and the logic of the jobs running on them, analyzes billions of data points per hour, learns normal behavior, detects unseen anomalies, and predicts failures. The platform includes a built-in AI assistant that can connect to external agent tools and supports troubleshooting cluster health and providing remediation suggestions in natural language. IREN, IBM, Jio, Lambda, Tesla, and others plan to deploy it, while Dell, Lenovo, and Supermicro will incorporate it into their AI factory operations product portfolios. On the same day, Unite.AI reported that Minerva includes a multi-agent AI assistant named Sherlock and provides an MCP server, and that IREN, Bitdeer AI, and others are already using it. There are currently no subsequent developments.

NVIDIA releases AIOps platform Minerva

On October 22, 2026, NVIDIA released Minerva, an AIOps platform for AI factories. According to NVIDIA's official blog, the platform collects telemetry data from infrastructure and AI workloads, then uses dedicated AI models to turn the data into actionable insights; it builds a dynamic knowledge graph that maps GPUs, SuperNICs, switches, cables, and the logic of the jobs running on them, analyzes billions of data points per hour, learns normal behavior, detects unseen anomalies, and predicts failures. The platform includes a built-in AI assistant that can connect to external agent tools and supports troubleshooting cluster health and providing remediation suggestions in natural language. IREN, IBM, Jio, Lambda, Tesla, and others plan to deploy it, while Dell, Lenovo, and Supermicro will incorporate it into their AI factory operations product portfolios. On the same day, Unite.AI reported that Minerva includes a multi-agent AI assistant named Sherlock and provides an MCP server, and that IREN, Bitdeer AI, and others are already using it. There are currently no subsequent developments.

View story:NVIDIA releases AIOps platform Minerva

Aravind Srinivas shares JOLT post-training research in collaboration with NVIDIA

Aravind Srinivas shared JOLT, a post-training method proposed by Perplexity in collaboration with NVIDIA. The method lets the same model serve as both teacher and student, and trains the teacher to learn from rewards from its own attempts while penalizing drift away from the student's behavior. On the MATH benchmark, JOLT reaches accuracy comparable to GRPO with 6.5 times fewer training tokens; JOLT+ additionally provides outcome-based rewards to the student model and clearly surpasses GRPO on math, coding, terminal tasks, and tool-use benchmarks. Currently this research is being shared publicly by Aravind Srinivas, with no further follow-up progress yet.