THE APEX TIMES
NVIDIA rolls out new Jetson Thor modules aimed at bringing humanoid and edge AI robots to scale
The company introduced the Jetson T3000 and T2000, two new compact computers built on its Thor architecture, designed to run foundation-model workloads closer to where robots operate, with planned software tools to streamline memory use and deployment.
NVIDIA on Wednesday announced two new edge-computing modules under its Jetson Thor platform, positioning the devices as a path for robotics companies to move from prototypes to mass deployments that rely on “foundation model” intelligence at the device level. The new Jetson T3000 and T2000 are intended for general-purpose robots and autonomous machines that need high compute in a small power envelope, without requiring every decision to be processed in the cloud.
At the center of NVIDIA’s pitch is the Jetson AGX Thor platform, which the company said is already seeing growing adoption across industries and is being used as the hardware backbone for next-generation humanoids and robotic systems. NVIDIA named a broad set of companies building on the platform, including 1X, Agility, Agile Robots, Amazon Robotics, Boston Dynamics, FANUC, Hitachi, Medtronic and Techman Robot.
The Jetson T3000 module is aimed at performance-rich robotics and other edge AI applications. NVIDIA said it delivers 865 FP4 teraflops of AI compute in a compact form factor that is roughly half the size and power of the company’s earlier T5000 module. The company also said the T3000 pairs an NVIDIA Blackwell GPU with an eight-core Arm Neoverse CPU, 32GB of LPDDR5X memory, 273GB/s of memory bandwidth and 25 GbE connectivity. For safety-oriented robotics, NVIDIA also described an IGX T3000 variant that includes integrated functional safety and supports NVIDIA Halos for Robotics, a software stack intended to help robots operate alongside humans.
Despite the smaller footprint, NVIDIA said the T3000 can match the T5000’s inference performance for multimodal workloads. The company cited the kinds of models robots typically use, including large language models, vision-language models, vision-language-action models and “world foundation models.” NVIDIA added that migrating to the T3000 can reduce total costs, pointing to high memory prices as a driver for why lower-memory configurations may matter commercially.
The Jetson T2000 expands the Thor architecture to a wider range of edge AI systems. NVIDIA said it offers 400 FP4 teraflops of compute and 16GB of memory, targeting what it called an entry point for developers building visual AI agents, autonomous mobile robots, industrial manipulators and other intelligent machines.
NVIDIA also tied the new hardware to software tools meant to reduce friction for developers. The company said “Jetson agent skills” are newly released, designed to optimize the software stack and achieve meaningful memory savings in days rather than weeks. NVIDIA described the goal as helping developers run more capable workloads on lower-memory configurations, which it said can translate into lower system cost, faster deployment and the ability to move down one memory SKU within the same performance tier without giving up performance.
To illustrate the impact, NVIDIA pointed to case studies in which teams reduced memory usage enough to shift to smaller Jetson configurations. It cited humanoid robotics and related hardware: UBTech and Agile Robots, and an industrial solutions provider, Connect Tech, which NVIDIA said reduced memory usage by up to 15GB, enabling a move from a 64GB Jetson AGX Orin module to the 32GB Jetson configuration. In smart retail, SandStar was said to reduce memory usage by up to 4GB, enabling deployment on a Jetson Orin NX 8GB module instead of a 16GB configuration. For companion robotics, GROOVE X, maker of the LOVOT robot, was cited for using heterogeneous AI accelerators to optimize workload distribution and reduce memory. NVIDIA also cited intelligent transportation work at NoTraffic, which it said reduced memory usage by 30% on a Jetson TX2 NX platform.
Beyond memory optimization, NVIDIA said Jetson is “agentic-ready” for physical AI, highlighting the idea of on-device reasoning, autonomous decision-making and task automation. The company referenced NVIDIA NemoClaw blueprints for orchestrating intelligent agents, and it said the Jetson portfolio will support a broader path for developers building embodied AI systems.
A related software and model announcement accompanied the hardware. NVIDIA expanded its NVIDIA Cosmos 3 “frontier open world foundation model” family with Cosmos 3 Edge, described as a lightweight, 4-billion-parameter robot foundation model built for embodied systems. NVIDIA said Cosmos 3 Edge is compatible with NVIDIA Thor platforms and is intended to support real-time on-device inference for seeing the world, reasoning over it and predicting and generating actions, including the ability to post-train with specific embodiments and sensors using the open Cosmos framework in about a day. NVIDIA also said the new Jetson modules will run as part of the Thor family’s shared chip architecture and software stack, including tools such as NVIDIA Isaac for robotics simulation and perception. NVIDIA added that developers can begin using T3000 emulation mode later this month with JetPack 7.2.1, with T2000 emulation mode to follow in a future release, and that the Jetson T3000 and T2000 modules are scheduled for availability in Q1 2027.
Why It Matters
- Edge AI for robotics often runs into cost and power constraints, so compact modules built for running multimodal and foundation-model workloads on-device could accelerate robot deployments outside labs.
- If NVIDIA’s memory-optimization tools work as described, they may reduce the amount of hardware needed to run common model pipelines, improving unit economics for robot makers.
- By pairing hardware releases with an emulation-to-migration pathway, NVIDIA is indicating an attempt to lower software integration risk for developers moving to Thor.
- The Cosmos 3 Edge update suggests NVIDIA wants to link robot-specific foundation models with the same platform strategy, potentially tightening the hardware-software-model stack for embodied AI.
Key Facts
- NVIDIA introduced Jetson T3000 and Jetson T2000 modules based on the Thor architecture to support mainstream robotics and edge AI.
- Jetson T3000 is rated at 865 FP4 teraflops, pairs a Blackwell GPU with an eight-core Arm Neoverse CPU, includes 32GB LPDDR5X memory, 273GB/s memory bandwidth, and 25 GbE connectivity; NVIDIA also described an IGX T3000 safety variant.
- NVIDIA said the T3000 can match T5000 inference performance for multimodal workloads and that migrating may reduce costs amid high memory prices.
- Jetson T2000 is positioned as an entry point with 400 FP4 teraflops and 16GB of memory for agents, mobile robots and industrial manipulators.
- NVIDIA said Jetson agent skills are newly released to optimize the software stack and save memory in days rather than weeks, enabling lower-memory deployments.
- NVIDIA cited examples of memory reductions that helped teams move to smaller Jetson configurations across humanoid robotics, smart retail, companion robotics and intelligent transportation.
- Emulation for T3000 is planned with JetPack 7.2.1 later in July, T2000 emulation is expected later, and both modules are scheduled for Q1 2027.
Technology Related
SBA expands Palantir use for fraud detection in pandemic-era relief programs, per report
The U.S. Small Business Administration says it is broadening adoption of Palantir Technologies software to support a nationwide fraud detection push tied to pandemic-era funding. The report also points to fresh alliances that could expand Palantir’s government footprint.
Broadcom’s AVGO rally revives a “fair value” debate after an 8x five-year run
After a large gain over the past five years, Broadcom shares are again drawing attention as investors weigh whether the latest deal-driven optimism is already reflected in the valuation.
ELVA shares surge to best day in more than 13 years as investors weigh an Amazon-related pact
Roth Capital’s Craig Irwin urged investors to look at Electrovaya’s agreement with Amazon alongside a comparable energy storage push tied to Ford and EDF.
Nvidia’s first H200 shipments to China may not move results, but they could matter to expectations
A reported opening in Nvidia’s China supply chain for its H200 data-center AI chip is unlikely to change near-term quarterly numbers by itself, but it arrives at a sensitive moment as investors focus on whether recent guidance fully captures export constraints.
Apple shares jump after China approval for “Apple Intelligence” using Alibaba’s Qwen model
Apple (AAPL) rose sharply in afternoon trading after a report said regulators cleared the company to launch its “Apple Intelligence” features in China, tying the rollout to Alibaba’s Qwen artificial intelligence model.
Japan builds industry-focused AI using NVIDIA’s Nemotron open models, company says
NVIDIA said Japanese enterprises, startups and research institutions are using its Nemotron open models to develop more specialized applications, highlighting a growing push toward domain-specific AI systems.
Oracle and DocuSign shares jumped after softer inflation data calmed budget-squeeze fears
Stocks in software and cloud-connected businesses rose in the afternoon session after a Producer Price Index reading came in soft, easing worries that corporate buyers would tighten spending. Investors also appeared to look past a fresh round of concerns sparked by IBM earlier.
Buffett’s take on Alphabet: Berkshire bet supports a bullish bar set for Wall Street
Warren Buffett said Alphabet, whose shares trade as GOOGL, can outperform most Wall Street stock picks over the next year, a comment that coincided with a sharp jump in the stock as Berkshire’s stake was described as topping $31 billion.
European cloud and hosting groups ask EU antitrust regulators to scrutinize Broadcom’s VMware licensing changes
Trade association CISPE and other industry players have urged the European Commission to examine whether Broadcom’s post-acquisition VMware licensing adjustments could disadvantage competitors in cloud services and infrastructure.
Jefferies flags Amazon as a ‘top pick’ among hyperscalers, pointing to AI-infrastructure exposure at a lower stock price
A brokerage argument for Amazon centers on the idea that investors can get exposure to artificial-intelligence infrastructure and retail operations at a valuation level it views as more attractive than two major peers.