iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
The EE Times GTC 2026 podcast describes an emerging NVIDIA-centered AI stack: accelerator systems, faster links between them, software agents, and robotics. Its most useful takeaway is how these pieces may fit together—and what remains unanswered. The conversation is an interview recap, not independent verification of product specifications, security performance, or market forecasts.
In episode five of AI with Sally, published March 20, 2026, host Sally Ward-Foxton speaks with Tirias Research principal analyst Jim McGregor at NVIDIA GTC in San Jose. They discuss Groq technology in NVIDIA systems, SpectrumX and optical networking, NemoClaw and OpenClaw-style agents, infrastructure costs, and robotics. The episode offers context about the direction of the ecosystem, but not a technical product comparison or deployment guide.
What McGregor says Groq V3 LPX is—and what is still unclear
McGregor describes Groq V3 LPX as a system-level offering that integrates Groq technology beyond the chip itself. He says systems were planned for release in Q3 and that Samsung was producing the chip. These are statements made in the interview; the episode does not independently verify the production status or release timing.
Recommended Free Tools
The conversation does not explain what changed from V2 to V3. McGregor says, “We did not get a clue on what the difference is and what happened to two.” It therefore leaves V3’s specifications and the V2-to-V3 story unresolved.
#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
At the system level, the speakers describe NVLink connecting NVIDIA and Groq and a modified, low-latency SpectrumX link between chassis. They also discuss Groq inference hardware and Rubin CPX as potentially coexisting in a broader rack-level value chain. The episode provides no configuration diagram, performance data, or independently checked deployment specification, so these descriptions should not be read as a confirmed rack design.
How SpectrumX and co-packaged optics fit into the discussion
McGregor describes SpectrumX as connectivity between chassis and says NVIDIA discussed co-packaged optics for both scale-up and scale-out designs. In this context, scale-up concerns connections within a larger system, while scale-out concerns connecting systems across a wider deployment.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
The interview emphasizes that putting optics inside the rack has cost implications, but supplies no detailed product specifications, pricing, or firm deployment schedule. Its point is architectural: moving data between components and systems is part of the infrastructure challenge, not an implementation detail with a cost-free answer.
What NemoClaw is meant to do for OpenClaw-style agents
McGregor characterizes OpenClaw as a tool for building agents that can work locally with a user’s information and data. He raises a practical risk: an agent may exceed its intended bounds or lose track of rules that were installed to constrain it.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
He presents NemoClaw as NVIDIA’s security layer or wrapper for bounding those agent deployments and mentions Nemotron as a possible supporting model. That is an interview description, not a demonstrated security guarantee. The episode includes no test results showing that NemoClaw prevents particular failures, nor does it establish the limits of its protections.
The broader concern is trust as agents gain the ability to use other agents. McGregor describes a future with agents assigned to different functions and agents calling on other agents, saying, “It’s likely to be multiple agents for different functions, different things. And it’s also going to be agents using other agents.” The conversation frames trust as a hurdle to adoption, rather than offering an implementation checklist.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
The economics: lower cost per token versus high upfront spending
The business case discussed is reducing cost per token through improved efficiency, throughput, and latency. That goal sits alongside substantial upfront costs for chips, systems, racks, and broader infrastructure. McGregor sums up that trade-off simply: “So, it’s a costly thing.” The interview does not provide a cost model or measured savings, so it cannot establish whether a particular deployment would lower total costs.
Free tools Windows power users keep installed
One-click scans. No signup required.
McGregor also cites a market estimate of $500 billion by the end of 2026 and a $1 trillion opportunity by the end of 2027. The transcript does not identify who produced the underlying forecast or how it was calculated; these are figures he discussed during the interview, not independently established market totals.
Best Value
- Powered by the NVIDIA Blackwell architecture and DLSS 4 OC mode: 2640MHz/Default mode: 2610MHz (Boost Clock)
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.125-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
Robotics spans edge constraints and NVIDIA’s simulation ecosystem
The speakers range from industrial robots to humanoids. McGregor stresses power constraints and suggests complex robots may use multiple control units and sensors. That highlights a different design pressure from centralized data-center compute: a robot must work within the power, sensing, and control requirements of its physical platform.
McGregor points to NVIDIA’s Cosmos, Isaac Sim, and a model he calls Root as elements of a software and simulation ecosystem. The conversation does not compare robot products or recommend a particular development kit. Ward-Foxton mentions 110 robots on the GTC floor, but the episode does not independently verify that count.
What the podcast does—and does not—establish
The episode is useful as a snapshot of how its speakers connect accelerator infrastructure, networking, agent controls, and robotics. It is not a product benchmark, security assessment, or buyer’s guide. One number raised by the host illustrates why attribution matters: Ward-Foxton asks whether Groq might account for 25% of a data center, but McGregor does not confirm that figure. It should not be treated as a forecast or established configuration.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchFor the full interview and transcript, see EE Times’ “GTC 2026 Review: NemoClaw, Groq, and SpectrumX”, published March 20, 2026.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
