# NVIDIA GTC 2026: what was announced, what ships, and what is roadmap

- **Published:** 2026-03-20
- **Updated:** 2026-10-12
- **Category:** Hardware & Infrastructure
- **Author:** Sinthetix
- **URL:** https://sinthetix.com/blog/nvidia-gtc-2026-what-was-announced

> NVIDIA's GTC 2026 announcements sorted into available now, partner availability in late 2026, and roadmap, with performance claims marked as vendor-stated.

NVIDIA's GTC conference ran March 16 to 19 in San Jose. Most of the headline news landed in the first-day keynote and a batch of same-day press releases. This analysis sorts what was announced into three groups: available now, announced with partner availability later in 2026, and roadmap. Nearly every performance figure comes from NVIDIA itself and has no independent test behind it yet.

**Key takeaways**

- NVIDIA now describes Vera Rubin as seven chips, including a newly integrated Groq 3 LPU, packaged as five rack types.
- NVIDIA says Vera Rubin is "in full production", but partner products are expected from the second half of 2026.
- Software was the part buyers can use today: Dynamo 1.0 is available now, and NemoClaw was formally announced.
- Kyber racks and the Feynman architecture are roadmap items with no firm NVIDIA specifications.

## The platform: seven chips, five racks

At the keynote, NVIDIA presented Vera Rubin as seven chips: the Vera CPU, Rubin GPU, NVLink 6 switch, ConnectX-9 SuperNIC, BlueField-4 DPU, Spectrum-6 Ethernet switch and a newly integrated Groq 3 LPU [1]. They are packaged into five rack types [1]:

- **Vera Rubin NVL72:** 72 Rubin GPUs and 36 Vera CPUs. NVIDIA claims large mixture-of-experts models train on one-fourth the GPUs of Blackwell, and up to 10x higher inference throughput per watt at one-tenth the cost per token. Only the training claim names Blackwell as the baseline [1].
- **Vera CPU rack:** 256 liquid-cooled CPUs. Each Vera CPU has 88 custom cores and up to 1.2 TB/s of memory bandwidth [2].
- **Groq 3 LPX:** 256 LPU processors with 128 GB of on-chip SRAM, claimed at up to 35x higher inference throughput per megawatt, with no baseline stated [1].
- **BlueField-4 STX storage:** a reference architecture for the key-value cache data that long-context agents generate, claimed at up to 5x more tokens per second than traditional storage [3].
- **Spectrum-6 SPX networking:** with co-packaged optics, claimed at up to 5x better optical power efficiency than pluggable transceivers [1].

The Rubin GPU itself, per NVIDIA's technical blog, has 336 billion transistors, up to 288 GB of HBM4 and 50 petaflops of NVFP4 inference compute [10]. The ratings are NVIDIA's and measured its way.

## Demand and the "$1 trillion" line

Jensen Huang said he now sees at least $1 trillion in revenue from 2025 through 2027 [8]. Jon Peddie Research described the same figure as Blackwell and Vera Rubin purchase orders through 2027, double a $500 billion figure from a year earlier [9]. Those are different things, so treat the number as a statement of NVIDIA's expectation, not a reported result.

## What ships, and what does not

| Item | Headline spec or claim (NVIDIA's) | Status as of March 20 |
| --- | --- | --- |
| Dynamo 1.0 | Open-source inference software; up to 7x Blackwell inference gain in "recent industry benchmarks," unnamed [4] | Available today [4] |
| Vera Rubin NVL72, Vera CPU rack | 72 GPUs and 36 CPUs; 256 CPUs per CPU rack [1][2] | "In full production"; partner products from H2 2026 [1][2] |
| BlueField-4 STX | Up to 5x tokens per second vs traditional storage [3] | Partner platforms in H2 2026 [3] |
| Groq 3 LPX | 256 LPUs; up to 35x throughput per MW [1] | H2 2026 per NVIDIA [1]; Q3 2026 per Jon Peddie Research [9] |
| NemoClaw and OpenShell | One-command install of Nemotron models and a policy-controlled sandbox for OpenClaw agents [5] | Announced; page states no maturity stage or date [5] |
| DLSS 5 | AI neural rendering for games [7] | "This fall" [7] |
| Space-1 Vera Rubin Module | Up to 25x the AI compute of an H100 for space use [6] | "At a later date" [6] |
| Vera Rubin Ultra with Kyber racks | Kyber is the rack design after Rubin [9] | Targeted for 2027 per Jon Peddie Research [9] |
| Feynman architecture | Next major architecture with a Rosa CPU and LP40 LPU [8] | No date in NVIDIA's blog; Jon Peddie Research reports 2028 [8][9] |

The "full production" language deserves care. It describes NVIDIA's manufacturing, not what a customer can order. NVIDIA's own release says partner products arrive from the second half of 2026 [1], and one analyst write-up says Rubin "ships to customers later in 2026" [9]. Azure was named as the first hyperscale cloud to power up Vera Rubin NVL72 systems [8], but that is not general availability.

## Other announcements

NVIDIA also announced DSX, a reference design and software for AI-factory power, claiming 30% more infrastructure in a fixed-power data center, and a claimed 100 gigawatts of unlocked grid capacity [1]. On the workstation side, DGX Station with GB300 offers 748 GB of coherent memory and models up to 1 trillion parameters, with first units delivered [8]. IGX Thor became generally available for industrial edge AI [8]. The press kit also lists open model families, robotics and autonomous-vehicle partnerships; we have not evaluated those [8].

## What this means for buyers

1. **Plan on cloud, not purchase orders.** Early Vera Rubin capacity is aimed at hyperscalers and large AI clouds, so most businesses will meet it as instance availability and pricing.
2. **Hold NVIDIA's multipliers loosely.** Claims such as 10x and 35x have unstated or partial baselines, and NVIDIA's own release says features may be offered "when-and-if-available" [1].
3. **Use the software now.** Dynamo is available today [4], and agent runtimes such as OpenShell are worth evaluating, but check their maturity first.
4. **Do not plan around 2027 and 2028 items.** Kyber and Feynman are direction, not products.

> **Why it matters:** The announcements raise the ceiling for large inference clusters, but for most organizations the immediate effect is indirect: cloud providers' capacity plans and per-token prices over the next year.

For most businesses, the practical implication is to track provider pricing and instance availability through the second half of 2026 and to keep model and hosting choices portable, rather than committing to any single vendor's hardware roadmap.

## Sources

1. [NVIDIA Vera Rubin Opens Agentic AI Frontier](https://nvidianews.nvidia.com/news/nvidia-vera-rubin-platform), NVIDIA Newsroom, 2026-03-16
2. [NVIDIA Launches Vera CPU, Purpose-Built for Agentic AI](https://nvidianews.nvidia.com/news/nvidia-launches-vera-cpu-purpose-built-for-agentic-ai), NVIDIA Newsroom, 2026-03-16
3. [NVIDIA Launches BlueField-4 STX Storage Architecture With Broad Industry Adoption](https://nvidianews.nvidia.com/news/nvidia-launches-bluefield-4-stx-storage-architecture-with-broad-industry-adoption), NVIDIA Newsroom, 2026-03-16
4. [NVIDIA Enters Production With Dynamo, the Broadly Adopted Inference Operating System for AI Factories](https://nvidianews.nvidia.com/news/dynamo-1-0), NVIDIA Newsroom, 2026-03-16
5. [NVIDIA Announces NemoClaw for the OpenClaw Community](https://nvidianews.nvidia.com/news/nvidia-announces-nemoclaw), NVIDIA Newsroom, 2026-03-16
6. [NVIDIA Launches Space Computing, Rocketing AI Into Orbit](https://nvidianews.nvidia.com/news/space-computing), NVIDIA Newsroom, 2026-03-16
7. [NVIDIA DLSS 5 Delivers AI-Powered Breakthrough in Visual Fidelity for Games](https://nvidianews.nvidia.com/news/nvidia-dlss-5-delivers-ai-powered-breakthrough-in-visual-fidelity-for-games), NVIDIA Newsroom, 2026-03-16
8. [NVIDIA GTC 2026: Live Updates on What's Next in AI](https://blogs.nvidia.com/blog/gtc-2026-news/), NVIDIA Blog, 2026-03-19
9. [Nvidia GTC 2026 keynote](https://www.jonpeddie.com/news/nvidia-gtc-2026-keynote/), Jon Peddie Research, 2026-03-17
10. [Inside the NVIDIA Rubin Platform: Six New Chips, One AI Supercomputer](https://developer.nvidia.com/blog/inside-the-nvidia-rubin-platform-six-new-chips-one-ai-supercomputer/), NVIDIA Technical Blog

## How Sinthetix helps

An engineering studio for AI agents, automation, and internal tooling. We build the systems that make operators scale.

[Our services](https://sinthetix.com/services) | [Talk to us](https://sinthetix.com/contact)

Sinthetix is an AI engineering studio. We ship agents, automation, and internal tools for operators who need results, not reports. [About Sinthetix](https://sinthetix.com/about)