Jensen All In AI! “Compute Race” to Physical AI: Do You Spot These Highlights & Lists?

If CES 2026 can be summed up in one sentence, it’s this: AI is no longer just about stronger models—it’s truly entering the real world, forcing a full re-pricing of energy, compute, and infrastructure.

NVIDIA opened strong but closed weaker, suggesting capital markets weren’t fully satisfied. But don’t overlook this all-in-AI starting point. Let’s walk through the three biggest CES highlights.

1. From “Single-Card Performance” to “System-Level AI Supercomputing”: NVIDIA Rubin vs. AMD Helios

NVIDIA broke with its CES tradition of launching new consumer GPUs—and went all in on AI.

1) Rubin lifts data-center revenue expectations

If Blackwell pushed the limits of single-GPU performance, Rubin tackles the problem of system-scale deployment. It’s expected to launch in 2H 2026

• Rubin delivers 5× inference performance and 3.5× training performance versus Blackwell

• 10× lower cost, already in mass production

The Rubin platform (including Rubin GPUs + Vera CPUs, six chips in total) will be deployed first by $Amazon.com(AMZN)$ AWS, $Microsoft(MSFT)$ Azure, $Alphabet(GOOG)$ Cloud, $Meta Platforms, Inc.(META)$, and OpenAI, before entering large-scale commercial adoption in the second half of 2026.

2) AMD: Helios takes a direct shot at data-center scale

$Advanced Micro Devices(AMD)$’s presence at CES was noticeably stronger than in previous years. Lisa Su unveiled the latest Helios AI computing rack, expected to launch later this year.

Helios isn’t trying to beat Rubin on “who’s faster.” Instead, AMD chose a different path: open architecture + scale, designed to absorb the compute demand of large models and AI agents.

Helios is AMD’s next-gen data-center rack platform for large-scale AI workloads. Built on the OCP open rack standard, co-developed with Meta

Each Helios rack includes 18,000+ CDNA 5 GPUs and 4,600 Zen 6 CPU cores； Delivers up to 2.9 ExaFLOPS of compute

2. Physical AI: Everything About Robotaxis and Robots

1) Autonomous driving: from “driver assistance” to “thinking systems”

NVIDIA officially launched its end-to-end autonomous driving AI, AlphaMayo—a system that can think and reason.

NVIDIA Drive AV software will debut in the new Mercedes-Benz CLA, delivering L2 end-to-end driving. In demonstrations, the vehicle completed route planning and driving with zero human takeover, successfully reaching its destination.

• AlphaMayo = model layer

• Mercedes-Benz = application layer

When it comes to smart driving, Tesla’s FSD rollout this year is also worth watching.

2) Robotics: the biggest “incremental narrative” at CES 2026

Even before CES, markets were buzzing about robots.

By integrating Omniverse simulation, synthetic data generation, and robot control models, NVIDIA is injecting AI capabilities from the cloud-based “soft world” into the physical “hard world”—cars, factories, and robots.

But $Qualcomm(QCOM)$ delivered the real surprise.

At CES Las Vegas, Qualcomm unveiled a next-generation full-stack robotics architecture, integrating hardware, software, and hybrid AI. Its goal: power everything from small home robots to full-size humanoids.

Qualcomm also launched its latest high-performance robotics processor—the Qualcomm Dragonwing™ IQ10 series. Its positioning is clear: turning robots from “able to move” into “able to work continuously.”

Other robots companies to look:

Robotics software: $PTC Inc(PTC)$, $PALLADYNE AI CORP(PDYN)$, $Mobilicom Limited(MOB)$

Industrial robotics: $Tesla Motors(TSLA)$ $Honeywell(HON)$ $Teradyne(TER)$ $Lincoln(LECO)$

3. When Compute and Physical AI Accelerate Together: Where Does the Power Come From?

U.S. nuclear stocks surged: $Nine Energy Service Inc.(NINE)$ +16%, $NuScale Power(SMR)$ , $Oklo Inc.(OKLO)$ +15%, $NANO Nuclear Energy Inc(NNE)$ +12%, $Energy Fuels(UUUU)$ +11%

Another key Rubin highlight: the architecture supports 45°C warm-water cooling, eliminating chillers and saving 6% of global data-center electricity usage. The stronger the emphasis on efficiency, the clearer it becomes that systems are approaching power limits.

The U.S. Department of Energy’s funding direction—and the collective rally in nuclear stocks—suggest this isn’t just short-term speculation. AI data centers are not cyclical demand; they are multi-year, continuously expanding infrastructure.

Questions to discuss

• CES’s Three Big Themes: Rubin, Physical AI, and Power: Which investment opportunity do you like most?

• Are you ready to all in on NVIDIA, or buying robots or nuclear stocks instead?

• Is Qualcomm’s robotics pivot a signal for a stock re-rating?

• Will 2026 be the breakout year for Tesla’s FSD + Optimus?

Leave your comments to win tiger coins!