Bloomberg Tech at CES, Day 2

Live from CES in Las Vegas, we teed up a fast hour on chips, robots, and streaming—Intel’s push to get competitive in PCs, media deal drama with QuickPlay’s Paul Pastor, and Mobileye’s big move into humanoids, with markets holding up despite geopolitics.

Tech names are steady, but the buzz is around Nvidia’s outlook and a rip in memory; investors are weighing China headlines on H200 against talk of stronger demand into Blackwell and Rubin.

Everyone wants to know if this is a bubble; the takeaway is demand keeps broadening as new models and sectors jump in, while memory has swung into scarcity, pushing prices higher.

We’re deepening a broad partnership with Siemens across EDA, simulation, and factory automation so we can design and run more complex systems faster; Rubin class GPUs will be dramatically more energy and cost efficient, but the design challenge is immense, which is why digital twins and accelerated tooling matter.

The edge is where this gets real—low‑latency inference brings physical AI to controllers and industrial PCs, turning cloud‑trained models into shop‑floor gains.

Intel’s at CES with fresh silicon, so what actually changed?

It’s been a long road, but on 18A we combined last year’s top performance with last year’s best efficiency to run the same workloads at much lower power, and early customer feedback is that it feels powerful while sipping battery.

AMD and Qualcomm claim the same; what proves you’re in front?

For mobile gamers, our Series 3 GPU with AI‑driven multi‑frame generation delivers notably higher frame rates and smoother play—I’ll stack that up against anyone.

Does the AI PC pitch actually move the market in 2026?

Customers keep asking us to ramp supply, and we’re seeing real on‑device AI demand from partners like Perplexity for performance, privacy, cost, and control; Series 3 is also pulling in day‑one edge orders for vision and robotics control.

Why your own foundry and what’s the benefit?

18A brings EUV, gate‑all‑around, and backside power to separate power and signals for better performance per watt and denser chips, which helps OEMs build slimmer systems with bigger batteries and longer life.

Memory’s tight; how are you handling pricing and the PC outlook?

Customers haven’t settled on how to offset memory costs, but they want us on plan because they see share wins on Series 3; we’re modeling the market roughly flat and ready to flex.

Cristiano, you’re calling robotics the next big AI wave—why now?

Robotics is inherently an edge AI problem with tight power and sensor integration, and we see industrial use cases ramping as early as 2026; full home humanoids will take longer, much like assisted driving arriving before robotaxis.

Sam, the floor is packed with humanoids—how close are we?

Live demos are improving, but dexterity and speed are still hurdles and homes are unpredictable; meanwhile health tech is exploding into new form factors, from a saliva‑aware toothbrush to a ‘longevity’ mirror and a smart night guard that tracks sleep.

Amnon, why take Mobileye from autonomous driving into humanoids?

Humanoids are a natural extension of physical AI and a long‑term growth path that leverages our vision, software, and talent; related‑party concerns were handled with recusal and my son’s role is immaterial.

What’s the go‑to‑market for Mente Robotics?

Phase one targets structured sites—warehouses, assembly, retail—where fleets can be customized and built for about twenty thousand dollars per unit at scale; phase two is homes, using a ‘mentoring’ approach where the robot learns a new task from video, trains in the cloud, then executes.

Nvidia showed a full autonomy stack; how do you stack up on cost and readiness?

Competition is healthy; we’re close to driverless with Volkswagen, aiming to scale to one hundred thousand robotaxis over eight years, with a lower‑cost sensor suite and compute and best‑in‑class imaging radar.

And that U.S. automaker deal?

It’s a roughly nine‑million‑vehicle safety and convenience program that consolidates multiple ECUs into one platform, alongside Volkswagen bringing the total to about nineteen million units in the past year.

Quick pivot to media M&A: which deal setup really serves viewers—streaming carve‑outs or full takeovers?

Linear still throws off cash, but the center of gravity moved to streaming and creator platforms; a Netflix‑Warner pairing would blend top‑tier tech with elite storytelling, though value also exists in a broader studio combination.

What’s the real risk while these giants negotiate?

Distraction—while you integrate org charts, TikTok, YouTube, and creators are sprinting ahead, so you can finish a deal only to find the market has shifted under you.

How does AI reshape the viewing experience this year?

Expect conversational onboarding and a unified data fabric that personalizes from the first touch and coordinates across apps and socials to drive acquisition, engagement, and retention.

For those new to QuickPlay?

We build white‑label OTT backends end‑to‑end, now infused with AI to reformat content, respond to social signals, and deliver hyper‑personalized experiences at enterprise scale.

Hemanth, on AI valuations—are prices justified or hype?

Big research efforts have uncertain business timing, while applied AI can grow into seemingly rich prices; the real question is durability on top of rapidly improving models, with standouts like Anthropic and developer‑tooling use cases looking resilient.

So, should companies like Anthropic head to the public markets soon?

They could in the next year or so if the business is predictable enough, but growth has been so volatile that timing will depend on how the market settles.

Discord quietly filed to go public—what matters here?

It’s grown beyond gaming chat to two hundred million monthly users and brought in a seasoned games executive to sharpen monetization, adding momentum to a warming tech IPO pipeline.

That’s it for this edition of Bloomberg Tech from CES—more guests tomorrow, from air taxis to policy voices.

Vegas has been a ride; we’ll keep the coverage rolling and post the recap on the pod—you know where to find it.