Nvidia Storms Two Conferences at Once: Physical AI Meets Consumer Desktop Ambitions

Nvidia unveils RTX Spark AI chip at Computex, targets edge computing and robotics at CVPR, expanding beyond data center dominance. Stock dips 2.94% but up 9.53% over 30 days.

Nvidia leveraged back-to-back appearances at industry events this month to project a broader vision than its core data center business. At the CVPR 2026 conference in Denver, the company showcased tools for robotics and autonomous driving. Days later at Computex in Taipei, it unveiled a consumer AI chip that jolted competitors. The message is clear: Nvidia intends to dominate computing from the cloud to the edge — and right onto the desktop.

The market's reaction has been mixed. In European trading, Nvidia shares slipped 2.94% to €185.96, though the stock remains up 9.53% over the past 30 days. The pullback leaves the shares 15.37% above their 200-day moving average but 8.17% below the 52-week high. In Frankfurt, the stock added 0.27% to €192.12 on the day of the Computex announcement. Year to date, Nvidia has gained roughly 19%, a more measured pace than the 54% rally over the past twelve months.

Rivals Feel the Heat from RTX Spark

The Computex keynote delivered the biggest shock to Nvidia's competitors. The RTX Spark, a system-on-chip co-developed with MediaTek, combines a Blackwell GPU with a MediaTek CPU core and unified memory, enabling large language models to run locally without cloud dependency. Nvidia claims 1 petaflop of FP4 performance. First laptops from Microsoft, Dell, HP, ASUS, Lenovo and MSI are due this year, with Morgan Stanley pegging starting prices at $2,899 for the N1X variant and $1,799 for the lower-tier N1.

Intel fell 4.7% to $109.33 on Monday, while AMD and Qualcomm also declined — Qualcomm dropped 8.8%. Nvidia itself rose 6.3%. Arm Holdings surged 16% on licensing windfalls from any ARM-based chip, including the RTX Spark. MediaTek gained over 5% in Taipei trading. Yet the annual performance tells a different story: AMD is up 130% year to date and Intel roughly 200%, partly on lower bases.

Physical AI Gains Traction Beyond the Data Center

At CVPR, Nvidia shifted the narrative from pure compute power to physical world applications. The company demonstrated open agent frameworks for Omniverse, Cosmos, Alpamayo and Metropolis, aiming to translate industrial workflows into tasks AI agents can execute. For autonomous driving, developers can port fleet data into simulations, generate photorealistic scenarios and expand closed-loop training without needing to collect every edge case on the road.

Robotics is emerging as a second growth pillar. Nvidia's stack spans Cosmos for physical world understanding, Omniverse for digital twins, Isaac for robot training, and Jetson for edge deployment. Metropolis handles vision AI. The strategy includes open models, code, pre-trained weights and datasets. Among the research highlights: NitroGen for generalist gaming agents, SAGE, FastFoundationStereo, and GraspGen-X — a grasping model trained on 395 million simulated grasps designed to handle unfamiliar objects.

Earnings Show Core Strength, but Edge Gets New Billing

For now, the data center remains the profit engine. In the first quarter of fiscal 2027, Nvidia posted $81.6 billion in revenue, up 85% year over year. Data Center alone contributed $75.2 billion, a 92% jump. Second-quarter guidance calls for $91.0 billion, plus or minus 2%.

A new reporting structure now separates Data Center and Edge Computing, making the second path more visible. Edge Computing revenue reached $6.4 billion, encompassing robotics, automotive, workstations, PCs, gaming consoles and AI-RAN base stations. That segment could become a significant growth driver as physical AI and local inference scale.

Analysts are cautious but not dismissive of the PC push. Seaport Research's Jay Goldberg called the RTX Spark technically promising but warned that Windows on ARM still lags behind x86, predicting adoption will take "multiple generations." Truist's William Stein reiterated a buy rating with a $307 target, and Cathie Wood bought 300,017 Nvidia shares following the Computex keynote. CEO Jensen Huang also confirmed that the next-generation Vera Rubin architecture is in full production, with first systems headed to Anthropic, OpenAI, xAI, Dell, Oracle and CoreWeave this autumn.

The dual offensive — physical AI at CVPR and consumer AI at Computex — underscores Nvidia's ambition to build a platform that spans data centers, factories, cars and laptops. The market is weighing high expectations for AI infrastructure against these new growth fields. For Nvidia, the real test will come not from chip upgrades alone, but from how deeply the ecosystem is integrated into real-world applications.