June 1, 2026, Taipei Music Center—Jensen Huang, clad in his iconic black leather jacket, took the stage at GTC Taipei. This nearly two-hour keynote, which NVIDIA defined as ushering in "a new era for the PC," marked not only the official launch of the RTX Spark PC processor, codenamed N1X, but also a pivotal assertion from Huang: the age of Agent AI has arrived. As the world’s largest edge computing device, the PC is poised for a fundamental architectural transformation.
The following day, the S&P 500 closed at 7,609.78, with an intraday high of 7,620.90—its first-ever close above 7,600. The Nasdaq finished at 27,093.90, also setting a new record. By the close, the S&P 500 had notched its 24th record high of 2026 and extended its winning streak to nine consecutive sessions.
The AI PC narrative has become one of the primary drivers behind the broader market rally. This article explores three key questions: Why does Jensen Huang say the PC is being "redefined"? What are the core technologies and market logic behind the N1X chip? And how will the AI PC narrative reshape competition across the chip and PC industry value chain?
Keynote Highlights: The PC Is Being Redefined
The Terminal Logic of the Agent AI Era
In his keynote, Huang explained that AI is transitioning from cloud-based training and inference to the era of Agent AI. The essence of Agent AI is that AI is no longer just passively responding to commands—it can understand context, interpret intent, reason, use tools, access memory, and complete multi-step tasks. This computational paradigm applies not only to cloud data centers but also to PCs, autonomous vehicles, robots, and other edge devices.
The critical takeaway: PCs must evolve from traditional tools into "Agentic Computers." Huang likened the future PC to a personal intelligent assistant—not just a passive device waiting for user input, but an active, interactive terminal that helps users accomplish complex tasks.
Why Current PC Architectures Can’t Support Agent AI
Huang further explained the hardware requirements for Agent AI at the architectural level. Traditional CPUs were designed for humans who operate in a "second-level world," while AI Agents operate in a "nanosecond-level world," where latency is extremely sensitive. If Agents need to control various applications, those programs must respond instantly—something serial CPU architectures alone cannot deliver.
This means future PCs must integrate GPUs, Tensor Cores, unified memory, and comprehensive accelerated computing capabilities. Huang called this "the first complete redesign of the PC industry in 40 years," forming the rationale for NVIDIA’s entry into the PC processor market and the launch of RTX Spark.
"Compute Is Revenue": Extending AI Economics to the Edge
Huang introduced a recurring economic principle: "Compute is revenue, compute is profit." He explained that in AI factories, electricity is the limiting resource, so the number of tokens generated per unit of power directly sets a company’s revenue ceiling. This logic has already been validated at the data center level—NVIDIA’s data center business has seen rapid growth in recent quarters—and now NVIDIA aims to extend this to PCs, making personal computers nodes for AI compute generation rather than mere consumption endpoints.
N1X Chip: Technical Specs and Strategic Intent
Technical Breakdown
RTX Spark (development codename N1X) is a joint effort between NVIDIA and MediaTek, with deep collaboration from Microsoft. According to official releases, its core specs include:
CPU: 20-core heterogeneous design, featuring 10 Cortex-X925 high-performance cores and 10 Cortex-A725 efficiency cores, max clock speed of 4.0GHz, based on ARMv9.2 architecture, custom-designed with MediaTek.
GPU: First integration of Blackwell architecture in a PC processor, equipped with 6,144 CUDA cores. Graphics performance matches desktop RTX 5070 discrete GPUs and supports fifth-generation Tensor Cores with FP4 precision.
AI Compute: Delivers 180–200 TOPS of local inference, supports FP4/INT4 compression, can run large-scale models like DeepSeek-70B and Qwen-32B locally, and natively supports Microsoft Copilot+ AI PC standards.
Memory: Supports up to 128GB LPDDR5X unified memory with 301GB/s bandwidth, meeting the demands of AI inference and high-load creative scenarios.
Process: Manufactured using TSMC’s 3nm process.
Power Range: Flagship models range from 45W to 80W; mainstream versions cover 18W to 45W. Mixed-use battery life of 12–15 hours is 1.5–2 times longer than comparable x86 devices, with full-load noise as low as 32dB.
From an architectural perspective, N1X compresses NVIDIA’s core AI training and graphics rendering capabilities into a low-power PC package. With 6,144 CUDA cores, laptops can now achieve graphics compute on par with mid-to-high-end discrete GPUs; 180–200 TOPS of local AI compute means users can run billion-parameter models locally, without relying on the cloud.
The Strategic Significance of Three-Way Collaboration
N1X is not a solo effort by NVIDIA. It’s backed by three strategic alliances: Microsoft ensures Copilot+ AI PC ecosystem compatibility, bringing top-tier AI compute to Windows on Arm for the first time; MediaTek custom-designs the N1X CPU, leveraging its experience in Arm SoC design and modem integration; and Arm provides deep instruction set and architectural optimization. This triple partnership gives N1X distinct advantages in the hardware-software ecosystem, contrasting with Qualcomm’s previous "CPU but no GPU" dilemma in Windows on Arm PCs.
The Business Logic of Edge AI
From a commercial perspective, NVIDIA’s entry into the PC processor market follows two recursive paths:
First, the decentralization of compute demand. As Agent AI migrates to endpoints, cloud inference faces increasing latency and cost challenges. Local AI inference enables PCs to shoulder part of the AI workload, reducing dependence on data center compute and lowering end-to-end inference latency.
Second, the necessity of edge-cloud collaboration. Huang emphasized that future AI systems will not fully replace the cloud, but endpoints must have sufficient local inference capability for real-time tasks. As the world’s largest edge computing device, the PC occupies a pivotal position in this architecture. NVIDIA also unveiled its product roadmap: N1X is a long-term architecture series, with next-gen N2X and N3X chips already in development, plus plans for a compact N1 variant to round out the AI PC product lineup.
Market Response: Data Validates the Narrative
S&P 500 Breaks 7,620: Macro Context
On June 2, Eastern Time, the S&P 500 closed up 9.82 points (0.13%) at 7,609.78, with an intraday high of 7,620.90—its first close above 7,600. The Nasdaq finished at 27,093.90, also setting a new closing record. The S&P 500 marked its 24th record close of 2026 and has risen for nine consecutive sessions, up 11.16% year-to-date.
This rally is not driven by a single event, but by a structural shift in the AI narrative—from "cloud compute dominance" to "expanding edge compute." The launch of NVIDIA RTX Spark is emblematic: it extends the AI investment thesis from data center infrastructure to consumer electronics, broadening the market’s valuation boundaries.
Divergent Performance Among AI PC Stocks
Following the RTX Spark launch, stocks across the industry value chain showed clear divergence.
Big gainers: NVIDIA surged 6.26% to $224.36. Arm Holdings soared 15.73% to $408.85, reflecting high market expectations for its architecture licensing business in the AI PC era. Dell Technologies jumped 10.70% to $465.96, and HP rose 8.51%. Both OEMs are confirmed as the first to adopt RTX Spark chips in their PCs.
ServiceNow saw significant gains around the keynote. On May 29, shares jumped about 14%, marking one of the strongest single-day performances in the software sector for 2026. The rally continued, with a 7.8% rise on June 1. This reflects a shift in market sentiment from "AI may disrupt software business models" to "enterprise AI adoption will drive revenue growth," bolstered by ServiceNow’s launch of new generative AI tools at Knowledge 2026, a $4.2 billion share buyback plan, and renewed coverage from Bank of America.
Underperformers: Intel fell 4.67% to $109.33, its lowest since May 18, ranking as the third-worst performer in the Philadelphia Semiconductor Index. Qualcomm plunged 8.78%, and AMD dropped 1.16%.
The market’s valuation logic varies sharply: the AI PC narrative benefits two types of companies—those supplying AI chips (NVIDIA and Arm), and OEMs that can differentiate with AI PC product lines. Traditional x86 CPU suppliers like Intel and mainstream Windows on Arm PC chip providers like Qualcomm face direct competitive pressure, prompting investors to reassess their market share floors.
How to Trade These Stocks on Gate
All the stocks mentioned above are available for trading on Gate. Gate allows users to trade over 10,000 US-listed stocks and ETFs with USDT, covering NYSE, Nasdaq, NYSE Arca, NYSE American, BATS, and other major US exchanges and liquidity networks. Fractional trading starts at just 0.01 shares, offering users a comprehensive global securities allocation. With stock trading services now live, Gate further bridges digital assets and traditional financial markets on a unified platform.
Investors can log in to Gate, enter the relevant ticker (e.g., DELL, HPQ, INTC, ARM, NVDA, NOW) in the search bar, and place real-time buy or sell orders with USDT—much like trading cryptocurrencies. The fractional share feature is ideal for users looking to build positions in AI PC stocks incrementally, lowering the entry barrier for US equity investment. All the AI PC concept stocks discussed are listed in Gate’s stock trading section, allowing users to tailor their allocation to their investment strategy.
Industry Landscape: From Data Centers to Personal Computers
Bidirectional Expansion: NVIDIA’s Full-Stack AI Strategy
Another highlight of the Computex 2026 keynote was NVIDIA’s full-stack AI strategy, which now extends both "cloud and edge."
For data centers, the Vera CPU—purpose-built for AI Agents—is now in mass production. Vera features 88 custom Olympus cores and 1.2TB/s memory bandwidth—about four times that of the RTX Spark laptop chip’s 301GB/s—and delivers roughly 1.8 times the compute efficiency of traditional x86 processors. Leading firms like Alibaba Cloud, ByteDance, Meta, and Oracle have confirmed Vera deployments. Jensen Huang told media at Computex that Vera could become NVIDIA’s new core growth engine, with market momentum potentially surpassing its current GPU lineup.
For endpoint PCs, RTX Spark brings AI compute to personal devices. NVIDIA plans to launch more than 30 RTX Spark-powered laptops and 10 desktop models. Dell, HP, ASUS, Lenovo, MSI, and Microsoft Surface are all confirmed to roll out these products in fall 2026.
From "Tool" to "Personal AI Assistant": A Shift in Positioning
Huang’s vision for the PC is crystal clear: the future PC is not just a container for running software, but an AI Agent with perception, understanding, reasoning, and action capabilities. Users may name their Agents, interact via messaging apps, and delegate complex, cross-application tasks. Huang described RTX Spark as "not just a hardware product, but a unified cloud-edge compute platform for intelligent agents."
This shift has profound implications for the software ecosystem. Huang stressed that Agents won’t replace Excel, SQL, browsers, or operating systems; instead, they’ll use these tools to accomplish more complex tasks. This means application software won’t be marginalized—it will see more frequent invocation as Agentization accelerates. In post-keynote discussions, Huang also clarified that NVIDIA currently has no plans to launch PC handhelds based on RTX Spark, with all R&D focus directed at rebuilding the AI PC ecosystem.
Quantitative Projections for Structural Changes in the PC Market
Based on disclosed information, the impact of the AI PC era on industry structure can be quantified across several dimensions.
Market share: As NVIDIA enters both data center CPU and PC CPU markets, x86’s share in computing may shrink even faster. NVIDIA’s move into PC CPUs signals that the battle for AI infrastructure dominance is expanding from server GPUs to CPUs, endpoints, and integrated hardware-software ecosystems. The company has announced a full product roadmap—N1X as the long-term architecture, N2X and N3X next-gen chips in development, and a compact N1 variant planned.
Endpoint penetration: OEMs are showing high acceptance of the new platform, with the first RTX Spark-powered products expected to launch en masse in fall 2026, covering mainstream laptops and desktops.
Data center linkage: NVIDIA has secured capacity with foundries like TSMC. Huang said current reserves are sufficient to meet global CPU and GPU demand. Meanwhile, NVIDIA plans to return at least 50% of annual free cash flow to shareholders in 2026, reflecting strong confidence in sustained future cash generation.
Conclusion
Jensen Huang’s Computex 2026 keynote was less a product launch and more a fundamental redefinition of "what a PC is." When the N1X chip brings 180–200 TOPS of local AI compute to thin and light laptops, when the Vera CPU drives Agent workloads in data centers at 1.8 times the speed of x86, when the S&P 500 hits a historic 7,609.78 and Arm and Dell both close with double-digit gains—these events all point to a clear trend: AI’s scale is expanding from cloud data centers to endpoint devices, and the PC sits at the heart of this structural shift.
For investors and market watchers, the value of AI PCs is not about "whether the PC market grows," but "whether the PC’s role changes." As PCs upgrade from tools to AI Agents, their centrality in the AI ecosystem will undergo a fundamental revaluation. The moment the S&P 500 closed at 7,609.78 after nine straight days of gains may well mark the market’s first systematic pricing of this trend.
For those looking to participate in this structural transformation via the secondary market, Gate’s stock trading service offers a convenient entry point—invest directly in upstream and downstream AI PC stocks with USDT, starting from just 0.01 shares. No need for complex traditional brokerage accounts; digital assets and US equities can be managed on a unified platform. As N1X chips reach mass production and the first endpoints launch in fall 2026, every milestone in the AI PC narrative’s journey from concept to reality is worth tracking closely.
