Stay connected with KayaToday—follow us on Instagram and Facebook for the latest news and reviews delivered straight to you.
There is a particular kind of corporate ambition that announces itself not through press releases but through architecture. Not the architecture of buildings, but the architecture of strategy — the way a company arrays its products, its partnerships, and its capital to occupy an entire market rather than merely competing within one. On the morning of June 1, in a packed auditorium in Taipei, Jensen Huang spent two hours revealing that Nvidia’s ambitions have become architectural in precisely this sense.
In the span of a single keynote, Huang unveiled six new data centre chips, an entirely new PC platform co-developed with Microsoft, a 500-billion-parameter open AI model, next-generation humanoid robotics software, and a declaration that Microsoft and Nvidia are going to “reinvent the personal computer.” The breadth of the announcement was not accidental. It was the explicit articulation of a strategy that has been in the making for three years: to own every layer of the AI economy, from the data centre silicon to the laptop in a student’s bag to the robots assembling the next generation of factories.
The Data Centre Layer: Vera Rubin
The centrepiece of Nvidia’s data centre roadmap is the Vera Rubin platform — the company’s most significant architectural advance since Blackwell. Six new chips form the complete platform: the Rubin GPU with 336 billion transistors on a dual-die TSMC 3nm process, the Vera Arm-based CPU, the NVLink 6 switch, the ConnectX-9 SuperNIC, the BlueField-4 DPU, and the Spectrum-6 Ethernet switch. Together, they constitute not merely a new chip generation but a complete rearchitecting of how AI computation is networked and organised at the rack level.
Huang announced that Nvidia’s Vera CPU for data centres is now in full production, with millions being manufactured for a market he described as one “that never existed before.” Early customers include Anthropic, OpenAI, SpaceX’s xAI, Dell, Oracle, and CoreWeave. The customer list is, in effect, a roster of every company building the AI infrastructure on which the next decade of the industry will run. Securing that list before competitors can offer alternatives is a first-order distribution moat.
The PC Layer: RTX Spark and the Reinvention Claim
The RTX Spark superchip — also referred to as the N1X — debuts in the autumn on a new line of Windows PCs from Microsoft, Dell, HP, ASUS, Lenovo, and MSI. The chip is Arm-based, a direct challenge to the x86 architecture that Intel and AMD have dominated for decades, and a signal that Nvidia intends to follow Apple’s path from data centre dominance to the consumer device market.
Nvidia’s plan to build system-on-chips for PCs sent shares of Advanced Micro Devices, Intel, and Qualcomm downward immediately following the announcement — a market reaction that was less about the RTX Spark’s immediate revenue contribution and more about the competitive signal it sends. A company with Nvidia’s brand credibility in AI performance, now entering the laptop market with a chip co-engineered with Microsoft and supported at launch by every major PC manufacturer, is not a marginal competitor. It is a structural threat to an incumbency that Intel and AMD have not seriously had to defend since Apple’s M-series chips demonstrated how comprehensively the Arm architecture could outperform x86 in the right hands.
The Deeper Claim
Nvidia at $5.23 trillion — the most valuable company in the world — is betting that the AI economy rewards vertical integration across every layer. The logic of the Computex keynote is that the same company supplying the data centre racks on which frontier AI models are trained should also supply the laptop chips on which those models are deployed locally, the robotics software on which they are embodied physically, and the open model weights on which developers build the next generation of applications.
This is a strategy with a historical precedent, and the precedent is instructive. Microsoft dominated the PC era not by building the best hardware but by controlling the software layer that made all hardware interchangeable. Intel dominated the semiconductor industry not by competing in every market but by ensuring that its architecture became the unavoidable foundation of every computing stack. Nvidia is attempting something more audacious than either: vertical integration across hardware and software, data centre and consumer device, cloud and edge, simultaneously.
The question that Computex 2026 does not answer is whether that degree of integration is regulatorily sustainable. The question is whether any competitor — or any combination of competitors — can build an alternative stack before Nvidia’s lead becomes permanently structural. Antitrust regulators in Washington, Brussels, and Beijing are watching a company accrue the kind of cross-layer market power that historically triggers intervention. Nvidia’s legal and government affairs teams are, one assumes, acutely aware of this.
Nvidia has emerged as the world’s most valuable company by dominating the market for artificial intelligence chips in the data centre. Now the company is expanding its prowess to chips that will serve as the main processor for personal computers, entering an arena that has long been ruled by Intel, Advanced Micro Devices, Qualcomm, and Apple. The ambition is extraordinary. What Computex confirmed is that the capital, the partnerships, and the production capacity to pursue it are all now in place. Whether Huang’s vision of owning every layer of the AI economy is a strategic masterpiece or an overreach of historic proportions is a question that the market — and eventually the regulators — will answer. Taipei was simply where the claim was formally made.
Read Also: Google Just Declared the Age of the AI Agent; The World Should Pay Attention
