Nvidia’s Reporting Overhaul Lays Bare the $200 Billion CPU Ambition Behind Record Earnings

Nvidia restructures into Data Center & Edge Computing, debuts Vera CPU for agentic AI, posts record $81.6B revenue, and invests $150B in Taiwan supply chain.

Nvidia is giving investors a sharper lens on where its next growth wave will come from. On May 28, the chip giant unveiled a new financial reporting structure that carves its business into two distinct segments: Data Center and Edge Computing. The move, outlined at the TD Cowen technology conference in New York, is more than a cosmetic tweak — it’s a signal that Nvidia expects autonomous, agent-based AI systems to generate a wholly new revenue stream, one that stretches well beyond the GPU-dominated data center.

At the core of that bet sits the Vera CPU architecture, purpose-built for “agentic AI” — applications that run continuously and make real-time decisions without human intervention. Nvidia estimates the addressable market for these processors at roughly $200 billion. Early benchmarks show that even under sustained full load, the platform maintains peak performance while slashing the cost per inference token by up to 90% against its predecessor. For enterprises deploying AI at scale, that kind of efficiency swing is a game-changer.

The numbers behind the strategy are already staggering. In the first quarter of fiscal 2027, Nvidia posted record revenue of $81.6 billion, a year-over-year jump of 85%. The data center segment alone contributed $75.2 billion, up 92%. Adjusted earnings per share hit $1.87, comfortably ahead of the $1.75 consensus. The board authorized an additional $80 billion in share repurchases, with $38.5 billion left over from the previous program, and raised the quarterly dividend from a token $0.01 to $0.25 — payable on June 26 to shareholders of record on June 4.

Yet the stock has struggled to build on the momentum. Shares recently changed hands at around €184, roughly 8% below the 52-week peak reached in mid-May. The relative strength index stands near 40, technically an oversold reading. On a weekly basis the equity has slipped about 3%, though it remains up 13.67% year-to-date and has gained 53.29% over the past twelve months. Part of the restraint comes from elevated bond yields: the 30-year U.S. Treasury touched 5.197% recently, putting pressure on high-multiple growth names. Analysts on the conference floor noted that, while the quarterly report was undeniably strong, the market is now pricing in flawless execution against Nvidia’s 2026 and 2027 product roadmap — the ramp of Blackwell Ultra and the planned launch of the Vera-Rubin platform.

CEO Jensen Huang underscored the industrial backbone behind those plans. Nvidia is funneling roughly $150 billion annually into its Taiwanese supply chain — about ten times the level of five years ago. That capital is flowing into a new regional headquarters and the addition of some 4,000 local jobs, cementing the island as the linchpin of global AI hardware production. Huang described the broader build-out of AI infrastructure as the largest investment wave in history, with annual spending projected to reach between $3 trillion and $4 trillion by the end of the decade.

For the current quarter, Nvidia forecasts revenue of roughly $91 billion, well above the analyst consensus of $87.2 billion. The product pipeline is already drawing interest: the Vera Rubin architecture, scheduled for a 2026 debut, is generating early demand signals that executives say are even stronger than those seen during the Blackwell cycle. Meanwhile, the company is still selling existing hardware at a brisk pace. A fresh two-year deal between QumulusAI and Shadeform, announced today, will see two H200 clusters with 680 GPUs go live in Kansas City — precisely the kind of deployment that Nvidia hopes Vera will redefine at a lower cost base.

The new reporting structure makes this two-pronged strategy more transparent. In Data Center, Nvidia will now break out hyperscale customers from the “ACIE” segment — AI Clouds, industry, and enterprises — giving a clearer view of public-sector AI programs and corporate adoption. Edge Computing bundles together devices for physical and agent-based AI: PCs, robotics, automotive components, and AI-RAN base stations. That consolidation brings into focus businesses that had previously been scattered across line items.

For investors, the message is clear. The record quarter has already been discounted. Nvidia’s next leg rests on its ability to execute the Vera rollout and prove that the $200 billion CPU target is not just ambition but deliverable reality. With an $80 billion buyback as a backstop and a dividend that now carries real weight, the board is betting the company can do exactly that.