40,000 Employees Today and 4 Million Agents Tomorrow: Nvidia CEO Jensen Huang Believes Agentic AI Will Soon Become the Norm in the Workplace

Nvidia says its newly released Vera CPU platform could generate $20 billion of revenue in its current fiscal year, with potential to more than double by fiscal 2028, as agentic AI increases demand for chips that execute automated workflows. CEO Jensen Huang expects AI agents to become pervasive across enterprises, complementing Nvidia GPUs in AI reasoning workloads. Bank of America estimates the data-center CPU market could grow nearly fivefold from $35 billion this year to $170 billion by 2030, though expanded use of autonomous agents raises cybersecurity and access-control risks.

Analysis

The investable read-through is less about a new CPU profit pool and more about Nvidia expanding its attach rate across the AI stack. If autonomous workloads move from episodic model queries to persistent orchestration, compute utilization, networking, storage I/O and enterprise software spend all rise; NVDA's upside would come from protecting system-level gross margin rather than merely taking CPU share from Intel (INTC) and AMD (AMD). The more vulnerable incumbents are x86 vendors whose installed-base advantage does not translate into tightly integrated AI racks, while ARM Holdings (ARM) is a beneficiary if custom ARM-based server designs proliferate.

The company-supplied revenue and performance assertions require independent confirmation in product shipment data, hyperscaler qualification schedules, and disclosed customer wins. A standalone CPU opportunity is unlikely to be valued at GPU-like multiples: CPUs are a mature, price-competitive market, and any material mix shift toward lower-margin processors could dilute NVDA's blended gross margin unless bundled with accelerators, InfiniBand/Ethernet networking, and software. Over the next 1-3 months, the relevant catalyst is evidence that hyperscalers are ordering complete agentic-AI infrastructure rather than substituting CPUs for accelerator spend.

The second-order beneficiary may be cybersecurity, not compute. Wider delegated access to internal systems raises identity, endpoint, data-loss-prevention and machine-to-machine authentication requirements, favoring PANW, CRWD and OKTA; security budgets can become a gating cost that delays broad enterprise deployment. The contrarian view is that enterprise agent adoption remains constrained by permissioning, auditability and error liability, making near-term revenue claims vulnerable to pilot-to-production conversion delays despite strong developer enthusiasm.

Key Decisions for Investors

• Maintain a core long NVDA only versus a semiconductor benchmark rather than add aggressively on this article; reassess after the next earnings call for disclosed CPU/system revenue, gross-margin trajectory and hyperscaler design wins. Falsifier: guidance implies lower blended gross margin or customers describe CPU purchases as substitutions for GPU/networking spend.
• Initiate a 3-6 month basket pair: long PANW and CRWD / short IGV, sized market-neutral. Agent deployment should redirect incremental enterprise spend toward security control planes before broad software productivity gains are monetized; exit if enterprise security billings or remaining-performance-obligation trends decelerate materially.
• Watch AMD, INTC and ARM for server-CPU order commentary and hyperscaler qualification announcements before expressing a relative-value trade. A long ARM / short INTC structure is attractive only if ARM server royalty growth accelerates while INTC data-center revenue guidance weakens; absent that evidence, the article alone is insufficient signal.
• Avoid treating the cited CPU revenue potential as a modeled NVDA earnings catalyst until independently corroborated. If NVDA's next two quarterly disclosures do not separately evidence meaningful platform attach and sustained system-margin economics, fade any valuation expansion attributable solely to the CPU narrative.