Nvidia's Balancing Act: Record Results and a Splashy PC Chip Face a Murky GPU Roadmap

Nvidia heads into Computex week with both a historic quarter under its belt and growing unease over its next-generation GPU lineup. The stock, which ended last week at 181.40 euros — roughly 10% below its 52-week high of 201.05 euros — is hovering near oversold territory with an RSI of 36. Yet the contrasting signals from the company’s product roadmap and its soaring financials have left investors parsing which narrative will prevail.

The headline numbers from the first quarter of fiscal 2027 are hard to ignore. Revenue hit a record $81.6 billion, up 85% year over year, with the data center segment alone contributing $75.2 billion — a 92% jump. Adjusted earnings per share surged 140% to $1.87. For the current quarter, Nvidia is guiding for around $91 billion in sales, and that forecast excludes any contribution from China. The board has also authorized an $80 billion share buyback program and raised the quarterly dividend from one cent to 25 cents per share, with the ex-dividend date falling in the coming days and payment scheduled for June 26.

But while the financial engine hums, the machinery underneath is showing cracks. Industry sources indicate that the Rubin CPX — an inference-optimized GPU originally slated for the second half of 2026 — is effectively dead. No orders have been placed for memory chips or substrates, and there has been zero activity around the planned GDDR7 memory component. One substrate industry representative put it bluntly: many in the supply chain now regard the project as shelved. Nvidia has not publicly confirmed the cancellation, but the product’s absence from roadmaps presented at GTC 2026 has amplified the belief that it has been either scrapped, postponed, or fundamentally reworked.

Nvidia’s vice president Ian Buck has hinted that the CPX concept hasn’t been abandoned entirely but has been pushed out to the Feynman generation — still years away. The shift reflects changing priorities in AI inference: the focus has moved from long context windows to "time to first token" — how quickly a model delivers its initial response. That architectural shift favors a different approach, boosting the profile of the Rubin LPX tray, which uses Groq’s LPU units to accelerate the decode phase of inference workloads.

Should investors sell immediately? Or is it worth buying Nvidia?

The Rubin generation as a whole is facing headwinds. TrendForce recently slashed its estimate of Rubin’s share of Nvidia’s high-end GPU shipments for 2026 from 29% to 22%. The research firm cites the complex validation of new HBM4 memory, the migration to ConnectX-9 networking cards, and heightened demands on power delivery and liquid cooling as key bottlenecks.

Yet even as doubts swirl around the GPU roadmap, Nvidia is making aggressive moves on other fronts. At Computex in Taipei (running June 2–5), the company is unveiling the N1X, its first ARM-based high-performance processor for Windows laptops. Built on TSMC’s 3-nanometer process, the chip packs a 20-core ARM CPU alongside a GPU comparable to an RTX 5070. Partners including Dell, Lenovo, ASUS, and MSI are expected to show initial devices, with a market launch planned before Christmas 2026. Notably, the N1X is designed to run large language models with over 100 billion parameters directly on the laptop — a direct challenge to the x86 stronghold of Intel and AMD.

Alongside the product push, Nvidia is deepening its ties to Taiwan. The company plans to ramp up annual spending in the region to $150 billion and is building a new headquarters in Taipei, with construction set to begin in June or July 2026 and operations starting in 2030. CEO Jensen Huang will kick off the week on June 1 with a keynote at GTC Taipei, where he is expected to provide more details on the "Vera Rubin" platform and AI agent applications.

One area of concern that may surface during the event is the rising cost of memory for the Vera Rubin generation. Analysts warn that memory costs could surge more than 400%, consuming a significant chunk of material expenses. That could dent margins even as revenue climbs.

Nvidia at a turning point? This analysis reveals what investors need to know now.

Institutional investors remain broadly bullish. Mitsubishi UFJ Asset Management recently increased its stake by 4.7% to roughly $9.93 billion. Among the 61 analysts surveyed by S&P Global, the consensus price target stands at around 270 euros per share — implying upside of more than 40% from current levels.

With Huang’s keynote and a flurry of product announcements set for the coming days, Nvidia has a chance to refocus the narrative. Whether it will use the platform to address the growing supply chain speculation — or let its silence on the Rubin CPX speak for itself — could determine whether the stock finally breaks out of its recent trading range.

Ad

Nvidia Stock: New Analysis - 31 May

Fresh Nvidia information released. What's the impact for investors? Our latest independent report examines recent figures and market trends.

Read our updated Nvidia analysis...

en | US67066G1040 | NVIDIAS | boerse | 69455135 |