Nvidia’s Unusual Discount: Surging Profits Drive P/ E to Seven-Year Low as Memory Bottleneck Shifts Focus

Nvidia's market cap dropped ~€1T but forward P/E fell to ~18, lowest in 7 years, as earnings rise amid AI supply chain shift to memory. Bulls cite pricing power; bears warn of in-house chips and DRAM costs.

The numbers don’t add up in the way they usually do. Nvidia’s market capitalisation has lost roughly a trillion euros since mid-May, shaving over 12% off the stock price. Yet the company’s projected earnings for the next twelve months have climbed so quickly that its forward price-to-earnings ratio has fallen to around 18 — the lowest in seven years. Investors are grappling with a paradox: a business that is earning more money than ever before has never looked cheaper on a relative basis.

The immediate cause of the valuation compression is a shifting view of where the next bottleneck lies in the AI supply chain. For the past two years, the constraint has been the sheer compute capacity of Nvidia’s GPUs. Now, attention is turning to memory. Spot prices for DRAM have roughly decupled over the past year, and the cost of High Bandwidth Memory is eating into the margins of data-centre operators. Nvidia still commands a 97% share of the server-GPU market, but the question is whether its ecosystem can navigate a supply squeeze that originates outside its own fabs.

Bulls argue that the market is missing the forest for the trees. At a forward P/E of 18, Nvidia trades below the S&P 500’s multiple of over 20 and the Nasdaq 100’s ratio of 23. Revenue is forecast to hit nearly $216 billion in fiscal 2026, a 65% year-over-year jump. Analysts see a 48% upside from current levels, with a consensus price target of €263.69, and the company has an $80 billion share buyback programme and a quarterly dividend of $0.25 in place. Moreover, the Chinese market is slowly reopening: Alibaba and ByteDance are now permitted to buy certain H200 chips, and Nvidia’s data-centre revenue alone is projected to reach $203 billion by the end of 2027.

Pricing power is the bull case’s ace. Bank of America and Citigroup have highlighted that as Nvidia transitions from the Blackwell architecture to its next-generation Rubin platform, rack-level prices could rise by $2–3 million per unit. That increase should more than offset the rising cost of memory, keeping gross margins in the 70–75% range even if DRAM prices remain elevated.

The bear camp counters that Nvidia’s biggest customers — Google, Amazon, Microsoft and Meta — are developing their own AI processors, a move that directly threatens the company’s pricing leverage. More immediately, rumours have surfaced about delays in the “Kyber” rack system. According to research firm SemiAnalysis, specialised circuit boards for the Kyber architecture are causing trouble, with some reports pushing the timeline to 2028. Nvidia has denied any such delay, insisting that the regular Rubin platform remains on track for the second half of 2026. But the mere speculation has injected a dose of anxiety. Meanwhile, the soaring cost of DRAM is a double-edged sword: it pressures data-centre operators’ margins, potentially deferring new GPU orders. The rental price for H100 chips has already dropped from its peak of $3.20 per hour, signalling that the era of extreme scarcity may be fading.

Technically, the stock is hunting for direction. It closed at €177.60, just below its 50-day moving average of €180.93 and 7.86% above the 200-day average of €164.66. The relative strength index sits at 51.3 — neutral territory that suggests the market is waiting for a catalyst rather than driving a trend.

That catalyst is scheduled for August 26, when Nvidia is expected to report fiscal second-quarter earnings. Analysts are looking for revenue of roughly $91.7 billion. The management team will need to convince investors that the memory bottleneck is manageable and, crucially, provide a concrete timetable for the Rubin architecture. If the roadmap is clear and margins hold, the stock could rally toward the analyst target. If delays are confirmed or pricing power looks shaky, a test of the 52-week low near €140.30 is plausible. For now, Nvidia remains the centrepiece of the AI infrastructure build-out — but the narrative is no longer about pure growth; it is about whether that growth can survive the constraints that come with it.