Nvidia’s $119 Billion Order Backlog and Seoul HBM Pacts Paint a Picture of Scarcity, Not Weak Demand

Nvidia locks in HBM4 memory for Vera Rubin amid warning of multi-year advanced chip shortage; stock dips 5% but analysts see 45% upside.

When Jensen Huang lands in Seoul this weekend, he is not coming to negotiate for customers. He is coming to secure factories. The Nvidia chief executive will sit down with SK Group chairman Chey Tae-won on Monday, June 8, to formalise a cooperation plan that locks in supply of fourth-generation High Bandwidth Memory (HBM4) for the coming Vera Rubin supercomputing platform. Samsung and Micron are also being lined up as secondary sources to diversify the chipmaker’s memory supply chain. The urgency reflects a warning Huang has been sounding: the world faces a multi-year shortage of advanced memory chips that could throttle the entire AI buildout.

That warning is backed by numbers that suggest the constraint is on the production side, not in demand. Nvidia’s purchase commitments for supply chain components hit $119 billion in the first quarter of fiscal year 2027. That is not a figure associated with a company struggling to sell its products — it is an order backlog that gives the business revenue visibility stretching years into the future. Huang said in Seoul that the supply crunch will last “several years” and extend well beyond 2027, spanning everything from wafers to cable connectors.

The stock market took a different view on Friday. Nvidia shares dropped 5.42% to €178.08, pulling the stock about 12% below its 52-week high of €202.50. The move was tied to US labour market data that dampened rate-cut hopes, a macro-driven selloff that left the stock’s relative strength index at 45.3 — neither overbought nor oversold. Year to date, the shares are still up roughly 10%, and over the past twelve months they have gained 45%.

Technical support now lies at the 50-day moving average of €174.40, a level the stock is hovering just 2% above. A break below that would open the path to the 200-day average at €161.46. For now, the correction looks more like a healthy cooling-off after a strong run than a structural breakdown, but the next few trading sessions will test that thesis.

Adding to the short-term noise, insider selling has picked up. Over the past 90 days, company insiders have sold shares worth approximately $387 million. The most notable transaction was director Mark A. Stevens unloading 500,000 shares. Institutional investors, however, have largely held their positions, pointing to a divergence between management’s portfolio adjustments and long-term conviction funds.

Analysts remain bullish. The consensus price target stands at €258.67, implying upside of roughly 45% from Friday’s close. Bank of America reaffirmed a “Buy” rating with a target of $350, driven by expectations that the value embedded in each gigawatt of AI infrastructure will double from $40 billion in the current Blackwell generation to as much as $80 billion when Vera Rubin ramps up. The operating numbers support that view: Nvidia’s latest quarterly revenue surged 85% to $81.6 billion, while the company raised its quarterly dividend to $0.25 per share and maintains a multibillion-dollar buyback programme.

All eyes now turn to Tuesday’s US consumer price index release, which will set the near-term trading rhythm. But the real strategic milestone comes Monday in Seoul. If Huang’s supply chain pacts with SK Group, Samsung and Micron demonstrate a credible path through the chip shortage — and particularly secure HBM4 for Vera Rubin — then Friday’s selloff may look like exactly what it technically resembles: a temporary gust in a long-term updraft. As Huang put it, the world is currently “busy building AI.” For Nvidia, building the infrastructure to supply that boom is what matters far more than the week’s closing price.