Nvidia has fired off two contradictory signals in the span of a few days. The chip giant is raising its quarterly dividend from a token $0.01 to $0.25 per share – a move that screams cash‑return confidence. At the same time, it sold $25 billion in bonds that investors devoured with $85 billion in orders, locking in cheap money for maturities stretching to 2056. A company that prints cash shouldn’t need to borrow, unless management sees an investment cycle so long and so deep that hoarding dry powder today is the prudent play.
The bond offering, the largest in the company’s history, was almost comically oversubscribed. Nvidia tripled its original target after order books swelled. The message to the market: the AI build‑out is still in its early innings, and Nvidia intends to finance a multi‑decade infrastructure boom with the cheapest capital available. That bullish signal is reinforced by Morgan Stanley’s forecast that the four largest cloud providers – the hyperscalers – will collectively issue $400 billion in bonds this year, nearly 2½ times last year’s tally. That wall of money flows directly into data‑center hardware, much of it powered by Nvidia’s chips.
Yet here lies the paradox. Those same hyperscalers – Google, Amazon, Microsoft and Meta – account for roughly 40 percent of Nvidia’s revenue. And every one of them is designing its own accelerators. Google’s TPUs run on a fully proprietary software stack; Amazon’s Trainium and Inferentia chips use a home‑grown compiler framework; Microsoft’s Maia silicon is optimised for its own workloads; Meta is scaling up its MTIA programme. Crucially, Google and Amazon already offer their chips to external customers, turning an internal hedge into a direct competitive threat.
For training the largest AI models, Nvidia’s CUDA ecosystem remains a formidable moat – two decades of development, more than 4 million developers, and every major machine‑learning framework built primarily around CUDA. Switching costs are measured in years. For inference, however, the picture is hazier. OpenAI’s Triton allows hardware‑agnostic GPU programming, PyTorch’s torch.compile abstracts chip differences away, and AMD’s ROCm 7 now rivals CUDA on most inference workloads. Parities not yet reached, but the direction of travel is unmistakable.
Should investors sell immediately? Or is it worth buying Nvidia?
China adds another layer of pressure. Nvidia’s market share for AI accelerators in the country has effectively collapsed. CEO Jensen Huang called that outcome “terrible for the United States” because it breaks the software‑dependency loop that CUDA had built over two decades. Export controls continue to redraw the addressable market, and no quick fix is in sight.
Wednesday’s virtual annual general meeting, set for 18:00 Central European Time, will put these tensions front and centre. The agenda covers the election of ten directors, executive compensation approval, the reappointment of PricewaterhouseCoopers as auditor, and four shareholder proposals. While AGMs rarely move stock prices, the timing is loaded. Micron reports earnings after the US close the same day, and its HBM memory sales are a direct barometer of GPU supply constraints. The next day brings the US PCE inflation data for May – the Federal Reserve’s preferred gauge. For a stock with an annualised volatility north of 40 percent, any macro surprise could whip the shares sharply.
Technically, Nvidia closed Friday at €181.96, barely above its 50‑day moving average of €180.04. The 200‑day average sits farther at €163.04, leaving the long‑term trend intact, but the near‑term picture is indecisive. A sustained break below the 50‑day line would signal a deeper consolidation; holding it keeps the path open toward the all‑time high around €202.50. Of the 62 analysts surveyed by S&P Global, the consensus is a “strong buy.”
This week will test whether the market believes Nvidia’s dominance is self‑reinforcing or whether the whispers of a structural unbundling – customers building their own chips, China slipping away, the CUDA grip loosening at the edges – are gaining weight. The $25 billion bond bet says the company itself is wagering on the first scenario. The next few days may begin to reveal which one is right.
Ad
Nvidia Stock: Buy or Sell?! New Nvidia Analysis from June 21 delivers the answer:
The latest Nvidia figures speak for themselves: Urgent action needed for Nvidia investors. Is it worth buying or should you sell? Find out what to do now in the current free analysis from June 21.
Nvidia: Buy or sell? Read more here...
