#美股财报季 #Gate股票暢聊季 The market battle ahead of NVIDIA’s earnings this time is a key turning point, as the AI computing-power rally officially shifts from “trading on expectations” to “verifying real demand.”

On AI demand: Can cloud providers’ capital spending support GPU orders?
The existing order base is solid, and new overseas purchases could push it up another notch. The 2025 GPU holdings plans of the world’s five largest tech companies have already locked in massive GB200 capacity: Microsoft has secured 700,000 to 1.4 million GB200 units, while Google and Amazon have finalized purchases of 400,000 and 360,000 units, respectively. OpenAI’s GB200 purchases are at least 400,000 units, underscoring highly inelastic demand. Saudi Arabia and the UAE have also recently been snapping up several thousand AI graphics cards in bulk, with incremental demand from new markets far exceeding previous expectations. Continued expansion by cloud providers can fully support GPU orders in the next phase, and NVIDIA’s production schedule is already booked through the second half of 2025, leaving no risk of near-term order cancellations.

On profitability: Which matters more, margins or growth guidance?
Margins above 70% provide confidence today, while clear long-term growth guidance anchors valuation; both are indispensable. NVIDIA’s AI GPU sales are expected to reach $40 billion in 2024, with a global market share above 90%. By comparison, AMD’s AI GPU sales totaled just $3.5 billion and Intel’s just $500 million, putting them in a completely different league. NVIDIA’s AI chip business has maintained gross margins above 70% over the long term, a level its competitors simply cannot replicate. Only a combination of revenue growth, high margins, and long-term guidance can support its current trillion-dollar market capitalization. High revenue growth alone, accompanied by declining margins, would instead signal that supply-demand conditions are starting to loosen, prompting the market to reprice the stock immediately.

On valuation disagreements: Is the pullback a release of sentiment or a valuation reset?
This is a rapid release of short-term sentiment, nowhere near a systemic valuation reset. NVIDIA’s P/E ratio previously surged to 220x, an extremely high level even among US tech stocks, and had already priced in some long-term expectations. In May, NVIDIA opened lower but then rallied, gaining 7% in a single day and adding RMB 520 billion in market capitalization, showing that core investors had not exited. As long as Blackwell chip deliveries meet expectations, subsequent earnings will quickly absorb the current high valuation. The real adjustment risk is not the high valuation itself, but if the earnings report’s guidance for the next quarter comes in far below market expectations.

On market opportunities: Which is most likely to recover first?
The main ranking is memory chip stocks rebounding first, followed by NVIDIA, with AMD having the weakest upside. For memory chips, HBM is currently the most critical bottleneck in AI computing power. The “father of HBM” has also said that AI dominance is shifting from GPUs to memory, and there are even rumors that NVIDIA may acquire memory companies to address its shortcomings. Supply-demand conditions are tightest here, giving the sector the strongest rebound potential after the pullback. For NVIDIA, its market share is overwhelmingly dominant and order visibility is extremely strong; once the earnings report is released, it can immediately open up room to rise. Its fundamentals are the most stable, with moderately strong upside. For AMD, its share of the AI GPU market is only around 8%, and its products still lag NVIDIA significantly. Without a product breakthrough that exceeds expectations, it will be difficult to stage an independent rally, leaving it with the weakest upside. $NVDA