For years, exceeding expectations was enough to excite the markets. Today, even spectacular results do not guarantee euphoria: one must surpass rumors, crush the most optimistic projections, and defy logic to keep investors happy. NVIDIA has just experienced this firsthand. On paper, its figures are impressive: annual revenues of $68.1 billion, a 73% increase. The data center segment reached $62.3 billion, and gross margins exceeded 75%. The guidance for the first fiscal quarter points to $78 billion, well above the consensus of $72.8 billion. At any other time, this would be cause for celebration. But the context has changed. The market no longer rewards just growth, but the promise of perpetual growth. NVIDIA is no longer seen as a mere chip manufacturer. It is now considered the backbone of the AI revolution, the equivalent of the electrical system for artificial intelligence. Investors are not asking if revenues are increasing, but whether that growth can be sustained at an exponential rate indefinitely. The $78 billion guidance did not disappoint in terms of fundamentals, but it did in terms of excessive expectations. Some analysts expected figures above $80 billion. When the "whisper number" becomes the benchmark, results are no longer negotiated, but myths. Jensen Huang, CEO of NVIDIA, insists that we are witnessing a reconstruction of global computing infrastructure that will take years. He is not wrong: replacing legacy systems with AI-capable hardware is a capital cycle, not a two-quarter phenomenon. It is modern electrification, where cables carry models instead of current. The problem is that capital cycles have their own rhythm, while euphoric markets do not. The situation with China adds uncertainty. NVIDIA excluded data center revenues in China from its forecasts. No contributions are expected, and although there is a licensing framework for the H200, revenues are zero and approval processes remain uncertain. Small shipments, U.S. inspections, and tariffs complicate the landscape. More than a growth engine, it is a regulatory hurdle. China is the largest chip market in the world. Considering it an option rather than a primary source of income is prudent, but it removes a tailwind just as expectations are at their peak. AI demand has turned high-bandwidth memory into a critical resource. Prices remain firm, supply is limited, and the entire semiconductor chain is now interdependent. NVIDIA claims to have inventory and capacity secured for the coming quarters, which is reassuring, but any restriction could alter the narrative. The data center remains the main engine. The gaming and automotive segments fell short of expectations. The former revenue pillars no longer lead; everything now depends on AI computing. Multi-billion dollar deals reinforce the story: Meta and AMD have signed multi-year purchase commitments, ensuring revenue visibility in the tens of billions. This suggests solid demand, but also a capital arms race. Some skeptics warn of circular ecosystems: suppliers finance customers and vice versa, which can lead to synchronized overinvestment. It is not about fraud, but about assuming that the cycle will never stop. We are not in 1999: there are real revenues, solid margins, and tangible deployment. But the market is now asking whether the AI economy is accelerating or simply pulling forward several years of spending in a short time. The market reaction indicates that we have moved from blind accumulation to meticulous analysis. When a 73% growth does not generate enthusiasm, it is a sign that dependence has replaced disbelief. NVIDIA remains the toll on the AI highway. As long as capital flows into the training and deployment infrastructure of models, revenues will continue to come in. The question is whether the growth curve can be maintained without collapsing under its own weight. For now, the fundamentals are solid: elite margins, a deep order book, and multi-year agreements that enhance visibility. But the valuation already discounts an industrial revolution without pauses, frictions, or capital fatigue. It was not a bad quarter, but a reality check. The market no longer asks if NVIDIA is winning, but how long this fever can last before the easiest veins run dry. In the early stages of a bull market, price follows earnings. In advanced stages, earnings do not meet expectations. The expansion of AI is real. Revenues and margins are too. The question is whether the expectations are. The close of Wall Street boosted Asian markets, and this time the catalyst was capital expenditure (Capex). NVIDIA presented revenue guidance for the first quarter between $76.4 and $79.6 billion, far exceeding forecasts. This confirms that the investment cycle in AI continues to accelerate. For Asia, what matters is not just the specific data, but what it implies for the entire semiconductor chain. If NVIDIA maintains this level of demand, the ecosystem—from foundries to chip exporters—retains revenue visibility, which is crucial for Asian markets. For weeks, the debate has revolved around whether AI valuations were too ahead of fundamentals. This report helps close that gap: high multiples and Capex are justified when revenue growth exceeds expectations by billions. This provides backing in both Capex and results for the bet on AI in Asia, reducing the risk of a sharp correction and reinforcing the idea that the current rally is supported by real cash flows. That is why South Korea, Taiwan, and Japan are confident: they are not debating whether AI destroys jobs, but supplying the hardware that drives it. For them, Capex is synonymous with revenue. Previously, every headline about AI seemed decisive. Now, the market watches cautiously, aware that this is not just a technological cycle, but an environment that has perfected the art of postponing consequences under the name of "price discovery." The recent rebound in software was more of a short-covering reaction than an institutional rotation. The S&P closed in the green, but credit did not accompany the stock market optimism, suggesting that the recovery lacks depth. Goldman Sachs summed it up well: it is not a question of growth versus value, but of checking who has structural advantages. AI eliminates frictions; asset-light models and bottlenecks should be priced differently when those bottlenecks disappear. Capital has shifted towards tangible and low-obsolescence assets: data centers, energy, chips. Hyperscalers are the exception because their Capex is the income of others. Asia understands this: they do not discuss the labor impact of AI, but export the equipment that makes it possible. NVIDIA led the recovery, but its call with investors generated concern: was it due to Chinese competition or the size of capital spending? The important thing is that expectations are compressed into a narrow range, with the entire sector using NVIDIA as a benchmark for visibility to 2027, exposure to China, and demand for new architectures. While stocks debate narratives, the credit market tells another story. Corporate debt is increasingly in the hands of price takers, not makers. Liquidity depends on flows, not fundamentals. Spreads remain tight, but credit volatility remains high. Insurance remains expensive, even as the stock market relaxes. Markets are experts at postponing adjustments. Multiples are relative, and once they wobble, they do not stabilize until expectations are readjusted. The AI catastrophe thesis is difficult to refute in the short term because it lives in the future, just like utopias. The immediate question is whether revenues, Capex, and deployment timelines justify the current premiums in software. For now, the market remains in recovery mode: rising hardware, pressured shorts, firm crypto and gold, stable oil. But the true arbiter is not the Nasdaq, but credit. (Source: fxstreet.com)