This impressive growth hides a paradox: the more profits surge, the greater the risk of a cyclical distortion in the semiconductor industry.
Nvidia is no longer just a stock, it’s a key indicator for the global market. Today, it’s the company with the highest market cap in the world, and to understand where we’re going, just look at one fundamental fact: its main customers include Microsoft, Amazon, Google, Oracle, and Meta—literally, the digital infrastructure of contemporary capitalism. And if such a company posts numbers like net income of $31.91 billion (+65% YoY), revenues of $57.01 billion (vs. expected $54.92 billion), and EPS of $1.30 (vs. $1.25 expected), it means these companies are continuing to purchase GPUs at unprecedented levels.
The intuitive conclusion would be this: Big Tech is increasing AI-related CAPEX, so everything is moving toward structural expansion. Yes, but there’s a problem. And it’s hidden. Because these extraordinary results aren’t just proof of Nvidia’s strength, they’re also a symptom of the cyclical overextension of an entire industry.
What Nvidia’s numbers really say
Formally, everything is perfect. Overwhelming growth, exceeding forecasts, and even more aggressive prospects for the following quarter. An impeccable sequence. Yet a 65% YoY increase in earnings isn’t "organic growth": it’s the expression of a demand boom concentrated in a single segment, the AI data-center segment.
This cycle, by definition, is not linear. It’s a hyper-accelerated investment loop driven by the need of large hyperscalers (Azure, AWS, Google Cloud, Oracle Cloud, Meta) to build computational capacity before their competitors.
The market is interpreting it too simplistically: "Earnings are growing, demand is growing, multiples are growing, AI is infinite."
But the question to ask is another: Is this pace sustainable?
A growing sector... but it’s cyclical
The dominant narrative portrays AI as a secular supercycle destined to continue for years without brakes.
In reality, at the microeconomic and supply chain levels, semiconductors follow cyclical logic:
- Client-driven CAPEX
- Orders concentrated in narrow windows
- Rapid capacity saturation
- Pauses for demand to recover
Hyperscalers make investments in blocks. When they decide to build AI data centers, they do so massively, compressing investments into one or more years. Then, once sufficient capacity is in place, they abruptly slow down orders.
This pattern is identical in historical semiconductor cycles: from DRAM to NAND, from mobile to cloud.
The only anomaly is scale. Because for the first time, "cyclical customers" are the most capitalized players on the planet, and precisely for this reason, the leverage effect is amplified. If Microsoft or Amazon decide to slow investment growth, even by just 10-15%, the entire semiconductor sector enters a mini-bear market.
And today these investments are at an all-time high.
The paradox: rising profits, increasing risk}
Nvidia’s record numbers are the result of a dangerous dynamic: the AI sector pushes CAPEX to the maximum, CAPEX fuels Nvidia’s growth, Nvidia beats estimates, sector multiples rise, the market prices in endless growth, Big Tech ramps up investments to avoid being left behind, and CAPEX expands further.
The result? A self-reinforcing cycle that appears linear only in quarterly charts.
The risk is that the market is mistaking this acceleration for structural growth.
And the proof is in the multiples:
- Semiconductor sector P/E consistently above 30x
- Forward multiples incorporating unrealistic growth rates
- Risk-premium compression across the Nasdaq
A significant portion of Nvidia’s performance doesn’t come from "AI adoption across the real economy," but from an infrastructure cycle consistently carried out by the same five companies.
It’s not monetization: it’s pre-monetization investment.
The key point the market ignores
The only question that matters is: are Big Tech monetizing AI investments or are they just investing in the hope of monetizing? As long as the answer remains "they are investing," Nvidia will continue to grow. But if the answer were to become "now we need to monetize what we’ve built," then the trajectory changes:
- CAPEX slows
- GPU demand reverts to trend
- Chipmakers’ data center revenues decline
- Industry multiples compress
- The narrative of infinite growth deflates
Today we’re in a phase where Nvidia’s results tell us more about the balance sheets and strategies of Big Tech companies than about the ultimate success of AI in real life. And this is the paradox: extraordinary profits today, greater fragility tomorrow, if not accompanied by concrete monetization at the end-user level.
So?
Big Tech companies are pushing CAPEX to unprecedented levels, and this, inevitably, favors Nvidia in a momentous way.
But the semiconductor industry remains cyclical by definition, and the current phase of hyper-expansion is supported only by the few companies with cash flows deep enough to afford it.
When the procurement cycle normalizes and AI data centers truly begin to generate tangible profits, then we’ll have the ultimate test: is this massive investment cycle producing sustainable returns or not?
If the answer is no, the entire industry is currently pricing in multiples that don’t reflect the real risk.
It’s not a question of FOMO or fear: it’s a question of understanding that Nvidia’s numbers, impressive as they are, don’t prove AI’s growth is infinite. They merely prove that the build-out phase is still underway.
And when the construction is over, the economic truth always arrives. And it arrives all at once.
Original article published on Money.it Italy 2025-11-20 13:32:38. Original title: Perché gli utili di Nvidia in realtà amplificano la bolla AI
