At Davos, Nvidia (NVDA) CEO Jensen Huang didn’t just talk his book; he effectively redrew the global capex budget for the next decade, arguing that the world is only a “few hundred billion” dollars into what will be a trillions‑of‑dollars AI infrastructure buildout. He framed the surge as “the largest infrastructure buildout in history,” a supercycle that stretches from power grids to data centers and all the way up to application developers.
Huang’s point is simple but market‑moving: the AI trade isn’t about one chip upgrade cycle, it’s a long-haul re-tooling of the digital and physical economy. Investors fretting about an “AI bubble,” he suggested, may be confusing a long-duration capex cycle with a short-lived mania.
The Five‑Layer AI “Cake” Wall Street Has to Model
Huang likens AI to a five‑layer “cake”: at the base is energy, then chips, cloud infrastructure, AI models, and finally the application layer where profits are most visible. Each layer has to scale in sync, which means this is less a story about one hero stock and more a coordinated global buildout.
Consultants and banks are now putting numbers to that cake. McKinsey estimates that data centers alone will require nearly 7 trillion dollars of capex by 2030, with about 5.2 trillion targeted specifically at AI‑ready facilities. J.P. Morgan ( JPM) projects AI infrastructure spending could reach 1.4 trillion dollars per year by 2030, underscoring how quickly this line item could rival traditional industrial capital outlays.
From Capex Line Item to Global Construction Boom
For once, Wall Street’s favorite growth story is also a boon for people in hard hats. Analysts describe AI infrastructure as the largest infrastructure supercycle in modern history, with 2026 capex on AI data centers and associated power needs projected at roughly 602 billion dollars, up more than a third from 2025. Moody’s sees at least 3 trillion dollars in data center investment over the next five years, driven by hyperscalers racing to add capacity.
That spending doesn’t just buy GPUs; it hires electricians, engineers, and construction crews. Recent analysis suggests AI‑driven data center construction could create a wave of six‑figure technical and construction jobs as capacity scales, turning what used to be a niche real‑estate segment into a macro‑relevant employer. If the last tech boom minted software millionaires, this one may do the same for professionals who know their way around megawatts and chilled water.
GPUs as the New “Toll Booths” of the Data Economy
Underneath the cranes and concrete, Nvidia remains the dominant toll collector. GPUs now represent roughly 39% of total data center spending, making them the single biggest cost driver in AI infrastructure builds. Nvidia controls an estimated 92% of the AI data center GPU market, a position reinforced by a rapid product cadence that keeps each new architecture meaningfully ahead of rivals.
That dominance has prompted some bolder forecasts. One analysis pegs AI capex climbing from around 600 billion dollars in 2025 toward 3–4 trillion by 2030, arguing that Nvidia’s role as the indispensable compute layer could support a path toward a 10 trillion dollar market capitalization if the cycle runs its course. On that math, every incremental dollar of AI data center spend is less a headline and more a recurring toll payment.
Is This an AI Bubble or the New Baseline?
Talk of trillions tends to summon ghosts of the dot‑com era, and skeptics warn that outsized data center and chip budgets could eventually collide with more sober demand. Huang’s rebuttal leans on two realities: first, AI workloads are rapidly becoming foundational across sectors from healthcare and finance to manufacturing; second, we’re still early in building the “application layer” that ultimately monetizes that compute.
Forecasts back up that shift. McKinsey estimates AI‑related workloads could command the majority of data center capacity by 2030, with AI‑optimized servers driving most of the new power demand. J.P. Morgan’s view that AI infrastructure could triple to 1.4 trillion dollars annually by the end of the decade reframes the current spending spike less as a bubble and more as the new baseline for digital-era infrastructure. If Huang is right, the real risk for investors isn’t that the AI buildout is too big—it’s that their models still assume it’s temporary.
The Sources
[1] Nvidia CEO Jensen Huang: ‘Trillions of dollars of AI infrastructure … https://finance.yahoo.com/news/nvidia-ceo-jensen-huang-trillions-of-dollars-of-ai-infrastructure-needs-to-be-built-144212721.html
[2] What Bubble? Nvidia CEO Says AI Needs Trillions More in … https://finance.yahoo.com/news/bubble-nvidia-ceo-says-ai-225727598.html
[3] Nvidia CEO says AI boom is fueling the ‘largest’ infrastructure … https://finance.yahoo.com/news/nvidia-ceo-says-ai-boom-154207631.html
[4] The cost of compute: A $7 trillion race to scale data centers – McKinsey https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-cost-of-compute-a-7-trillion-dollar-race-to-scale-data-centers
[5] 13 Data Center Growth Projections That Will Shape 2026-2030 https://avidsolutionsinc.com/13-data-center-growth-projections-that-will-shape-2026-2030/
[6] AI Infrastructure Could Triple to $1.4 Trillion by 2030 – Nasdaq https://www.nasdaq.com/articles/ai-infrastructure-could-triple-14-trillion-2030-heres-best-stock-buy-2026
[7] Whether To Follow $602 Billion Flowing To AI Data Centers In 2026 https://www.forbes.com/sites/petercohan/2026/01/23/whether-to-follow-602-billion-flowing-to-ai-data-centers-in-2026/
[8] Moody’s sees $3T in data center spending by 2030 | Construction Dive https://www.constructiondive.com/news/moodys-data-center-spending-2031/810055/
[9] Nvidia CEO says AI boom will create ‘six-figure’ construction jobs https://finance.yahoo.com/news/nvidia-ceo-says-ai-boom-172142457.html
[10] Nvidia’s S-Curve: The Physical Buildout Driving Exponential Demand https://www.ainvest.com/news/nvidia-curve-physical-buildout-driving-exponential-demand-2601/
[11] Is Nvidia Still a Buy in 2026? Can NVDA Reach a $10 Trillion … https://www.tradingkey.com/analysis/stocks/us-stocks/261505954-nvda-nvidia-stock-price-2026-2030-tradingkey
[12] Nvidia CEO Jensen Huang: AI infrastructure buildout is ‘sensible’ https://finance.yahoo.com/video/nvidia-ceo-jensen-huang-ai-143500353.html
[13] Trillions of dollars of AI infrastructure needs to be built https://x.com/YahooFinance/status/2015184375772573974
[14] Nvidia CEO Jensen Huang says AI buildout still needs trillions of … https://www.youtube.com/watch?v=3Y8PvfUwaEw
[15] Nvidia CEO urges continued spending into AI: Top takeaways https://finance.yahoo.com/video/nvidia-ceo-urges-continued-spending-213600478.html
