When investors look at the technology landscape today, they are really staring at the early scaffolding of an industrial revolution built not on steam or electricity, but on artificial intelligence. Nvidia, Broadcom, and AMD are not just chipmakers anymore; they are the architects of the compute backbone upon which every AI application depends. The market caps these companies command today—trillions in combined value—may look extreme to skeptics, but they pale in comparison to what the next decade could bring if the AI “factory” thesis plays out in full. Analysts like Tom Lee have argued that we are still in the first third of the AI cycle, and this framing makes clear why valuations in the tens of trillions are no longer fanciful science fiction, but a plausible destination.
Nvidia sits at the very center of this storm. Its GPUs are effectively the picks and shovels of the AI gold rush, with Blackwell and future architectures already booked years in advance by hyperscalers. AMD, with its MI350 and its growing footprint in cloud deployments, is carving a strong second-mover advantage—its valuation may be smaller, but it rides the same structural wave of demand for compute. Broadcom, often overlooked outside of the semiconductor enthusiast community, provides the interconnects and networking silicon that make AI scale. Without its Jericho and Tomahawk chips, racks of GPUs would remain isolated islands of power rather than being woven into planetary-scale compute fabrics. Together, these firms create the trinity that transforms AI hype into working infrastructure.
The magnitude of demand is difficult to overstate. AI training and inference workloads are growing at rates faster than Moore’s Law could have predicted, leading to what OpenAI’s CFO Sarah Friar called “under-compute”—a chronic shortage of processing power relative to AI’s ambitions. The only solution is massive, relentless capital expenditure by cloud providers, governments, and enterprises. This CAPEX wave is now spilling into the hundreds of billions annually, and the beneficiaries are the companies that design and sell the silicon. If this trajectory continues, it is not inconceivable that Nvidia alone could reach a $10 trillion valuation, with AMD and Broadcom scaling into multi-trillion territory as well. At that point, these firms would rival the combined weight of the entire energy sector in today’s S&P 500.
The danger, of course, is that markets tend to front-run narratives and overshoot in their enthusiasm. AI optimism risks being priced in before revenues fully materialize, creating the conditions for brutal corrections along the way. But just as the internet bubble of 2000 still seeded the trillion-dollar platforms of today, the AI cycle could support far higher sustained valuations than critics assume. The essential difference is that AI is not an optional consumer application; it is becoming the core infrastructure of defense, healthcare, finance, logistics, and even government operations. The stakes are systemic, which is why policymakers in Washington, Brussels, and Beijing are all maneuvering to secure their share of the AI industrial base.
If we zoom forward to 2035, the plausible scenario is that the world’s first $10 trillion company will not be a bank, oil major, or retailer, but a semiconductor firm whose chips power thinking machines. In that light, the current rally in Nvidia, AMD, and Broadcom is not the end of a cycle but its prelude. The market has barely begun to price the scale of transformation that AI will demand, and these companies are positioned to capture it more directly than any others.
