Been thinking about where to really concentrate capital right now, and honestly the macro picture keeps pushing me back to the same two names in the semiconductor space. Both have that rare combination of explosive current momentum plus multi-year visibility that's hard to ignore.

Let's start with what's actually happening in the market. We're seeing AI shift from this experimental phase into real production workloads. That's the inflection point everyone's been waiting for, and the infrastructure plays are the ones capturing that transition. Companies aren't just testing anymore—they're deploying at scale, which means sustained capex cycles.

Nvidia is the obvious one, but the numbers actually justify the attention. Their latest quarterly showed $68.1 billion in revenue, up 73% year-over-year. More importantly, data center hit $62.3B, growing 75% annually. But here's what caught my eye: management's highlighting that demand isn't just from training anymore. The real money is flowing into inference workloads—actually running AI in production. That's customers generating revenue from these systems, which completely changes the durability of demand.

They're also expanding beyond just selling chips. Networking revenue crossed $11 billion in the most recent quarter, and they're moving toward rack-scale systems. When you're not selling standalone components but integrated solutions, you lock in customer dependency. That's the kind of moat that compounds over time. Plus they generated $97 billion in free cash flow last year, so they have the firepower to stay ahead on R&D.

The forward guidance is interesting too—$78 billion expected for the next quarter. That's not some pie-in-the-sky projection; it's anchored in actual customer commitments for Blackwell platforms.

On the other side, TSMC is the manufacturing backbone for all of this. They control over 70% of the advanced foundry market, and the AI wave is flowing directly through their fabs. Their most recent quarter showed 25.6% year-over-year growth to $33.1 billion in revenue, with operating margins hitting 54%. That's the kind of margin profile you see when supply and demand are completely out of balance in your favor.

What's notable is the product mix. Advanced nodes (7-nanometer and below) represent 74% of their wafer revenue. Three-nanometer alone was 24% of output last year. More importantly, they just started high-volume production on 2-nanometer in Q4, and they're planning an aggressive ramp throughout this year. Management is actually expecting the 2-nanometer ramp to be bigger than what they saw with 3-nanometer—that's saying something.

The guidance for next quarter is $34.6 to $35.8 billion, implying 38% growth at the midpoint. Given the 2-nanometer scaling curve, that seems totally achievable. They're also projecting AI accelerator revenues to compound at mid-to-high 50% annually through 2029, with total company revenue growing at 25% CAGR over that window.

The reason I keep coming back to both of these is that we're at this rare moment where the secular tailwind is accelerating rather than plateauing. AI isn't slowing down—if anything, the move from experimentation to production is just beginning. Both companies are positioned at the top of their respective value chains, and both have the financial muscle to invest ahead of demand.

If you're looking at the semiconductor sector right now, these are the two names that have the double advantage of current growth plus credible multi-year visibility. The infrastructure buildout for AI is just getting started, and these are the companies capturing that cycle.