The best investors don’t try to build the engine. They figure out who owns the road it drives on.
Your post nails the first rule of sound investing: stay inside your circle of competence. You’re not claiming to code LLMs. You’re reading phase transitions and capital allocation patterns. That’s exactly where investors belong. History rhymes: railroads, electricity, the internet. The Cambrian Explosion analogy works because adoption accelerates suddenly when conditions align.
The moat test: Castle and picks & shovels
You’re right to focus on picks and shovels, not the hype-y app layer. That’s the “toll bridge” with high switching costs.
GPU makers: Proprietary software ecosystems = high switching costs.
Hyperscalers: Microsoft, Amazon, Alphabet, Meta spending $800B-$900B combined in 2026. That capex is a moat. No startup can match it.
Boring infrastructure: Power, cooling, networking. Less glamorous, but pricing power lives here.
The risk you flagged is spot on: 2026 is the year of revenue realization. Moats mean nothing without cash flow. The shift from infra spend to margin expansion is the real exam.
Management & capital allocation: Who’s swimming naked?
“When the tide goes out…” The hyperscaler CEOs are placing massive bets: Microsoft ∼$190B, Amazon ∼$200B, Meta $125-$145B. Discipline shows up in 3 questions:
Is return on incremental capital > cost of capital?
Do they have more high-return projects than cash? AI capex says yes.
Do they think in decades, not quarters? “Build a kingdom, not just be king” answers that.
Financials: Facts > forecasts
Ignore predictions. Focus on what’s already in the numbers:
Inference costs down 280x from 2022-2024 = deflation → future margin expansion.
AI adoption: 55% → 78% of orgs in one year. Faster than the 1990s internet.
Data centre spend growth: 5% → 29% annually. Real inflection.
The only question that matters now: After $1.4T/year capex by 2030, what does free cash flow look like? That’s why “Year of Proof” and “Show me the free cash flow” are the right frames.
Valuation & temperament
Price is what you pay, value is what you get. Cut the superlatives about “greatest wealth window ever.” You can’t time it. What matters: Can you buy wonderful businesses at fair prices with high certainty their earning power grows 10+ years?
Margin of safety isn’t timing the Cambrian Explosion. It’s owning the infrastructure every winner must rent, at prices that don’t require AI to be perfect.
Temperament is the edge. “We are not spectators, we are positioned, we are patient, disciplined.” The market is a voting machine short term, weighing machine long term. Vote with capital now. Weigh with patience later.
Bottom line
Your thesis is strong because you bet on toll-bridge assets, not speculative apps. For full credibility: 1) Drop timing superlatives, 2) Emphasise FCF and margins over just capex.
Time in the market beats timing the market. If you own businesses that benefit from this buildout and have temperament for compounding, it’s a good time to be an owner, not a trader. Stay the course.
You said 2026 is the “Year of Proof” for AI revenue. Which metric will you watch first to know if the moat is actually producing cash — free cash flow margin, ROIC on incremental capex, or something else? And why?
