All Insights

AlphaFold won a Nobel Prize, but its own drug pipeline hasn't dosed a single patient. Meanwhile, a 21-day AI-designed molecule just posted real Phase IIa results in Nature Medicine. A clear-eyed look at what's proven versus still promised.

Equinix, Digital Realty, and Iron Mountain returned 39-45% over the past year — but that still trailed the AI companies whose demand drives it. REITs capture the landlord's slice; the tenant running AI workloads keeps the rest.

China already runs a commercial underwater data center cutting cooling costs 90% and has AI models running on 24 orbiting satellites. Yet combined, underwater and orbital capacity through 2040 stays under 1% of total AI infrastructure — land still wins.

SpaceX debuted at a $2.1 trillion valuation on $18.7B revenue and no profit — with analyst price targets already split between $63 and $165 on day one. What's real (Falcon 9, Starlink) versus what's still a bet (Starship, Mars) after the largest IPO ever.

Tesla trades at 100-200x earnings while automotive gross profit has declined for two straight years. The valuation gap isn't explained by the current business — it's a bet on FSD, robotaxis, and Optimus, none of which meaningfully show up in the financials yet.

$10,000 in VGT grew to ~$241,800 over 20 years; the same amount in VOX reached just $52,600 — a 4.6x gap. There's one live event that could change everything: if OpenAI's IPO classification lands in Communication Services, VOX inherits the world's most valuable AI consumer platform overnight.

Anthropic's revenue model routes cleanly into VGT with no ambiguity. OpenAI's classification is genuinely uncertain — its $100M advertising business is tiny today, but if it scales toward its stated $100B 2030 target, a 2027 reclassification into VOX becomes plausible.

VGT's top 5 holdings make up ~53% of the fund — yet Alphabet, Meta, and Amazon aren't among them, excluded by a GICS classification rule, not an oversight. What "diversified tech exposure" actually means once you look at what's in, and what's out.

Cell and gene therapy generated over $14B in 2024, yet pharma spent $10B+ acquiring pre-commercial companies in just seven months. The "CGT is dead" narrative confuses a broken pricing model with failed science — the biology already works.

Investment banks can't predict whether a drug will work — 50-70% of Phase 2 drugs fail before Phase 3, yet banks price pre-Phase 3 biotech IPOs on a 16-20 week timeline anyway. VC, IB, and PE each play a genuinely different role — only one is actually built to judge the science.