The Signals That Move Stocks Aren't Public
When the future starts in private markets, analysts need new tools. Enter STAX.
Dec 91997 · 2011 · 2025 · Now · Next
Former research analyst, hedge-fund investor, two-time founder, EIR, AWS & Stripe operator — now advising and investing at the intersection of AI infrastructure and capital markets. Writing it all down From the Porch.
From the Porch · No ads · No fluff · Unsubscribe anytime
Read here before you decide whether to subscribe. No obligation.
When the future starts in private markets, analysts need new tools. Enter STAX.
Dec 9And that changes everything about how public markets get their signals.
Dec 15The third time I've felt this feeling about a technology wave. And what I think it means.
Nov 20Ask anything about AI infrastructure, market cycles, deals, or what I'm watching. This AI is trained on my writing and thinks the way I think.
Ask about the STAX framework, the three waves thesis, what I think about specific companies or sectors, my take on the CoreWeave IPO, whether the AI boom is a bubble — anything you'd normally have to email me about.

Each cycle taught me something different. Click an era to explore.
1997–2004
ING Barings and JPMorgan H&Q during the dotcom boom. I was one of those smart-ass inexperienced analysts — Ferragamo tie, big paycheck, bigger ego, no experience, no clue. I covered Internet infrastructure companies, earned Institutional Investor Homerun Hitter in 1998, and watched the whole thing rise and fall from inside the machine. The lesson: cycles are real, timing is not. And the companies that build the actual physical infrastructure — the Equinix types — survive every wave.
2004–2013
MLH Capital and Ardsley Partners on the hedge fund side. Comcast Ventures and Broadway Video Ventures as VC/EIR. Fund investments in Eniac Ventures, Factorial Capital, and Symmetry Peak. Early calls on social, mobile, and cloud computing. This is where I learned the difference between seeing something early and having the conviction to act on it — they're not the same skill.
2010–2014
Tigerbow: real gifts sent to email addresses. Two to three years of blood and fundraising. Failed. Philo: social television check-in. Launched in 2010, pioneered the category, sold the company. Also failed in the sense that mattered. Both post-mortems are public because I think transparency about failure is underrated in a world that mostly celebrates wins. And by the way — I could rebuild both working prototypes in a weekend now. That delta is the thesis.
2014–2023
Co-built the startup team at AWS. When I joined, ~100% of startups ran on cloud but cloud was under 10% of enterprise IT spend. I co-authored the playbook that helped flip that ratio. Then Stripe — same mission, different layer of the stack. Nine years inside two of the defining infrastructure companies of the cloud era. The best seat I've had for understanding how technology gets adopted at scale, and why the gap between startup adoption and enterprise adoption is the most reliable signal in tech.
2023–Now
Porch Capital. Advising and investing at the intersection of AI infrastructure and capital markets. CoreWeave (IPO 2025), Groq (mostly acquired by Nvidia 2025), Crusoe, and a growing roster of companies building the physical and computational layer of the AI era. Writing "From the Porch" for investors, operators, and people who like to get in early. And building STAX — a system for reading startup architecture as a forward curve for public market earnings.
Updated when my views shift. Not a static page.
Software is now model-driven and running directly on GPU-like logic. That shift is forcing a full rebuild of the world's digital infrastructure — from applications down through cloud and networking.
Updated April 2026 · See related posts →
Inference is now continuous and accelerating. Agentic systems — models calling other models — turn that loop exponential, running faster and more often than any human ever could.
Updated April 2026 · See related posts →
Compute is becoming a financeable asset class — more like electricity, oil, railroads, and bandwidth than traditional hardware. Capital is already organizing around it.
Updated April 2026 · See related posts →
AI infrastructure is a sequence of bottlenecks. GPUs gave way to power, to memory, to capital, and the whack-a-mole persists. SRAM systems help HBM shortages, agentic workloads hit CPU-bound orchestration and networking constraints. The opportunity is in spotting what's starting to strain before the market does.
Updated April 2026 · See related posts →
Podcasts, interviews, and coverage — updated regularly.
HBM, SRAM, and the Next AI Bottlenecks
The Information TV (TITV) · Mar 23, 2026
How Startups Signal Tech Adoption — Before Anyone Else
AI Marketers Guild · Jan 28, 2026
Nvidia's Investment in CoreWeave and What It Signals
The Information TV (TITV) · Jan 27, 2026
I work with companies building at the intersection of AI infrastructure and capital markets — at the moments that actually matter.
Public markets react. Private markets predict. And in 2026, the first real signals come from startups — meaning research needs a new way to see the future before it shows up in earnings calls.
Get notified when STAX launches
🔒 Full feed unlocked for subscribers