Being early Is now the only thing to be

Tim O'Reilly once said that being too early is indistinguishable from being wrong. We know this from personal experience. One of us invented responsive web design a decade before it mattered. The other invented VRML thirty years before the Meta Quest.

Being early used to cost you years. The gap between when something emerged and when it mattered could be measured in decades. You could be directionally right and still lose, because the world moved slowly enough that the consensus had time to form without you.

That's no longer true.

DeepSeek went from paper to market shock in weeks. MCP went from spec to universal standard in months. This week at GTC, Jensen Huang described a revenue trajectory for NVIDIA — $500 billion to $1 trillion — that dwarfs Apple. Overnight, Stripe launched an open standard for agents to pay each other. Data centre construction has overtaken office construction. And Apple started blocking vibe coding apps from the App Store.

All of that happened in the span of a few days.

Being early is no longer the same as being wrong. In fact, being early is the only time to be.

What NOOPS is

For fourteen months, the two of us — John Allsopp and Mark Pesce — have been doing this work. Every day. Sharing a dozen or more links via iMessage with running commentary. Going on fortnightly walks along the Cooks River, recorded, going deep on whatever's been accumulating. Building things with these tools in production. Watching what breaks, what holds, and what disappears.

We've been doing the work of a research operation without calling it one.

Today we're calling it one.

NOOPS is a strategic advisory in the shape of a real-time intelligence service tracking the AI transformation as it happens — not after the fact, not in quarterly reports, not in conference keynotes and decks that arrive months after the window has closed. We track the signals that move markets, identify the theses that explain them, and deliver analysis at the speed the transformation is actually happening.

Long infra. Short spoon.

Those four words describe our investment thesis. It's not complicated. But there is a lot of nuance.

Infrastructure is the layer that doesn't go away when AI rewrites the rules. Compute. Semiconductors. Agent runtimes. Payment rails. Data gravity. Identity systems. Trust. These are the things that become more valuable as AI capability increases.

Spoons are the things built on top — the applications, platforms, and services that can be vibed into existence by anyone with enough focused effort, evaporating all of their value.

What does that look like? Ask a copywriter. Oh wait, you can't.

That's what it looks like: now you see it, now you don't.

We track this in real time. Right now, our SaaS Apocalypse thesis is playing out in the stock market: Atlassian is down 51% year to date. Adobe down 25%. Salesforce down 24%. These aren't random selloffs — they're the market starting to price in what we've been saying since January: the application layer is dissolving.

Meanwhile, Micron is up 35% on AI memory demand. NVIDIA just described a future where it's four times the size of Apple. The infrastructure layer is hardening.

Long infra. Short spoon.

Why us

We're not analysts in the traditional sense. We're practitioners. We build with these tools every day. We see the cracks in a platform's model because we hit them in production. We see what an emerging tool actually does because we're using it, not reading the press release.

Between us, we have eight decades of building professionally with technology. John founded Web Directions and the AI Engineer Conference. He wrote "A Dao of Web Design." He originated the ideas behind Adobe Fonts. Mark invented VRML, wrote the books on the Metaverse, and has spent four decades at the intersection of technology, culture, and the market forces that connect them.

But the real credential is simpler than that. We've been tracking this transformation daily, together, for over a year. Thousands of links shared, interpreted, connected. The DeepSeek efficiency thesis identified in January 2025 — before the NVIDIA selloff. MCP's emergence as the standard spotted in early 2025. The SaaS pricing collapse tracked in real time. Anthropic's strategic positioning understood months before the market caught up.

And one more thing: NOOPS is a product of the AI-augmented workflows we analyse. Our conversations become structured analysis through an automated pipeline involving Claude, Obsidian, and a daily signal processing system. The production process is itself a proof of thesis. No other research service can make that claim without proving our point.

What you get

We publish at four speeds, because the transformation doesn't move at one tempo.

Daily signals — short, interpretive observations. The repos gaining traction. The token flows shifting. The earnings calls that confirm or challenge our theses. Not news — signal.

Weekly roundup — we step back from the daily flow and make sense of it. What emerged, what connected, what shifted. The context you won't find anywhere else, delivered while it's still actionable.

Monthly deep dives — when a signal resolves into a trend, we write the complete analysis. Full evidence, full working, what we got right, what we missed.

Quarterly salon — a curated gathering in Sydney and Melbourne. Practitioners, founders, investors. The conversation you wish you were having.

We also maintain a live signal dashboard tracking the stocks, repos, models, and data points feeding our theses. Updated daily.

Why today

We're launching into a market panic. The SaaS stocks we track are deep in the red. The broader market is selling off. Jensen Huang just put a number on the infrastructure buildout that makes the internet boom look miniscule. Stripe just gave agents the ability to pay each other. And Apple just started trying to close the gates - long after the horse bolted. (Samsung encourages vibing Galaxy apps.)

Every one of those is a signal. Every one of them connects to a thesis. And the window between the signal appearing and the market pricing it in is narrower than it's ever been: Not months. Minutes.

If your analysis shows up after the window closes, it's not analysis. It's a history lesson.

We'd rather be early.

John Allsopp & Mark Pesce Sydney, 19 March 2026