The Easy Money Was Made in Panic

Welcome 👋🏽

1. The AI Trade is Growing Up

2. NASDAQ has Already Started to Believe Again

3. Risk-On Risk-Off (RORO) Tool

4. The Most Important Chart

5. Capital Deployment Framework

6. What 5 February Actually Told Us

7. Current State Of The CDF Indicators

I’ve tried to keep this one as simple as possible, because I think the market is currently giving us three very clear signals:

1. AI demand is no longer theoretical

2. Risk assets are absorbing bad news better than they were earlier this year

3. My framework still says Risk-On, but that does not mean blindly chasing every green candle like an intoxicated degen at a casino on a Friday night

For context, this note builds on these major pieces:

• Latest deployment (14 April 2026)

• My base case (is a short-duration conflict, a temporary oil and gas spike, markets digest it, and we move on).

• The bottom is likely in for crypto (except if there is a major existential event)

• Why 5 Feb 2026 was THE buying opportunity (which we capitalised on)

• 2026 Macro Outlook

The simplest way I’d frame the regime we’re in is this:

The market is no longer just pricing AI hype. It is starting to price real AI revenue, real infrastructure demand, real energy constraints, and real capital flowing into the companies that can actually deliver it.

That matters for tech, energy, commodities and markets.

One investor sees “inflation”, while the other sees a “ deflationary boom” - which one is right?

And yes, it matters for crypto, because crypto tends to do very well when liquidity, risk appetite and technological speculation are all moving in the same direction (hint: they’re moving in the same direction!)

Let me reiterate what I laid out last week.

THE trade for the next several years is the AI buildout.

Remember, at first (2023-2025) the market was hyped about AI.

Then it got scared that the hyperscalers (Google, Meta, Amazon, Microsoft etc) were lighting free cash flow on fire chasing something that may never generate a return.

Fair. It reminded the oldies of the dot com bubble where we got “dark fiber” (unused optical fiber cables that were laid but never “lit up” with equipment to carry data).

Today we see actual usage, actual revenue, actual customers and actual constraints.

First, we are seeing other income growth across hyperscalers 👇🏽

Second, we are seeing the annualised run-rate revenue of companies like Anthropic go vertical.

If AI was just another dot com or tulip bubble, revenue would disappoint, CAPEX wouldn’t yield “other income”, compute demand would soften, and the whole trade would start looking like the 2021 metaverse all over again.

Instead, you can’t rent a Blackwell Chip (B200) if your life depended on it.

Instead, we are seeing:

• AI labs need more compute

• hyperscalers need more data centres

• data centres need more power

• power needs more grid capacity

• grid capacity takes years, not quarters

That is the bottleneck.

And when the bottleneck is physical infrastructure, the trade does not end because someone on X says “bubble”.

It ends when capacity catches up with demand, or when demand disappoints.

Right now, neither of those things appear to be happening, so… all aboard the AI train!

Back to Top 👆🏽

We have now rallied off the lows, with NDX ripping to new highs after ~10 months of accumulation.

We have since had a small pullback. To be honest, after a move like that, I would not be shocked to see a -10% move.

and when we see those sorts of moves, we’ll get the skeptics saying

it’s all over.

I won’t be paying any attention to these folks, because as an avid user of the frontier agentic models - and I mean proper agentic workflows through Claude, Codex, Replit or similar tools - you can clearly feel the direction of travel.

• I can one-shot CRM systems.

• I can one-shot indicators.

• I can one-shot web-apps.

• I can one-shot critical analysis.

and… this is the worst this technology will ever be.

So, what are our tools telling us? 👇🏽