Hey Income Traders,

Imagine stepping onto one of those moving walkways at the airport.

At first, you barely notice you’re being carried.

Then you realize you’re moving faster than the people beside you. Then the walkway starts accelerating, and now you’re running just to stay upright.

That’s what’s happening with artificial intelligence right now. And most people haven’t looked down at their feet yet.

I don’t think we’re approaching the singularity. I think we’ve already crossed it. But I don’t even like the word “singularity” anymore. It sounds like a lightning strike, some loud sci-fi moment where everything changes at once. What’s actually happening is quieter than that, and more dangerous.

I call it the Event Horizon.

Because an event horizon isn’t loud. You cross it without realizing it.

And only later do you understand the rules have changed.

When the Cost of Thinking Hits Six Cents

Here’s what changed the rules: the cost of intelligence is collapsing at a rate humanity has never experienced with any input this foundational. Andreessen Horowitz published research showing that the cost of AI inference, the price of getting an answer from a model, has been dropping by roughly 10x every year. Epoch AI’s data is even more dramatic: since January 2024, the median rate of decline has accelerated to 200x per year. To put that in real terms, a query that cost $60 per million tokens when GPT-3 launched in 2021 now costs six cents for equivalent performance. That’s a 1,000x cost collapse in three years.

This isn’t a marginal improvement in a niche technology. This is the underlying engine of problem-solving getting radically cheaper. And when an input that foundational deflates at that speed, it doesn’t stay in one lane. It seeps into everything.

It’s already happening. In drug discovery, AI is compressing preclinical development timelines from three to four years down to 13–18 months. In 2025, the first drug with both target and molecule designed entirely by AI completed Phase IIa trials, an 18-month journey from target identification that would have taken traditional methods years.

In chip design, a startup called Ricursive Intelligence just raised $300 million at a $4 billion valuation to do exactly what the name implies: use AI to design better chips, which train better models, which design better chips.

The founders, who co-created Google’s AlphaChip, are explicitly building the recursive improvement loop that most people still think is theoretical.

The Loop That’s Already Closing

And that’s the part people keep missing. Former Google CEO Eric Schmidt has been telling world leaders that 10–20 percent of the research code at OpenAI and Anthropic is already being written by AI. He calls it recursive self-improvement and says the window is three to six years before these systems are meaningfully learning on their own.

Tyler Cowen, writing just this week, pointed out that OpenAI went from one major Codex release to a dramatically more powerful one in under two months, a pace that used to be six months to a year. His argument: once the next couple of iterations land, the model will be able to update itself even faster. The feedback loop isn’t theoretical. It’s tightening in real time.

Once intelligence helps design better chips, which train better models, which design better robots, which build better factories, the loop closes.

That’s the Event Horizon

We’re entering a period where entire domains won’t slowly evolve.

They’ll flip.

Fields that used to take decades of incremental work could get compressed into months. And our institutions aren’t built for that kind of clock speed. Governance moves in election cycles. Culture absorbs change generationally. Markets price in quarters.

None of that matches a world where the core input to every knowledge-intensive industry is deflating 10x per year.

For anyone paying attention to markets, this means the next three to five years matter more than the next thirty.

The companies building the infrastructure for this shift, the chips, the models, the data centers, the robotics platforms, aren’t speculative bets anymore. They’re the toll booths on a highway that’s about to see traffic nobody planned for. The question isn’t whether this is happening. It’s whether you’re positioned for the speed at which it’s arriving.

This isn’t some distant sci-fi moment. It’s a structural shift in gravity. And from where I’m sitting, the line isn’t ahead of us.

It’s already behind us.

I’m hosting a free webinar today at 4:30 eastern where I’ll break down exactly how I’m positioning for this shift and 9 specific trades I’m watching. Tap here to sign up for The Event Horizon.

May the income be with you,

Hans

Warm Porridge and Where to Hide

Yesterday’s action gave us a classic reversal lower following an early surge after the payrolls report.

Jobs came in weaker than expected… but why is weak jobs data actually good news for stocks?

Here’s the reason.

The jobs report was weak, but not alarmingly so. Bad news that’s mild enough can actually be good news, since it nudges the Fed closer to cutting rates.

This delicate balance is the tightrope we’ll be walking for a while: hoping for more “just right” or “warm porridge” data. Too cold, meaning real cracks in GDP or jobs, and bad news becomes genuinely bad news.

Too hot, and rates stay high, which isn’t great for the market. For now, the market just wants things to stay lukewarm.

What Do I Like and Where Are We Making Money in Turbo Income?

• Amazon (AMZN) & Apple (AAPL): still my go-to places to hide. They’ve lagged the Mag7 rally a bit and have room to catch up. Amazon has the Anthropic kicker, with the AI company now valued at $350 billion and closing in on a $20 billion funding round. AMZN owns an estimated 15 to 19 percent of Anthropic, making it the largest outside stakeholder. Apple continues to dominate the “AI in your pocket” theme with fat margins. I keep saying it, and I won’t stop. Apple haters can keep hating.
• Alphabet (GOOGL): clear runway now that a federal judge rejected the DOJ’s push to force a Chrome divestiture last September. The DOJ filed a cross-appeal in early February to revisit that decision, but that process will take 12 to 18 months to play out, and the market has already shrugged it off. Cloud, Gemini, YouTube, Waymo. This ecosystem is a fortress. Valuation is cheap. I like adding on pullbacks. I put out a nice GOOGL income trade in the Turbo chat yesterday.

The One Name I Love But Won’t Touch Yet

• Nvidia (NVDA): pressure showing up with OpenAI developing its own chip. Cheap valuation on paper, but sentiment is heavy. I love it, but not yet.

The Rate Play That Keeps Paying

• iShares 20+ Year Treasury Bond ETF (TLT), Rocket Companies (RKT), Zillow Group (Z): rate-sensitive plays working and still good places to be.
• Goldman Sachs (GS) and banks: also solid beneficiaries as the curve dynamics play out.

Next week’s inflation data feels like a non-event. The stagflation talk is definitely overblown. With more “warm porridge” data expected, my base case remains a strong year-end rally. We just need to get through a few weeks of choppy, uncertain trading first.

May the income be with you,

Hans