SCOTT NEWTON.

Strategic bets / September 2026

Here’s where
I’m placing my bets.

I don’t buy the idea that AI is all hype. I think we’re seeing a change in what people and companies can do, and I’m willing to put my own expectations on the record.

First recorded here: 11 September 2026 · Updated 15 September 2026 · Personal views

These are my bets, not promises about what will happen. The dates matter. I’ll record changes as my thinking develops. On 15 September, I removed the June 2028 ASI deadline from these bets and replaced it with my longer-term vision for Kairo. My current thinking on ASI is here. Read the reasoning
Now

My current position

AGI, by my functional definition

I think we have effectively crossed the threshold: systems with broad cognitive capability that can use tools, operate computers, research, code, communicate and pursue goals across domains. That is my definition and my assessment, not a claim of universal agreement.

Late 2026

75% by December · My subjective estimate

A political and social “red alert” on AI and employment

My bet is that disruption to computer-based work becomes much harder for politicians and the public to ignore by the end of 2026. I put my own probability at 75%. If it takes longer, I expect the alarm to be much louder by mid-2027.

2027

My current bet

A major financial correction, while AI keeps improving

I expect a substantial correction in AI-related valuations and investment expectations. The problem won’t be that the technology is fake. It will be that spending and expectations have run ahead of the financial returns.

2027

Economic consequence I expect to watch

Recessionary pressure from an AI correction

I think an AI-led market correction could create serious pressure on the US economy and spill over globally. An outright recession is part of the scenario I’m concerned about, rather than something I regard as guaranteed.

2028

My infrastructure timing bet

Delayed infrastructure becomes economically important

I expect a significant wave of capacity from the current buildout to become useful around 2028, arriving alongside better models, hardware, agent systems and inference efficiency.

After a correction

Conditional on the correction scenario

A renewed AI investment cycle

I expect a reinflation rather than a permanent collapse. The next expansion could be driven increasingly by demonstrated economic work, with capability continuing to develop through the correction.

Towards 2030

My longer-term vision for Kairo

AI that helps you navigate your possible futures

My bet is that AI will use a deep, evolving understanding of your ambitions and circumstances to explore thousands of possible routes, then keep adapting them as your life and the world change. Making this accessible within an ordinary subscription depends on the cost of useful AI computation falling.

Read the full thinking