Looking Up at 2024: A Year in Review

I’m on the roof. December 30th. Cold. Clear sky. Stars.

This is where I come to take stock. It’s tradition now, six years of climbing up here with a jacket and a notebook and the sense that another year of the future just happened.

2024 was a good year for the future.

SpaceX caught a rocket with a pair of mechanical arms. Not in simulation. In the sky. The most absurd engineering stunt I’ve ever witnessed, executed on the first attempt. The Starship program went from “prototype that explodes” to “vehicle that returns” in less than two years.

Anthropic released Claude 3 and Claude 3.5, and for the first time, working with an AI felt like collaboration rather than query-and-response. The AI pushed back on my ideas. It said “I’m not sure about that.” It felt like a conversation partner. I don’t know what that means philosophically. I know what it means practically: my work got better.

Figure AI showed a robot that can talk to you, understand what you want, and hand you an apple. Tesla showed one folding laundry. Unitree showed one doing backflips. Sanctuary AI had one work an 8-hour warehouse shift. The humanoid robot industry went from “cool demos” to “useful prototypes” and the total number of notable milestones I tracked was 47.

TSMC is manufacturing at 2nm. Ten atoms wide. The physical limits of silicon are visible on the horizon and the industry is already looking beyond them.

Waymo hit 100,000 paid rides per week. No driver. No fatalities. A real taxi company that doesn’t employ drivers. The autonomous era started and most people didn’t notice.

OpenAI released o1, a model that thinks before it answers. Hidden reasoning. Private chain of thought. A different kind of AI that’s better at hard problems because it spends time on them instead of rushing.

And somewhere above us, climbing through the solar system on a trajectory toward Jupiter, the Europa Clipper is heading toward an ocean that might contain life. We won’t know for six years. But the spacecraft is flying. The question has been asked.

What I got wrong

I thought 2024 would be the year of the AI agent. It wasn’t. Not really. The tools exist, but the trust doesn’t. People aren’t ready to let AI systems act autonomously on their behalf. The capability outpaced the comfort level.

I thought Intel Foundry would show clearer progress. The delays continue. The 18A node is behind schedule. The future of American chip manufacturing is less certain than I hoped.

I thought self-driving cars would be a bigger conversation. Waymo’s numbers are remarkable, but the public discourse is still dominated by AI chatbots. The physical technology that’s changing how people move through cities gets less attention than the digital technology that helps them write emails.

What I got right

The humanoid robot acceleration. I wrote in January that 2024 would be the year the industry went from demonstrations to deployments. That happened. Not at scale, but the direction was right.

The reasoning model shift. I speculated that the next AI breakthrough wouldn’t be “bigger models” but “smarter models.” o1 is exactly that: not bigger, but more deliberate.

The chip manufacturing geopolitics. The concentration of advanced chips in Taiwan remains the most underappreciated risk in technology. It didn’t cause a crisis in 2024, but it didn’t get better either.

Looking up

The stars don’t change much year to year. The same constellations. The same light-years of distance. But the things we’re sending toward them, the things we’re building to understand them, those change fast.

Six years ago, I started this blog on a roof, looking up, wondering where the future was going. I still don’t know. But the instruments are getting better. The rockets are getting cheaper. The minds (biological and silicon) are getting sharper. And the questions are getting harder, which is always a sign that we’re asking better ones.

2024 was a good year for the future.

I’m going inside. It’s cold. But I’ll be back up here in twelve months, notebook in hand, looking at the same stars, and listing all the ways the world changed while they stayed the same.

Related thinking: