"I expect the ideas here to remain fresh and forward-looking until at least 2030. Beyond that, who knows?" 100x, 2025

I wrote that in 100x. What I was saying is that trends observed at the time are predictable for up to five years, and beyond that, it will need an update. The 100x thinking will continue from there. 

On October 7 I sent my team a line Aleksa Gordić, an AI researcher who used to work at Google DeepMind, had just posted on X: "the 2030s will probably be more different from today than today is from the 19th century." I don't know if he's right, and I don't think anyone does.

In 1900 the average person on Earth could expect to live to about 32. Today it's more than 70 (Our World in Data), and Gordić is saying the 2030s could move further than those 125 years did.

He posted that a day after OpenAI published math research from an unreleased internal model, so some thinking work now runs on a clock of hours, while the industries that run the physical world still change on a scale of years.

We exist to collapse that time, at scale, in the supply chains and operations of manufacturing, energy and infrastructure, from 5–10 year time frames to 1–2 year time frames. But a faster decade won't choose its own direction, and neither will an AI program. Deloitte's State of AI report from October (they surveyed 3,500 leaders in 25 countries, and Deloitte sells AI services itself, so keep that in mind) says 81 percent of companies have seen AI speed up their decisions, but only 15 percent saw a significant shift in decision-making authority. So someone has to decide what all that speed is for. My answer is symbiosis. The sooner we get to transform our current industries to more optimal states, the higher chance we have at real sustainable civilization, which I now simply call the Symbiocene.

In 2023, at an event we ran in Sintra, one of the speakers, a renowned UN advisor, shared the idea: a period we should strive to have after the Anthropocene. For me it was a huge a-ha! Finally! I knew the future I want.

Glenn Albrecht, an Australian environmental philosopher, coined the word in 2011 and built it from the Greek sumbiosis, "companionship".

I understood making life symbiotic would be my own life's purpose. Today symbiosis is the first line in some of our internal documents at Future Works.

At Future Works we see three factions deciding the future of the planet: Nature, Humanity and Thinking Machines. The Symbiocene is a state where the three coexist in a symbiotic balance, and we've built the company around two stages to get there.

The first is between humanity and thinking machines. "What's less optimal than billions of (already general intelligence!) people repeating boring, badly managed, often doubled tasks every day?" I asked that in January, and the first symbiosis is my answer: using the right kind of intelligence for the right task, and freeing up people to live out their real full potential.

In practice, that's how we work with clients today: named senior experts, each orchestrating a fleet of AI agents inside the client's operation, in twelve-week cycles. The experts scope, deliver and sign off.

The second stage brings in nature: we direct the power and resources generated by the first stage toward protecting and enriching it. We haven't reached the second stage yet. 

We sell outcomes. The mission is what drives our team. When an investor asked me in September what Future Works is actually for, I said climate, then spent the rest of the call not saying the word again. We build intelligent systems into the operations of large companies, so they run on less. It shows up on the P&L first, and the emissions drop with it. All it needs is a COO who wants the cost out!

Look wherever you pay for something twice, like trucks driving back empty or inventory sitting in the wrong warehouse. Each one is a cost line and an emissions line. The American Transportation Research Institute's annual report, out in July, puts the per-mile cost of running a truck in 2025 at the highest in the report's history, and says deadhead mileage (trucks driving empty) stayed high. Lasting symbiosis needs sustainability, and sustainability is really just optimization: more optimized means using less resources, using less resources means more sustainable.

There's a catch, though. Today, most people in our industry are very familiar with Jevons' paradox. For those who are not yet familiar, it's from 1865. The economist William Stanley Jevons argued that better steam engines were raising Britain's coal use, because cheaper power kept finding new uses. In his words, from The Coal Question: "It is wholly a confusion of ideas to suppose that the economical use of fuel is equivalent to a diminished consumption. The very contrary is the truth."

Modern economists find the effect is usually partial. For energy, direct rebound runs on the order of 5 to 40 percent of the saving, with most studies at 5 to 25 percent (Gillingham, Rapson and Wagner, 2016). So efficiency rarely backfires completely, but it reliably hands a slice of the saving back to more consumption, and that slice is bigger than anyone budgets for.

AI is a big part of why people talk about Jevons now. In June, Fortune reported that the price of a single token (the unit AI models are priced in) has dropped more than 90 percent since 2023, while spending on large language models has doubled since late 2025, and Apollo's chief economist, Torsten Slok, called it "Jevons paradox in action." Electricity use is climbing too: the IEA has data center demand up 17 percent in 2025, against 3 percent for electricity overall.

AI can do the same to us. Make a factory, a fleet or a grid cheaper to run, and the default is to run more of it. Left alone, the first symbiosis feeds volume and the second one never starts, which is why the second stage has to be a decision somebody makes.

So this is the one test I'd add to every AI program review. Next to every number that says something got faster, put one number that went down for the same output: energy per unit, scrap, freight miles per order, or inventory per unit of sales. Ask Finance to validate it. If speed went up and none of those moved, the gain went into volume.

In twelve-week cycles that number gets checked every twelve weeks, while the program can still change direction. Once Finance has validated it, the saving is yours to point somewhere: back into more volume, or toward the second stage.

So 2030 is when the trends get their update. Gordić might be right about the 2030s, or the physical world might hold out longer than any of us expect. 

What I do know is where I'd like the speed to go. Everyone has their purpose, ours collectively is to bend time until we see the Symbiocene emerge.

So, my friend, a question for your own operation. If it changes more in the next ten years than it did in the last hundred, what's the one thing you want it to use less of by 2035?

Reply here. I'd love to hear from you.

Much Love,
Matt