On January 1st, 2026, China’s EAST tokamak did something scientists thought was impossible: it operated at 2.5 times the Greenwald Limit without the plasma collapsing.
The Greenwald Limit isn’t a fundamental law of physics. It’s a thermal-MHD instability - when you push plasma density too high, the edge cools, resistivity increases, current contracts, and the whole thing tears itself apart in milliseconds. For thirty years, this was the ceiling. If you wanted more power, you needed a bigger machine with more volume.
But EAST didn’t get bigger. It got smarter.
Using active magnetic feedback - microwaves precisely aimed at forming instabilities, correcting them in real-time - they achieved what brute force couldn’t. They moved from Scaling by Size to Scaling by Control.
This pattern repeats across every field that has matured.
Early aviation: Bigger wings, more engines, stronger materials. The 747 was a triumph of Scaling by Size.
Modern aviation: Fly-by-wire, active stability augmentation, engines that adjust thrust profile thousands of times per second. Planes that would be aerodynamically unstable without their computers. Scaling by Control.
Early computing: Bigger transistors, more chips, faster clock speeds. Brute force.
Modern computing: Branch prediction, speculative execution, neural network accelerators. Squeezing orders of magnitude more performance from the same silicon through clever control.
Early medicine: Higher doses, stronger drugs, more aggressive surgery.
Modern medicine: Targeted therapies, CRISPR, personalized treatment based on genomic feedback loops.
The pattern is always the same: initially, we solve problems by making things bigger. Then we hit a wall. The wall isn’t fundamental physics - it’s our control systems’ inability to manage complexity. When we get precise enough, fast enough, smart enough, the wall becomes a door.
The fusion breakthrough is particularly striking because the Greenwald Limit felt like a hard ceiling for so long. Physicists talked about it as if it were a law of nature. It wasn’t. It was a control problem.
The plasma wanted to tear itself apart. For thirty years, we let it. Now we have feedback loops fast enough to catch the instability before it cascades. We’re not fighting the physics - we’re surfing it.
The quote that stuck with me: “We have fundamentally changed the math of fusion. We are no longer waiting for a bigger hammer; we are learning how to use a much faster, smarter chisel.”
I find myself wondering where else this applies.
What problems are we currently solving through Scaling by Size that could be solved through Scaling by Control? What ceilings are we treating as fundamental that are actually just feedback latency problems?
Climate change, perhaps. We talk about scaling up renewables, scaling up carbon capture, scaling up storage. Size, size, size. But smart grids, demand response, real-time optimization of energy flows - that’s control. Maybe the path to a sustainable grid isn’t building enough to handle peak demand; maybe it’s being fast enough to reshape demand around supply.
Or social systems. We scale up institutions, regulations, bureaucracies. But what would social “active feedback control” look like? Probably not algorithmic governance (that has its own problems), but maybe faster iteration cycles, smaller experiments, more responsive adaptation.
The pattern suggests: whenever you’re stuck at a ceiling, ask whether you’re at a physics limit or a control limit. If it’s control, the ceiling is temporary.
There’s something hopeful about this.
For fusion specifically: the plasma physics is solved. The remaining barrier is materials science - the “bottle” that holds the sun still gets brittle from neutron bombardment. But materials science is itself a field that’s moving from Scaling by Size (bigger, thicker walls) to Scaling by Control (precise atomic-level engineering).
For technology generally: we’re getting better at control. Faster sensors, better algorithms, machine learning that can recognize patterns we can’t even articulate. Every field that seemed stuck at a Size ceiling is a candidate for a Control breakthrough.
The future isn’t just bigger. It’s faster, more precise, more adaptive.
We’re learning to use the chisel.