UNSCARCITY

Robots + AI + Cheap Energy → Sustainable Abundance

Edition #5 · February 2026

Agility Robotics just made the humanoid hardware race irrelevant. The American software company announced a standardised operating platform that works across manufacturers — including Chinese ones — turning proprietary robots into interoperable machines. If that sounds familiar, it should. Google did the same thing to smartphones in 2008. Within five years, Android ran on 80% of the world's mobile devices. The humanoid market is about to compress on the same timeline, and almost nobody is pricing that in.

Here's why this is the most important robotics story of the year.

Every humanoid company — Figure, Apptronik, Unitree, Fourier Intelligence — has been building vertically. Custom hardware, custom software, custom integration. That approach made sense when the market was pre-commercial. You can't standardise what doesn't exist yet. But the market crossed into commercial territory over the past eighteen months. BMW, Mercedes, and Amazon all signed deployment contracts. Warehouses and factory floors became the proving ground.

And that proving ground exposed the real bottleneck: not hardware, but software fragmentation.

A warehouse operator running Figure robots on one line and Apptronik units on another needs two separate control systems, two training pipelines, two maintenance protocols. That's expensive. It slows adoption. It locks customers into single vendors. Agility's platform dissolves that friction by providing a common software layer — task planning, safety protocols, sensor integration, fleet management — that sits on top of any compliant hardware.

The Android analogy isn't just convenient. It's structurally precise.

When Google released Android, handset makers stopped competing on operating systems and started competing on price, form factor, and distribution. Samsung, HTC, and Motorola didn't need to build iOS clones. They plugged into Android and focused on what they were good at: manufacturing at scale. Hardware margins compressed. Devices flooded the market. Smartphone penetration went from luxury to ubiquity in under a decade.

Apply that pattern to humanoids. Chinese manufacturers already have a massive cost advantage in hardware production. What they lacked was a sophisticated, safety-certified software stack that enterprise customers in the US and Europe would trust. Agility just handed it to them. Expect Chinese humanoid prices — already aggressive — to drop further as software integration stops being a differentiator and hardware volume becomes the only game.

My prediction: by 2030, a warehouse-ready humanoid with Agility's software stack will cost under $30,000, and three Chinese manufacturers will command over 50% of global unit shipments.

Now, the sceptical read. Isn't Agility handing away its competitive advantage? Why would an American company arm Chinese competitors? Because Agility isn't a hardware company. It makes money on software licensing, fleet management tools, and the data layer that improves robot performance across every deployment. More hardware running Agility software means more data, better models, stickier customers. Google didn't care who made the phone. Agility doesn't care who builds the robot.

The losers in this shift are companies still betting that vertical integration wins. Tesla's Optimus programme, for all its ambition, runs on a fully proprietary stack. That's a strength when you control the factory and the customer relationship — which Tesla does. But it's a weakness in an open-platform world where enterprises want vendor flexibility. Tesla won't adopt Agility's platform. The question is whether enough of the market goes open to make Tesla's walled garden a niche play rather than the default.

For the abundance thesis, this moment is pivotal. Humanoid deployment was always going to depend on cost curves. Hardware costs follow manufacturing scale. Software costs follow standardisation. Agility just fired the starting gun on the software side. When both curves accelerate together, you get what mobile phones got: a technology that starts as a curiosity for rich companies and becomes infrastructure everyone takes for granted.

Consider what happens downstream. Standardised humanoids mean standardised training. A robot taught to pick and pack in an Amazon warehouse in Ohio can transfer those learned behaviours to a logistics centre in Rotterdam running different hardware. Skills become portable. Deployment lead times collapse. The friction between "we bought a robot" and "it's doing useful work" shrinks from months to days.

That's the transition from product to utility. And utilities, by definition, are abundant.

TODAY'S EDITION IS BROUGHT TO YOU BY

Unscarcity

Every week, we break down the three forces reshaping the global economy: robots getting cheaper than labour, AI getting cheaper than thought, and energy getting cheaper than ever. No hype. No hedge. Just the signals that matter. If someone forwarded this to you, now's the time.

Subscribe to Unscarcity →

TRIFECTA UPDATE

Robots: China's overhead charging robots are a quiet masterpiece of practical automation. Engineers deployed ceiling-rail systems in parking garages where robots glide above parked cars and drop down to plug in EVs — no dedicated charger spots, no expensive concrete retrofits, no human intervention. It solves one of electrification's most stubborn constraints: the capital cost of charging infrastructure in dense urban buildings. Instead of wiring every parking spot, you hang one rail and let a robot handle the rest. This isn't flashy humanoid territory. It's purpose-built automation attacking a real bottleneck. When adding one more charging point costs almost nothing, range anxiety becomes a memory, not a debate.

AI: Taalas is doing something genuinely radical — etching AI model weights directly into custom silicon instead of loading them into GPU memory at runtime. The result: inference costs that could drop from dollars per million tokens to cents. If that holds at production scale, the implications cascade everywhere. Real-time AI companions that don't need cloud connections. Continuous monitoring systems that run on milliwatts. Edge devices with frontier-model intelligence baked into the chip itself. Taalas hasn't proven this works at scale yet. But the physics checks out, and the economics are compelling enough that inference-heavy companies should be paying close attention. Cheap inference was already coming. Taalas wants to make it nearly free.

Energy: Equinor just cancelled its blue hydrogen plant in Groningen. Not because of permits. Not because of engineering problems. Because nobody wanted to buy the hydrogen. Zero customers. Blue hydrogen was the fossil fuel industry's transition bet: reform natural gas, capture the carbon, call it clean. But customers skipped straight to green hydrogen or — more often — direct electrification. Why build elaborate hydrogen infrastructure when you can plug in? According to Equinor's own project review, the economics no longer justified construction. The bridge fuel got stranded before the bridge was built. Abundance doesn't wait for halfway solutions.

THE NUMBER

$0.00 per mile.

An EV owner running rooftop solar published twelve months of real charging data showing zero marginal fuel cost. Not nearly zero. Literally zero. One household is a data point. A million of them rewrites the economics of personal transport, grid demand, and utility pricing. When your car's fuel comes from your roof, the old model — centralised generation, monthly bills, scarcity pricing — becomes structurally obsolete.

WHAT I'M WATCHING

Agility's first licensing deals. The platform is announced, but revenue depends on who signs. If a top-five logistics company commits to multi-vendor humanoid fleets running Agility's stack in 2026, the standardisation thesis moves from theory to deployment faster than anyone expects.

Taalas production timelines. Etching weights into silicon is elegant in the lab. Producing chips at volume with competitive yield rates is a different problem. First commercial benchmarks — likely late 2026 — will determine whether this is a real cost revolution or an impressive demo.

EU fuel cell bus wind-down. Deliveries peaked in 2025 and are now declining as battery electric buses dominate on total cost of ownership. Watch which cities manage the hydrogen-to-electric transition cleanly and which ones get stuck nursing stranded fuelling infrastructure through the decade.

THE QUESTION

If humanoid robot software standardises the way smartphone software did, what happens to the companies that bet everything on vertical integration — and what happens to the countries that only manufacture hardware?

UNSCARCITY

A weekly newsletter on the convergence of robots, AI, and cheap energy driving sustainable abundance.

Was this forwarded? Subscribe here.

© 2026 Unscarcity. All rights reserved.