Issue #023 — Weekend Notes

Published 2026-08-08 · Updated 2026-08-17

I've been staring at the same coffee stain on my desk for twenty minutes. It's shaped vaguely like a bipedal robot stumbling under the weight of a 50lb tote bag. Or maybe that's just my brain projecting after the week I've had.

Usually, when I sit down to write these notes, I look for the "Big Story." The shiny thing. The multi-billion dollar valuation that makes everyone gasp and check their portfolio allocations. But this weekend? I can't stop thinking about the unsexy stuff. Cardboard boxes, warehouses, and the distinct lack of humanoid robots actually doing the dishes in my apartment.

The Anti-Hype IPO

It started on Tuesday. I was doomscrolling through the usual hype cycle — "General Purpose Robots are the Next iPhone," "Sentience by 2029" — when I landed on the news that Agility Robotics is going public. Via a SPAC, which feels very 2021 of them, but let's park that (The Robot Report, June 25, 2026; TechCrunch, June 24, 2026).

The headline wasn't what grabbed me. It was the positioning from their CEO. In a world where every founder is trying to sell you a robot butler to fold your laundry, Agility's Damion Shelton — co-founder, now chairman; note the company has since brought in new leadership — has been notably "focused on the here and now," and the company's public messaging explicitly declines to promise a home robot anytime soon.

I paused. I re-read it. I actually laughed out loud.

Here is a company spun out of Oregon State University's Dynamic Robotics Lab in 2015 (per Wikipedia and the SPAC coverage), planning a SPAC merger at a $2.5B valuation and expecting roughly $620 million in proceeds (updated per FinBrain 2026-08-19), and its pitch to Wall Street is essentially: "We are going to move boxes in warehouses."

It feels counter-intuitive, doesn't it? We're constantly sold the Westworld-style servant. But Agility is betting the farm — literally, its public market survival — on the most boring application imaginable: logistics.

And I think they might be the only ones in the room who are sober.

The Pattern Forming

You can't throw a rock in Silicon Valley without hitting a "humanoid" demo reel. Robots walking in parks, picking up apples, making coffee. It's captivating. It's viral content. Meanwhile, Xiaomi's Lei Jun was posting robot work videos this week that were, tellingly, framed around a robot working rather than a robot performing (gnews-zh, Aug 16–17, 2026).

And then there's Avatar Robotics, which raised a $6.5M seed round — small money, but note the angle: "addressing industrial labor constraints," with humanoids and teleoperation for supply chains (The Robot Report). See the pattern?

We are slowly, painfully, waking up from the "General Purpose" hangover. For years the thesis was: build a body that looks human, shove an LLM into its head, and boom. It's a beautiful thesis. It's also incredibly hard.

If you are a manufacturing plant manager, you don't care if your robot has two arms and two legs because it's anthropomorphic. You care if it can move 500 pallets an hour for less than it costs to pay Dave, who calls in sick on Mondays. If the robot costs $100,000 and needs a PhD student to reset it every time it trips over a loose cable, you're sticking with Dave — or buying a standard automated forklift. Unit economics, not anthropomorphism, wins factory floors.

Geofencing the Humanoid

This brings me back to Agility and their anti-hype hype. In July, Agility opened a new Fremont, California facility explicitly designed to accelerate its "Physical AI" development (Agility press release / PRNewswire, July 16, 2026) — Tesla's backyard, but they aren't trying to sell a car. They're training Digit — the humanoid they've been shipping in some form since 2017, originally to academic labs, now aimed squarely at industrial tote-handling — to be a warehouse cog.

Because the home is a hostile, chaotic environment. Stairs, pets, clutter, narrow doorways. A physics nightmare. But a warehouse? Flat floors, standardized boxes, low variability. It's the self-driving lesson all over again: we thought we'd have Level 5 everywhere by 2020, then companies figured out geofencing was the only way to make the math work.

Agility is essentially geofencing the humanoid: dominate the warehouse, be the best at moving totes, actually ship product and generate revenue. The "boring" strategy that builds empires.

Dirty Jobs Are Getting Funded

The shift isn't confined to logistics. The University of Florida just launched a robotics lab dedicated to industrialized construction — alongside what The Robot Report calls the world's first Industrialized Construction degree. Construction robots. Industrial humanoids. Not the Jetsons. Concrete, rebar, heavy lifting.

That's the shift I'm seeing: the money and the brains are moving from "Sci-Fi Fantasy" to "Dirty Jobs."

The SPAC structure matters here because it forces transparency that VC funding doesn't. When you're public, you have to show earnings and a path to profitability. Agility betting on logistics suggests they've run the numbers on the home-robot market and the margins aren't there yet. The CAC of convincing a suburban family to trust a $30k robot with their fine china is astronomical. The CAC of convincing Amazon to put a robot in a depot is an Excel spreadsheet with ROI calculations.

Don't get me wrong — I still want a robot butler. And Avatar's teleoperation approach, using remote human pilots to bridge the gap between "AI isn't ready" and "we need labor now" while collecting training data, is a clever stepping stone.

Vertical Humanoids

But the "General Purpose" narrative is fracturing. We're entering the era of Vertical Humanoids: the Warehouse Humanoid, the Construction Humanoid, the Hazmat Humanoid. I wonder if we'll even call them "humanoids" in five years. Once they're covered in dust, holding rivet guns or stacking pallets, they just become... machinery.

That's harder to pitch a movie about. But for VCs tired of burn rates with no revenue, it should sound like music.

So here we are in mid-2026. The hype cycle is cooling, replaced by the cold splash of water that is commercialization. Agility is going public not because it solved general intelligence, but because it solved the tote-moving problem. That's worth more right now than a robot that can walk a dog.

I'm going to keep watching Fremont. If Agility can actually scale this — prove a biped is economically superior to a wheeled conveyor in specific use cases — it will be the biggest validation this industry has seen in years. Not because it's cool, but because it works.

I'm placing my bets on the warehouses. The floors are flat, the coffee is bad, and the robots don't mind the long hours.

Catch you next week.

— Ken