YIMBYLAND

Co-Founder @americanhousing.
Rebuilding the American Dream.

[YIMBYLAND @YIMBYLAND]

Last year I decided I was no longer satisfied just posting about housing.

So I started a company with 3 other incredible dudes with the mission to build missing-middle housing at scale.

Introducing... The American Housing Corporation

It’s time to rebuild the American Dream.

[The American Housing Corporation @americanhousing]

For a half-century, America has critically under-built family-sized housing in our most dynamic cities and neighborhoods, rendering them childless and unaffordable.

It's time to save the American Dream.

Introducing The American Housing Corporation.

[Extend & Pretend Cowboy @Cashflow_Cowboy]

Cool things happening in Austin. Only going to get better

[YIMBYLAND @YIMBYLAND]

@chrishume_ We’re going to build so many homes.

[YIMBYLAND @YIMBYLAND]

NIMBYs when they experience the horrific violence of an… apartment building.

[Boom 💥 @LoveCodeTrade]

San Francisco: If you look at this and cannot empathize with the people who will be harmed by the addition of this monstrosity...

you're an f'ing sociopath!

[YIMBYLAND @YIMBYLAND]

@kgast99 🙏

[YIMBYLAND @YIMBYLAND]

9 months from launch to first machine in our 1,200 home/year factory.

We’re going to build so many homes and I’m genuinely having the funnest time of my life doing it.

[The American Housing Corporation @americanhousing]

We're turning 225,000 sqft of empty warehouse into a machine that can build 3+ homes a day, 365 days out of the year.

Today marks the first step in that journey.

[YIMBYLAND @YIMBYLAND]

@Greygrouse all of it

[YIMBYLAND @YIMBYLAND]

- Jensen resigns
- 3rd Jensen mistress unveiled
- China training on Blackwell chips

Relax folks. Everything is going to be ok.

Now watch us accept delivery of our first press:

[The American Housing Corporation @americanhousing]

We're turning 225,000 sqft of empty warehouse into a machine that can build 3+ homes a day, 365 days out of the year.

Today marks the first step in that journey.

[YIMBYLAND @YIMBYLAND]

@ADoricko lmaooo

[YIMBYLAND @YIMBYLAND]

If this works, I want a line in the wikipedia article for starting this trend with the Trump Train High Speed Rail!

[Dr. Abdul El-Sayed @AbdulElSayed]

This is a message for @realDonaldTrump.

Donald, this is on the table. Guarantee healthcare for every American, from cradle to grave.

[Atoms Not Bits @AtomsNotBits]

3D printing enters commercial construction

Hard Tech Headlines:

- The Finnish Defence Forces have partnered with Varjo to test and develop mixed-reality C-UAS training for soldiers.

- UFORCE has secured a $700M Series A. The Ukrainian startup develops autonomous ground vehicles, ships, and drones, along with command-and-control software. The round was led by Valor Equity Partners and the UK National Wealth Fund.

- Maritime Tactical Systems (MARTAC) has signed an agreement with the Taiwan Coast Guard to deliver autonomous surface vessels, which can reach speeds exceeding 50 knots.

- @KairoxSystems has launched out of YC. The startup develops a fixed-wing drone that can fly through the air and dive underwater, using a pressure vessel that enables it to operate underwater.

- VulcanForms has secured $500M in financing led by J.P. Morgan and Apollo-managed funds to build three new U.S. factories. The startup specializes in large-scale industrial manufacturing using metal 3D printing, precision machining, and automation.

- Dunedain has secured a prototype OTA with the Office of the Assistant Secretary of War for Special Operations and Low-Intensity Conflict, focused on enabling Special Operations Forces to run agentic AI locally at the edge, even in communications-denied environments.

- The @americanhousing Corporation has begun construction on its second factory, a 225,000-square-foot facility designed to eventually produce four modular homes per day.

- Kraken Technology Group has signed an LOI with Umoe Mandal AS to explore the development of autonomous maritime systems. Umoe Mandal specializes in composite shipbuilding and operates a shipyard in southern Norway.

- @ICON3DTech, the 3D-printing construction startup, has begun construction on its largest commercial project to date, a 12,649-square-foot exhibition and training center in San Antonio, Texas, using its robotic 3D-printing technology.

- Ailit has launched out of YC. The startup is looking to convert diesel trains to battery power, enabling them to operate electrically.

- @mantlemachines has launched out of YC. The startup is developing an automated maintenance service for industrial machinery and robots, analyzing equipment data and sensor information to identify issues and deploying augmented technicians across the country to perform repairs.

- Serco has agreed to acquire QinetiQ's U.S. arm for $292M. The business specializes in autonomous maritime systems and military space systems supporting missile tracking and space domain awareness.

- @allencontrol has partnered with Attalon, a laser and optics manufacturer, to develop high-energy laser weapons for counter-drone applications.

[YIMBYLAND @YIMBYLAND]

This old Folgers coffee plant in Houston is apparently getting redeveloped into a Music and Arts public charter school.

One of the coolest redevelopment projects I've ever seen.

[John Arnold @johnarnold]

An underrated benefit of living in a growing city is that blighted houses and abandoned buildings get redeveloped rather than remaining eyesores.

[YIMBYLAND @YIMBYLAND]

@MFG_SMB and yet…

[YIMBYLAND @YIMBYLAND]

@emmasteuer1 East Texans can claim it, but no Texan in their right mind west of the piney woods would consider themselves southerners. They’re Texans.

[will depue @willdepue]

I have a major result to share.

I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.

This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.

Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.

I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.

I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.

It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.

These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.

I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at https://t.co/YOtzND0Jqk.

There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

[will depue @willdepue]

If you'd like to see more work like this, Unemployment Labs is raising $5M to fund our research for the next 6 months. As a subsidiary of https://t.co/wTO29zk3KK, we're a 501c3 pending non-profit. Please reach out to [email protected].

[YIMBYLAND @YIMBYLAND]

@dmitry_alexin Thanks!

[YIMBYLAND @YIMBYLAND]

@iononrecourse Loro is so good too

[YIMBYLAND @YIMBYLAND]

@CNN Based

[YIMBYLAND @YIMBYLAND]

@LombardTrucking @the_transit_guy Well absolutely need the truck. The large majority will be shipped by truck.

[YIMBYLAND @YIMBYLAND]

@floinkus 🔥🔥🔥

[YIMBYLAND @YIMBYLAND]

@NathanSRobinson whattttt, that's awesome!