You are a hardware company that needs power, height, a loading dock and machines you cannot justify buying alone. The proposition here is that a group of startups and entrepreneurs like you buy one building together instead of signing separate leases: you hold equity rather than a lease, you outfit the bay you need most, and you rent time on everyone else's. Below is every building for sale across San Francisco, San Mateo and Santa Clara counties that could hold it. Ninety-nine scored, twenty-nine that clear the bar.
Capability first, in the pattern of the facility playbook's Model B, recalibrated for a fabrication shop rather than an office relocation. Electrical service replaces parking ratio as the dominant factor, and whole-building deliverability is weighted far higher, because a group buying together needs the whole box rather than a suite.
Founder reach is measured to the nearer of two centres of gravity for Bay Area hardware teams, Pier 70 in San Francisco and Palo Alto. It deliberately does not penalise the South Bay against San Francisco. It penalises only genuine outer-county distance, which is what removed Gilroy and Morgan Hill. If your team is somewhere else entirely, this is the one weight you should argue with.
Match bands, not raw scores. Strong is 72 and above, Good is 62 to 71.9. Possible and Weak sit in Archive, never deleted. Every grade on a card carries the evidence behind it under Full details, so you can check our reading against the source rather than taking the band on trust.
Clear height is in the model at 10 points. It was absent from the spreadsheet exports on all 643 records. CoStar property summaries pulled on 4 September closed that gap for the shortlist, and it reordered the ranking materially. Buildings without a summary are scored at a neutral 4 of 10 rather than penalised, exactly as undisclosed electrical service is. The animation above is a concept study, not a survey of any real building.
This is a concept under active development, not a closed round. The five points below are the shape of the deal as we understand it today. The open questions underneath are real and unresolved, and your answers to them are part of why we are showing you this now rather than after the terms are set.
Participants take ownership in the entity that buys the building. Your monthly cost is your share of debt service and operating expense rather than a market rent carrying a landlord's margin. If the building appreciates, that accrues to the companies working inside it.
No early company can justify a five-axis mill, a fume hood, a reflow oven and a netted test volume at once. Each participant outfits the zone it needs most. The rest of the floor is there the day you need it, without a second capital raise.
Idle capacity is the point, not a side effect. Time on equipment you funded is bookable by other members at an internal rate that pays you back. A machine running at 30% for one company can run at 80% for the floor.
Six zones, shared docks, one break room. Your machinist meets someone else's controls engineer, your supply chain lead finds out which contract manufacturer burned the team two bays over. That adjacency is the part a lease cannot buy.
A large, credit-worthy anchor committing to a block of the floor changes the financing terms for everyone. Y Combinator is the natural candidate given the density of hardware companies in its cohorts. To be explicit: YC is not a party to this work and has not been approached.
Stated plainly, because you would find them anyway and they change the arithmetic.
All twenty-nine Strong and Good candidates, plotted from the export's own coordinates. Circles sized by building area, coloured by match band. Twenty-five are in Santa Clara County and four are in San Francisco. No San Mateo County building cleared the bar, which is a finding rather than an omission.
If you are considering putting capital into one of these buildings, you need the limits of the evidence as much as the ranking itself. Two source layers sit behind these cards: three CoStar spreadsheet exports pulled 3 September, and 29 CoStar property summaries pulled 4 September that confirmed the physical specs. Every card says which layer it rests on, and the summary for each property opens from the card.