Possible / Civic idea / Research explorer 1.0 · October 23, 2026
The Productive City.
Computing, homes and heat
Seattle has a surplus of office space and a shortage of homes. What else could our downtown buildings do?
More than a third of downtown office space is vacant. Cushman & Wakefield measured 35.8% vacancy in Q3 2026, up from 35.1% a year earlier. Its surveyed downtown market contained about 18.3 million square feet of vacant space, including direct and sublease vacancies. Read the Q3 report ↗
That is a substantial stock of existing space in a city that needs more homes. Seattle’s office-to-residential program already addresses those two problems together. But converting an office is expensive, and a cheaper building does not automatically make the work affordable.
For buildings that no longer work well as offices, waiting for the old demand to return is only one option. The Productive City asks whether homes, a computing tenant and useful recovered heat could together support another future. The test is whether the housing and public benefits justify the space, electricity and construction they require.
Start with a hypothetical 1 MW computing installation. Choose a real downtown building, specify a heat-demand scenario, and see what happens as more installations are added. Try smaller capacities, different numbers of homes or a sheltered outdoor seating area.
02 / The city is more than a sum
Useful heat. Real electrical demand.
At our default assumptions, 1 MW of installed computing could supply about 0.91 MW of heat after recovery and a heat pump. That comes close to the modeled winter peak for roughly 360 homes. Hot water is part of the annual demand; its simultaneous peak and temperature requirements still need design.
But a winter-sized heat source is too large for most summer days. In our illustrative residential profile, much of its annual heat still needs to be rejected. A nearby year-round customer could help, provided the connection, delivery temperatures and contracts work. A thermal network and a shared computing operator are separate choices.
Ten 1 MW installations imply about 13 MW of compute-facility demand at full load; fifty imply 65 MW. Heat pumps and other building loads add to that. Distributing computing does not make its cumulative demand disappear. City Light’s 10 MVA service policy is separate from the over-20-MVA land-use moratorium definition; neither provides a promise of available service.
A warmer place to sit.
Recovered heat might also serve sheltered outdoor seats or nearby radiant surfaces. Begin with wind and rain protection, then target occupied places. The explorer’s 20 kW seating zone is a modest experiment, not a reason to enlarge computing simply to use its heat. It receives only what remains after residential demand, and creates no assumed heat-sales income.
03 / First building investigation
Warm Tower, Fourth & Blanchard.
This prototype begins with a building whose future is already in question. Colliers’ public sale listing reports Fourth & Blanchard as 54% leased (checked October 23, 2026; the figure is undated). Its seven-building portfolio went through foreclosure on August 8, 2025. That combination of underused space and an ownership reset is one reason we chose it to test another future. Existing tenants and lease terms remain part of that test.
The original three-floor case mostly fails the thermal-synergy test: its apartments could absorb only about 3% of estimated annual recoverable heat, before timing constraints. Most of its modeled financial improvement comes from compute tenancy and changes in space and cost. A large financing gap remains.
The next version asks a sharper question: what changes with a smaller installation, how much housing space does it consume, and does heat recovery itself earn its additional cost? Compare housing only, computing without recovery, and the same building with recovery.