{ "blyg": "0.3", "id": "1mn57sr8gcqmmw64hvp7bt80de", "kind": "fragment", "origin": "https://www.msweet.net/notes/", "page": "925-sage-wrens", "author": { "name": "Matthew McDowell-Sweet", "url": "https://www.msweet.net/" }, "created": "2026-10-25T16:05:47Z", "updated": "2026-10-25T16:05:47Z", "version": 1, "content_md": "Hmmmm.\n\n> This is the problem that SF Compute solves, by building a physically settled forwards market for compute. We make it possible for lenders to use the GPU hour as collateral, on a liquid book, rather than the GPU cluster, which is non-fungible and requires a servicer that might blow up right at the moment they need it the most. If a lender gets a GPU hour as collateral, the credit risk of the customer matters less; if the two person AI lab in a San Francisco Victorian fails to make their payments, the lender has a real path to selling the collateral. That makes startups possible at all, a path that seems increasingly difficult in AI.\n\n[The end of startups | Evan Conrad ↗](https://evanjconrad.com/posts/end-of-startups)", "content_html": "

Hmmmm.

\n

This is the problem that SF Compute solves, by building a physically settled forwards market for compute. We make it possible for lenders to use the GPU hour as collateral, on a liquid book, rather than the GPU cluster, which is non-fungible and requires a servicer that might blow up right at the moment they need it the most. If a lender gets a GPU hour as collateral, the credit risk of the customer matters less; if the two person AI lab in a San Francisco Victorian fails to make their payments, the lender has a real path to selling the collateral. That makes startups possible at all, a path that seems increasingly difficult in AI.

\n

The end of startups | Evan Conrad ↗

", "content_hash": "sha256:4b3f55f15917e39103f525215eb7153cafc2892224b7c489d3704c2ec18a25e1", "media": [], "changelog": [ { "version": 1, "at": "2026-10-25T16:05:47Z", "note": null } ] }