B3

Introducing B3IQ.

A new model to own, run and monetize compute.

Already being used by faculty, researchers, and student teams at Stanford, NYU, Penn, UChicago, Dartmouth, Waterloo, and the University of Hawaiʻi.

Financing a GPU cluster today can cost you 7-15% a year, and the rate depends heavily on who’s renting your GPUs.

On deals under $20M:
→ 7–10% with an investment-grade offtaker
→ 10–12% with a strong corporate offtaker
→ 12–15% with a weaker customer
→ 15–18% with a less established customer or operator

That’s because lenders mostly price the contract, not the machine itself.

But serving inference varies a lot from machine to machine, and lenders now want to track it live, down to tokens/sec.

@darylX24 explains how putting that data onchain could scale GPU credit at @PanteraCapital's Blockchain Summit.

Spare GPU hours don't have to sit idle.

See what owning a node can earn you each month on B3IQ.

Choose your machine and compare earnings at different occupancy levels. ↓

AI is outgrowing the supercomputers built for research.

Per Times Higher Education, Australia’s top supercomputers are ~3x oversubscribed, with researchers requesting 2.2B+ compute hours this year.

And it’s not just one country. Research labs doing critical work now compete with everyone else for the same GPUs.

Commercial compute could help, but access and cost are still barriers. Many researchers say they’d use it if those hurdles came down.

With B3IQ, we’re opening that access up in two ways.
→ Own a dedicated machine. We build, host, and run it.
→ Or rent whole machines by the day.

So teams can spend their time on research, not waiting on compute.

Older GPUs are outlasting their depreciation schedules, per @a16z.

On a 6-year schedule, H200s should be down ~30% but are instead renting at 143% above the line.

A100s from 2020 are still under contract through 2029.

The lifecycle of a GPU is set by demand now.