Great Things with Great Tech Podcast

The episode contains a moderate amount of useful technical information - GPU fractionalization for dev workloads, policy-based workload routing, cloud performance parity inflection around 2019-2020 - but is heavily padded with explanatory throat-clearing and the host narrating back what the guest just said. Genuine insight per minute is low.

“cloud was kind of really becoming competitive performance wise...Azure had Infiniband on the floor around that time. They just started rolling that out. You could actually get those things to perform in a way that was similar performance to these on PREM supercomputing systems”

“We're doing MIG fractionalization and we're partnering with a company called Juice Labs for non MIG supported uh, pooling of GPUs. So fractionalizing GPUs out.”

The episode recycles broadly familiar narratives - democratizing HPC, hybrid cloud complexity, cloud repatriation - without any contrarian or first-principles arguments. The quantum mention at the end is generic hype with no substantive framing.

“democratize, you know, large computing environments was our slogan. Wake up 10 years ago.”

“it's like a cyclical world, right?”

Matthew Saxted is a genuine practitioner - an engineer who built HPC clusters from salvaged render farms, spun up 20,000-core jobs at Argonne, and bootstrapped a company on customer revenue and grants. He has real operational depth, though the company is small and niche enough that the perspective stays narrow.

“I remember very clearly, you know, first time running a large HPC job, I was in my office late at night, clicked a button, spun up 20,000 cores on one of a crane machine.”

“we were a fairly small team that was delivering, you know, building this product out...five, six people maybe, uh, for you know, years”

There are some concrete specifics - 400 FedRAMP controls, a 2,000-page documentation package, a 2.5-year authorization timeline, named infrastructure vendors like Juice Labs and CoreWeave - but no customer counts, revenue figures, performance benchmarks, or deal sizes to ground the broader claims.

“a FedRAMP High Impact Level 5 authorization, which is what we got about not even a month ago, took us about two and a half years...it's really a checklist of about 400 items”

“I think our documentation package is like 2,000 pages.”

The host is affable and occasionally surfaces useful topics like cloud repatriation and scheduler abstraction, but mostly narrates back what the guest said and asks leading questions the guest simply confirms. There is no pushback, no probing of unverified claims, and no productive disagreement throughout.

“So what you've done is come in and create that overlay and that UI and that user experience to democratize that effectively.”

“Yeah, that's interesting because obviously it creates headaches on both sides, right?”