AI infra capital partner. We originate, underwrite & manage financing for AI compute economy. Backed by @hack_vc, @FactionVC, @ambergroup_io, @AntalphaGlobal.
gaib.ai
Joined June 2024
- Tweets2.6K
- Following286
- Followers143K
- Likes2.7K
Next: a $140M pipeline of qualified GPU deals ready to deploy.
Full Q3 breakdown on the blog 👇
blog.gaib.ai/said-q3-2026-10…
As agents begin procuring compute directly, providers are losing the ability to identify who is on the other side of a transaction. Travis Cannell @traviscannell, COO of @vast_ai, on what the usage data reveals.
How do you tell whether a customer is a person or an agent?
Travis Cannell @traviscannell, COO of @vast_ai, says the clearest signal they have is API usage patterns over time, and those are shifting. Beyond the data, his point is simpler: the way people approach problems has changed, and the systems on the other end are getting longer memories and running for longer stretches.
On building the procurement model for it: “we built this system for what was called the singularity, and now we’re calling it this agent thing.”
Full episode available now.
Where compute demand actually comes from?
“Humans love intelligence, and we can manufacture it now.”
Travis Cannell @traviscannell, COO of @vast_ai, on what’s actually driving compute demand: not hype, but how much intelligence people want to produce. As the big labs absorb more capacity, the mid-tier buyers scramble for what’s left, a waterfall effect running down the market.
He thinks it may be a little frothy, with cycles that let some error out, but sees it as underpinned by real breakthroughs rather than a bubble.
Full episode available now.
GAIB retweeted
How a compute business describes itself shapes what buyers expect from it. @traviscannell explains why Vast doesn't fit the usual definition.
Travis Cannell @traviscannell, COO of @vast_ai, on why “marketplace” is a confusing word for what Vast does. Some GPU marketplaces really do sell physical hardware.
His preferred description is an infrastructure layer, because Vast does more than match buyers and sellers, starting with virtualization.
Full episode available now.
Renting capacity and selling hardware are different businesses, even when both get called a marketplace. Travis Cannell @traviscannell, COO of @vast_ai, on where Vast sits.
Travis Cannell @traviscannell, COO of @vast_ai, on why “marketplace” is a confusing word for what Vast does. Some GPU marketplaces really do sell physical hardware.
His preferred description is an infrastructure layer, because Vast does more than match buyers and sellers, starting with virtualization.
Full episode available now.
Gonka is the first platform GAIC serves, not the last.
We're looking for the next AI projects that need GPU capacity sourced and operated through GAIB's global GPU supply.
More on what's next in the article: blog.gaib.ai/gaics-gonka-pha…
Vast doesn't own compute. It organizes it.
Travis Cannell @traviscannell, COO of @vast_ai, on what that means.
“Google is for information. Vast is for computation.”
Travis Cannell @traviscannell, COO of @vast_ai, on how he describes the company: Google doesn’t make websites, it organizes information. Vast doesn’t own compute, it organizes, tests and standardizes it.
His five-to-ten-year view: training may start to look more like inference, which means fewer long contracts and more on-demand compute from different places and GPU types.
Full episode available now.
Same servers, same data center, but 5% variation in GPU performance?
Even identical servers, in the same data center, can vary by about 5% in GPU performance.
Travis Cannell @traviscannell, COO of @vast_ai, citing a study, on why: the silicon lottery, heat dissipation, and where a GPU physically sits relative to the others. If compute varies even when you control for everything, pricing it all the same means you either underprice or overprice it.
Full episode available now.
New episode of Where Capital Meets Compute: Travis Cannell @traviscannell, COO of Vast AI @vast_ai, on why no two GPUs should be priced the same and why he doesn’t think today’s AI buildout is a bubble.
New episode of Where Capital Meets Compute: Travis Cannell @traviscannell, COO of Vast AI @vast_ai, on why he thinks of every GPU as a “snowflake” rather than a commodity.
His example: a Nebraska farmer who went from cattle to Bitcoin to AI compute, filling his barn with GPUs, and undercutting CoreWeave on price by skipping the backup generator.
GAIB retweeted
What's a GPU worth in year three? The collateral behind every GPU loan depends on that number, and our guests disagree on where to find it.
@carmenli (Silicon Data) argues residual value should be highly correlated to the rental rate, and that lenders can use the forward curve to see how their collateral is trending.
@berniebiased (American Compute) ran a correlation study comparing resale prices to rental prices over the same period. They moved in different directions, so his underwriting relies on real resale transactions.
@Ishanee (Architect) supplies a likely reason for the gap: roughly 85% of the non-hyperscaler market is locked into reserved contracts, while rental indexes like Carmen's track on-demand rates.
GAIB retweeted
"The AI labs themselves are just infinitely hungry for compute." Ishanee Nagpurkar @Ishanee, Head of Compute Markets at Architect @Architect_Fi, on why the build-outs still aren't keeping up.
Her point: compute demand looks like one persona from the outside, but break it down and it's enterprises, SMEs, AI labs, net-new startups, even exchanges and lenders, all consuming compute in some form.
Full episode available now.