Locator: 51894COMPUTE.
Compute: basically relates to whom can afford to buy the most of the highest capacity chips available.
Previously posted (Locator: 51549CRAMER):
New asset class? Some analysts still support. A financial asset class is being built by Jensen Huang. Folks may want to discuss with Gemini the concept of a financial asset class.
- the asset: specifically AI chips (GPUs) and AI compute capacity, not tokens;
- mechanism: Wall Street is creating structured financial instruments, lease backstops, and asset-backed debt tied directly to physical chips and data center compute power;
- the analogy: prominent investors like Mark Cuban and Jeff Gundlac have noted that Wall Street is trying to treat silicon chips like a tradable commodity (akin to oil or gold), prompting major exchanges like the CME group to plan future contracts tracking GPU rental indices
- asset-backed debt: financial insitutions are issuing long-dated loans and bonds using physical AI chips as the underlying collateral
- One may want to use this prompt: one may recall that Larry Ellison bought a huge amount of compute (chips) for which he was criticized but now it turns out that he can indeed rent them, even if they are seen by some as "old."
- in hindsight, one might argue that Larry Ellison was the first big investor to note this "new" financial asset class.
- Larry Ellison was absolutely the first major tech titan to aggressively treat AI compute as an investable infrastructure asset class, essentially proving the concept years before Wall Street formalized it.
- Again, not chips, but compute: capacity / chip x chip = compute. But that's a theoretical number, similar to "nameplate capacity" used by utilities. Compute is also affected by optimum wiring, optimum cooling, optimum power availability, etc. So, analysts must consider all these factors to determine "compute."
- but there's more -- see the BOX below:
Tag: Ellison
