Rosa Del Mar

Daily Brief

Issue 7 2026-01-07

Evaluation-And-Measurement-As-Primary-Lever

  • Foody says he is most interested in identifying which rubrics and data types most improve model performance and how to apply human labor more efficiently to frontier AI training problems.
  • Foody argues employers should evaluate what candidates can accomplish while using powerful AI tools (e.g., via screen-recorded timed build tasks) rather than trying to prevent AI assistance in assessments.
  • Mercor pays high expert wages (e.g., $150/hour for poets) because a one-time expert teaching effort can be amortized across very large numbers of model users.

Patient-As-Operator Precision Oncology (Diagnostics, Combinations, Radioligands)

  • As of the interview, Sid Sijbrandij’s cancer was not detectable, he had a planned escalation path with roughly 30 therapeutic options, he was continuing development of about 10 drugs/diagnostics, and he had started six companies to scale treatments for others.
  • The conversation cites willingness to pay high monthly amounts for AI tooling, including a future scenario of $10,000 per month with human-plus-agents and an example of someone paying $1,400 per month for Cursor.
  • Sid Sijbrandij says Kilo’s paid features target managers/executives, including organization-wide AI adoption analytics priced at $15 per user per month, plus an enterprise tier with centralized BYOK and Bedrock-in-VPC support.

Compute Economics: Shortages, Buildout, And Commoditization Expectations

  • Supply shortages in GPUs and data center capacity are likely to trigger large-scale replication and buildout that reduces per-unit costs over time.
  • State-level AI regulation is proliferating, with roughly 1,200 bills being tracked across all 50 states, raising fragmentation risk.
  • Enterprise AI monetization is trending toward usage-based pricing where customers buy tokens as metered units of intelligence.

Tokenized Collateral As The Operational Bottleneck And Primary Institutional Use Case

  • Collateral rating and liquidity assessment tooling is still needed to support safe adoption of tokenized collateral.
  • The CFTC is exploring whether MiCA-regulated spot exchanges could be recognized using the Foreign Board of Trade framework.
  • Roughly 10% of the U.S. repo market is approaching being put on-chain, and a partner cited by the host does about $400 billion of repo daily.

Debt Fragility, Fiscal Dominance Pathways, And Policy Constraints

  • The trigger for a systemic tipping point is likely psychological, occurring when markets collectively recognize unsustainable debt dynamics that had been ignored in plain sight.
  • AI capex concentration risks a boom-bust with rapid depreciation and malinvestment if expected profits do not materialize.
  • The economy is a complex adaptive system and is not simple, predictable, or controllable in the way linear deterministic models assume.

Ai And Investment Promises Create Funding And Issuance Pressure

  • Global and U.S. yield curves are already reflecting issuance and growth pressure, and keeping rates low would likely require meaningful disinflation while an inflation-neutral backdrop implies somewhat higher rates.
  • QE-like reserve provision can fail to increase lending when banks are able but not willing, while broad credit expansion requires borrower demand at attractive prices for banks.
  • Macro outcomes are driven by money creation, credit creation, and the risk/term structure of liabilities used to fund activity.