Steven Woods
AI Briefing
Slowly, then fast. Every morning.
Steven Woods · public projects ↗
Share✉ Subscribe

Deep dive today: How a Convertible Note Creates a $72B Paper Loss

AI Briefing · Sunday, October 25, 2026

The day in brief. First, the cost of capital and compute. The Information frames CoreWeave as a test of whether the AI build-out can be financed. Replit reportedly moved $100M+ a year of spend to OpenAI for a 30%+ discount after Anthropic ended its own. Both rest on headlines we could not read in full. Second, robotics capital: Eka's $150M Series A prices sim-trained picking before outside proof. A question worth asking in diligence this week: how much gross margin depends on a model-provider discount, and what happens if it ends?

Steven's takeLooks like head to head discounts are getting easier - perhaps model quality is converging across a wider array of useful dimensions... it is increasingly obvious that for a lot and increasing amount of highly valueable use cases that free models are going to become the de facto standard in general while true frontier models are going to do a lot of planning, exploring and more complex tasks - at least for now.

Top of mind

  • Robotics Series A hits $150M. Eka Robotics raised from a16z and F-Prime. Its claim of 1,500 picks/hour on unseen objects is the company's own, without outside confirmation. 1
  • Model discounts are a margin line. Replit reportedly shifted most of a $100M+/year AI budget to OpenAI for a 30%+ discount after Anthropic ended its own; the full report has not been read. 2
  • CoreWeave as financing bellwether. The Information frames CoreWeave as the test of AI build-out financing; Bloomberg traces chip-backed debt to its 2023 $2.3B Blackstone-led loan. 4 6
  • Altman on slowing AI. Altman told OpenAI staff the industry needs ways to slow dangerous AI, listing regulation, lab mergers or one firm dominating compute; endorsement of any is not stated. 3
Steven's takeAltman is hardly leading the way on care, slowing or regulation - afaik.

Must read

1. Eka Robotics secures $150M Series A for VFA robotics model

The Robot Report · report · source · 2026-10-24

  • What happened: Eka Robotics says it raised a $150M Series A led by a16z and F-Prime. It claims robots using its simulation-trained VFA model pick objects outside the training distribution at roughly 1,500 operations per hour.
  • Agenda: Sponsored post by Eka Robotics with investor quotes
  • Evidence check: Mixed. The $150M raise is reported; claims of 1,500 picks/hour on unseen objects and sim-to-real rest on Eka and its backers.
  • Our read: A $150M Series A resets the price of robotics foundation models and may pull capital toward simulation-data suppliers. The 1,500 picks/hour on unseen objects is self-reported; customer-site throughput and error rates would change our view.
  • The market view: a16z and F-Prime, which led the round, are pricing sim-trained models as the route to general picking. Capital at this scale can buy the deployments that prove it.
Steven's takeThe devil is in the details - looking forward to seeing more magical results come out of these models - and practical repeatable ones especially.

Worth knowing

  • 2. Replit shifts AI spend from Anthropic to OpenAI for 30% discount. Replit moved most AI spending from Anthropic to OpenAI, securing a 30%+ discount on a deal worth $100M+/year after Anthropic discontinued its discount. (theinformation.com via Techmeme)
  • 3. Sam Altman proposes slowing AI development via regulation or consolidation. Altman told OpenAI employees the industry needs better ways to slow potentially dangerous AI. He described three paths the industry has discussed: international regulation, merging leading labs, or one firm dominating compute. The excerpt does not say he endorses any. (thein.fo via The Information)
  • 4. CoreWeave's financing fate signals AI build-out viability. CoreWeave is becoming a bellwether for whether the entire AI build-out can be financed, with investors betting it cannot. Developing story · day 4 · first in the briefing Oct 15. (thein.fo via The Information)
  • 5. Apollo's Torsten Slok argues AI shows no financial payoff outside tech. Apollo chief economist Torsten Slok claims AI shows no financial payoff outside tech (Techmeme), citing flat S&P 493 revenue per worker and margins versus rising Mag 7 metrics. Evidence check: overstated. (apollo.com via Techmeme, Gary Marcus)
  • 6. Blackstone-led $2.3B loan to CoreWeave sets chip-backed debt blueprint. $2.3B Blackstone-led private credit loan to CoreWeave in 2023 established a template for chip-backed debt, projected to reach $1T market size by Bloomberg. (bloomberg.com via Techmeme)

Recommended reading

Eka Robotics secures $150M Series A for VFA robotics model · The Robot Report · report

Why it matters: It is today's only fully readable report. It gives Eka's own account of VFA, its simulation training and the 1,500 picks/hour claim. That account is the baseline for checking any customer deployment data that follows.

In brief

Open questions

  • Did Replit actually move $100M+/year to OpenAI at a 30%+ discount, and is Anthropic ending discounts for other large customers? Model spend is the largest cost line for coding-agent companies. Settled by: The full report from The Information, or other large customers describing Anthropic pricing changes. 2
  • Can CoreWeave raise or refinance debt on terms that do not signal stress for chip-backed lending? Its 2023 loan is the template for a debt market Bloomberg projects at $1T. Settled by: Pricing, spreads and collateral terms on CoreWeave's next debt raise. 4 6
  • Will Eka publish customer-site throughput and error rates for picking unseen objects? Sim-to-real generalisation is the whole basis of the $150M valuation. Settled by: A named customer deployment with independently reported picks/hour and failure rates. 1

31 items analyzed from 15 sources · 44 filtered as off-topic · 16 more stories in the dashboard. Analysis: claude_code:default, local:qwen3:30b. Our read, the other views and the day in brief are machine-written. Sources are weighed by how they know: a primary source, a report or a hands-on test establishes a fact; analysis, opinion, summaries and posts are commentary, however many there are. Generated 2026-10-25 06:09 EDT.

Rate this briefing

★★★★★1 poor · 5 excellent

Get it by email

One email each morning when the briefing is published. Confirm once; unsubscribe with one click.