Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

Wednesday, October 28, 2026

Back to today

From X

@convequity 2 clicks

Convequity on X

The usual explanation for CoreWeave’s credit pressure is that its loans outlast its customer contracts. The filings say otherwise. All of $CRWV’s debt is due before its current contracts end, though only just: $14.6B of notes falls due in 2031-32, as the last contract cash comes in. That distinction matters. For a GPU cloud, debt that outlives its contracts is the dangerous kind. When a contract ends the chips are old, and there is no contract left to borrow against. CoreWeave’s real problem is speed. Each new build needs fresh capital before the last one has paid for itself, and almost all of the debt sits with the parent company. $IREN has the opposite shape. Its Microsoft project loan is matched to the contract. The risk is $6.3B of convertible bonds due 2031-33, mostly after its contracted cash has ended. If IREN’s stock is high by then, bondholders will likely swap the bonds for shares. If it’s low, IREN has to repay them in cash. Which leads to an odd conclusion: slower AI demand growth could help both. Fewer new commitments would give finished projects time to earn back their cost.
@convequity 1 click

Convequity on X

$UMC's nearer-term driver is its 12nm process with $INTC, and our longer-term view is that logic folding could make chips on its existing processes faster and more energy-efficient. UMC reports Q3 earnings tomorrow, while production is targeted for 2027 at Intel's Arizona factories. The partnership combines UMC's foundry experience with Intel's manufacturing capacity, giving UMC access to U.S. production without building an American factory. The immediate attraction is winning customers and expanding its business. Folding replaces long horizontal connections with shorter vertical ones, so signals need less time and energy to move. That could make chips built on established processes more capable without matching the newest transistor nodes. Older processes are, in our view, a more manageable starting point for this prospective thesis: UMC would still need to solve bonding, heat removal and reliable production, but it would not also be introducing a new transistor process.

Investor Letters

All letters

My Last GFL Update - Hopefully

GFL’s take-private structure hinges on BC Partners rolling 60% of its stake to preserve debt portability under the 40% permitted-holder…

GFL $16B
310 Value’s Newsletter 5 clicks

Are Neoclouds Dead?

Neoclouds face tighter debt financing as AI infrastructure capex outruns cash flow, leaving cash-burning GPU providers more exposed than…

NBIS $46B
Daniel Romero 9 clicks

Reddit: Deep Dive Analysis

Reddit generated $805 million in Q2 2026 revenue, 95% from ads, but slowing Google search referrals expose AI-driven traffic and…

RDDT $26B
Hated Moats 6 clicks
WULF · long

Terawulf INC

TeraWulf is shifting from bitcoin mining toward data centers, seeking a higher-quality and more defensive business model.

WULF $6.3B
Value Investors Club

Publications we follow

Every publication on this site, hand-picked.

129 publications

Newsletters 109

YouTube 4

Blogs 15

Seeking Alpha 67 of 268

Nominate a publication

X accounts we follow

Every account on this site, hand-picked.

91 accounts
Nominate an account

Investor letters we follow

190 firms

Sectors

Ordered by how often our editor reaches for them.