Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

Thursday, October 1, 2026

Back to today

From X

@convequity 1 click

Convequity on X

Inference repeatedly reads the same model weights, with far fewer writes than training. 1. In training, weights are constantly updated, so the memory sees heavy read and write traffic. 2. In inference, the dominant pattern is repeated reading of the same weights, and the key-value cache alongside them, with far fewer writes. That is the case for putting AI model weights on flash. Flash has traditionally been weaker than DRAM on write endurance and write speed. A read-heavy, write-light workload plays to its strengths and gets around exactly those weaknesses. Today AI chips sit beside HBM built from stacked DRAM, and it is the expensive part of the memory bill. High-Bandwidth Flash (HBF) is the product built on the read-heavy observation. It aims to deliver HBM-like read bandwidth at far higher capacity and lower cost by using NAND flash instead of DRAM. Two caveats. These are aims, not measured specifications. As of our early-August note, the first prototype from Sandisk ($SNDK), developed with SK Hynix, was expected later in the year. HBF is also more specialised than commodity NAND, where Chinese producers are expanding capacity aggressively.
@david_katunaric 1 click

Mikro Kap David on X

Q3 absolutely sucked, and my performance currently stands at 37.8% YTD versus 65.8% at the end of Q2, marking the worst drawdown from a high since I started sharing my buy/sell decisions on Substack. Apparently, the Market Gods decided this is what you get for calling your last portfolio update “Half-time Lead.” …I’m just glad I didn’t use a Buffett quote. That might’ve finished me off. While I barely felt the pre-2024 election volatility, the spring 2025 tariff scare, or the start of the Iran war in 2026, Q3 broke that pattern. $EVC, my largest position, fell more than 40% from its highs and made sure I felt this one. The portfolio’s relative resilience, something I’d come to appreciate, disappeared with it. Of course, I have plenty to say about Entravision, but I’ll save the fundamental developments for later in the article. Q3 was a season of portfolio pruning (wow, I never thought I’d use that phrase). I exited two winners, realized one loss, added to a large position, saw a rare billion-dollar-plus company find its way into my portfolio, and ended the quarter with 6% in cash versus 0% at the end of Q2. I ended Q3 with 11 holdings, with the top three accounting for 53% of capital and the top five for 70%. Portfolio beta also came down from 0.95 at the end of Q2 to 0.80 at quarter end. That is still elevated relative to the portfolio’s historical level, largely because my biggest position carries a beta of 1.68, while every other holding sits below 1. I’ll save the longer performance review for year end, when we can hopefully put what I still consider a satisfactory 2026 into proper perspective. For now, I think it’s better to focus on what actually matters for a concentrated stock picker like me: what has changed fundamentally, how those developments affect the risk/reward, and where my conviction stands today. As usual, at the link below I went through the portfolio from the smallest position to the largest.

Investor Letters

All letters

Not So Fast

Entravision, the portfolio’s largest position, fell more than 40% from its highs as Q3 returns dropped to 37.8% YTD.

EVC $682M
The Mikro Kap 2 clicks

The Next Consumer is a Computer

Meta’s Muse illustrates how persistent AI agents shift compute from per-query to per-user, driving heavier GPU, CPU and memory demand.

META $1.8TNVDA $5.5T
SixSigmaCapital: Navigating the Financial Markets 6 clicks

Publications we follow

Every publication on this site, hand-picked.

110 publications

Newsletters 93

YouTube 4

Blogs 12

Seeking Alpha 41 of 198

Nominate a publication

X accounts we follow

Every account on this site, hand-picked.

72 accounts
Nominate an account

Investor letters we follow

189 firms

Sectors

Ordered by how often our editor reaches for them.