Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

Saturday, October 3, 2026

Back to today

From X

@convequity 3 clicks

Convequity on X

Huawei just updated the Ascend 960 & 970 roadmap. $NVDA's competition pressure is increasing from both $AVGO and Huawei continuously as we've anticipated. → 960DT rollout pulled forward from 4Q27 to 1Q27. (Seems like they are finally getting the HBC-like, tailored-memory approach closer to high-volume production.) → 960PR by 3Q27 with 8 PFLOPS FP4—double the 960DT’s FP4 compute. (Closing in on NVDA’s Blackwell Ultra, which tops out at 15 PFLOPS dense FP4. Rubin moves that target to 35 PFLOPS dense FP4.) → 970 by 2028 with 14 PFLOPS FP4, up from the previous 8 PFLOPS target. Overall, this looks consistent with our Tau Scaling thesis: a higher compute ceiling, more tailored chip designs, and a faster path to production. Not every specification has doubled, but a near-doubling of the 970’s planned FP4 compute alongside an accelerated 960 rollout is a meaningful change. The bigger story, in our view, is Huawei’s CIDM model. Optimizing the entire vertical—not just the logic die—creates more room to change the compute design and shorten the development cycle. Huawei’s disclosed co-optimization already spans devices, circuits, chips and systems. This is the kind of vertical integration Elon Musk is targeting with Tesla and SpaceX’s Terafab. But Huawei is already putting its ecosystem version to work. (Would be quite funny if the Chinese ecosystem gets accelerator-based EUV into production lithography first, too.) Against NVDA, our thesis is that the single-chip gap narrows from roughly three years toward two—and potentially less as the roadmap accelerates. From a PPAC perspective, Tau Scaling could bring the logic die within a year—or less—while memory remains further behind. Bandwidth, not simply capacity, looks like the harder bottleneck. But Huawei may be able to lean into memory capacity and co-design the LLM and serving strategy around that constraint, rather than simply copy something optimized for NVDA’s hardware balance. System-level performance still needs to be proven at the cluster sizes Huawei is targeting. But we would not be surprised if it works better than expected. Even before the full UnifiedBus vision is realized, we think Huawei could gain system-level performance leadership on some large-model workloads sooner than many expect.
@BrokenMoats

Broken Moats on X

SMR reactors last year or quantum...appears the current moments basket of trash stocks with a narrative that doesn't fit the companies is 3rd derivative ai drug discovery plays (genome, simulation software, etc). $TWST $TEM $SDGR $TXG and theres more My favorite is $WGS - opko rehash with key players from there somehow sitting on a multi billion dollar valuation currently. There were caught with billing schemes multiple times at Opko as well as other serious problems (see Philip frost and opko). A member of the board actually still sits on Frosts charity and the CEO was the CEO of Cathie Wood favorite Invitae, that went bankrupt a few years ago. The short story writes itself, seems like a CVNA almost where the setup is pretty opposite, and yet it probably continues to work against you
@KEDM_COM 2 clicks

KEDM.com on X

Alpine Capital @jdsdebruyn reduced semiconductor exposure and added hyperscalers after the June pullback, citing combined hyperscaler capex of approximately $416bn in 2025 projected to exceed $750bn in 2026. It initiated Oracle $ORCL, and concedes the entry has not worked yet: “Oracle’s price action since initiation has been disappointing, but we believe the valuation is attractive and the current view on Larry Ellison (co-founder) is probabilistically asymmetric given the entry point. Oracle can slow spending if AI reverses (which we do not think will happen).”
@BlueDuckCap

BDC on X

Just surfacing this one to demonstrate we are not and never were $snap absolutists. We will go long and short and when facts and circumstances change our opinion changes too. And NO - Specs will not be a mass market hit this holiday season. Too expensive and awk form. But here's the bull case from here: 1. FCF/Share is the metric mgmt now focused on. By definition that requires fresh cost and SBC discipline. 16% hc reduction and flat share count already in place. $500mn in opex savings about to flow through. 2. Ads biz stable enough. 9% growth in Q2. Guide implies so decel. But ad checks stable vs Q2. Decent enough. Some progress on ad tech. Only needs 5% growth next year to hit our numbs. 3. Sub biz $1.2B rr, growing 85%. Needs 50% growth next year to hit numbs. 25mn paid subs, still well below penetration threshold of 7-12%. 4. Specs tech is real. 7k+ patents. Real use cases. What's that worth? $META or $MSFT or $AMZN could buy Specs for the patents alone and pay $2B and it would be nothing to them but all the call optionality which could incubate until form factor is mass market. 5. Lets just say Specs is worth $2B to a strategic buyer or partner. That implies $snap 7.5B or $4.5/ share. In this scenario then FCF/share at SNAP goes to over $1/share, easily. So 4.5x FCF on a business still growing revenue mid teens. @evanspiegel Specs are cool. But can't keep spending $500mn a year on them. Bring in a minority investor please and lets make SNAP stock great again. @IrenicCap

Investor Letters

All letters

Nike, Part Two: The Deep Dive

Nike’s second Latticework program examines the company and the global athletic footwear and apparel industry for professional investors.

NKE $51B
Latticework by MOI Global 1 click

Nike, Part One: The Story

Nike grew from a $1,000 shoe-distribution venture in 1964 into the largest sports brand, using athlete endorsements, product innovation and…

NKE $51B
Latticework by MOI Global 2 clicks

Publications we follow

Every publication on this site, hand-picked.

109 publications

Newsletters 92

YouTube 4

Blogs 12

Seeking Alpha 46 of 222

Nominate a publication

X accounts we follow

Every account on this site, hand-picked.

74 accounts
Nominate an account

Investor letters we follow

189 firms

Sectors

Ordered by how often our editor reaches for them.