Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

Sunday, November 8, 2026

Back to the front page

NEW POSITION: Google LEAPS

Alphabet bought long-dated GOOGL calls after a 12.6% earnings drop, betting AI monetization and distribution overcome capex pressure and lift shares above $400 within two years.

GOOGL $3.5T
Hated Moats 1 click

From X

@ParadisLabs

Paradis on X

Quick thoughts ahead of $META earnings shortly: Key question: What does the monetization roadmap look like for Muse? Despite app downloads showing that AI capex is translating into real demand, my gut feeling is that Zuck will give a more hollistic view on AI monetization that lacks any real detail. This ultimately boils down to Meta's strategy of scale first, monetize later. Which then raises the usual capex questions particularly with Meta now positioning themselves as a frontier AI lab with their new Muse Wonder models. They'll need more compute if they're really taking this AI thing seriously. The current FY26 capex guide is $130-145B, so any increase to this without a clear AI monetization roadmap could be detrimental if I take a glass-half-empty view...particularly with consensus FCF deeply in the red at roughly -$9B for Q3. So how do Meta finance this exactly? Is it via their Blackrock agreement, or some combination of debt/equity? Again, something we'll need clarity on, layered on top of their monetization plans. I ultimately think that Zuck needs to really sell the longer-term AI vision today.
@convequity 5 clicks

Convequity on X

AI models are now writing the kernels that made CUDA so hard to leave, and we think this breaks another part of $NVDA's moat. Kernels are the code that tells a chip how to carry out each basic step an AI model repeats, and how well that code is written decides how fast the model runs. Nvidia prebuilt a huge range of these through CUDA. Anyone building their own chip had to write that software from scratch, which was slow, hard and a big reason to think twice before starting a custom-chip project. OpenAI showed at Hot Chips in August that this is changing. On Jalapeño, its Broadcom ($AVGO) built chip, an internal version of Codex wrote attention and mixture-of-experts kernels that ran 1.5 to 1.8x faster than ones written by human experts. Within about three months of getting its first working chips, OpenAI had three open-weight models running on Jalapeño, even though none of them was designed for it. Our view: if AI can write the kernels, the software barrier that held back custom chips gets much lower, and we expect more companies to start their own chip projects. Investors still do not pay enough attention to that consequence. Deeper dive available at Convequity.
@convequity 6 clicks

Convequity on X

We sold TSMC ($TSM) earlier this year and are staying out, despite strong demand for its leading-edge chips. Our view is that investors may be too confident that leadership in shrinking transistors will keep earning the same share of industry profits as more performance gains come from stacking circuits, connecting chips and integrating optics. Logic folding could also make older manufacturing processes more competitive, which would weaken the pricing power that TSMC's leading-edge investments depend on. In 2009 TSMC's leadership initially rejected advanced packaging as low-margin work before being persuaded otherwise. Together with today's cautious approach to expanding capacity, we read that as a deeply risk-averse culture. That discipline protects returns. It can also leave TSMC slow to back technologies that challenge its existing business, and by the time their value is clear, other suppliers may have secured the key customer relationships.

Investor Letters

All letters

Axon 3Q26 Business Update

Axon cut full-year guidance after delayed federal Dedrone contracts slowed 3Q revenue growth to 28%, while software and services grew 43%.

AXON $31B
Speedwell Memos 4 clicks

AMD Q3 FY26, Annotated

AMD's Q3 revenue rose 45% to $13.4 billion as data center reached $8.5 billion, 64% of sales, driven by EPYC and Instinct ramps.

AMD $660B
Chipstrat 2 clicks

Publications we follow

Every publication on this site, hand-picked.

129 publications

Newsletters 109

YouTube 4

Blogs 15

Seeking Alpha 75 of 279

Nominate a publication

X accounts we follow

Every account on this site, hand-picked.

91 accounts
Nominate an account

Investor letters we follow

359 firms

Sectors

Ordered by how often our editor reaches for them.