Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

Saturday, October 24, 2026

Back to today

Tower Semiconductor Deep Dive

Tower Semiconductor’s low-loss photonics chips halve lasers in 1.6T optical modules, supporting a $680 million annualized photonics run rate and customer wafer prepayments through 2027.

TSEM $18B
CV's Substack 10 clicks

From X

@HaydenCapital

Fred Liu on X

It feels like Bromberg might be able to turn-around the Unity story after-all. Was impressed with his admission of Unity's prior mistakes, and fact is, the new Vector product is working. Also I wasn't aware of the long-relationship between JR and Bromberg. In these turn-around situations, it always seems to work better when the new guy is familiar / trusted, instead of a complete outsider without buy-in from management or the team. Curious to see how this unfolds... $U https://t.co/u9ntzroZnR ** So, as part of the runtime fee, you raised prices right off the bat. So, you gave them what they wanted right away. MB: Yes. But as you think about the business going forward, I mentioned before the fundamental tension does still exist. We don’t have smartphones growing at the rate they were, where you’re basically surfing this secular wave. You are delivering ongoing value, every time these games are run, 80% of games or whatever it is, they’re running on a Unity Runtime, which you’re not benefiting from. At the same time, it is zero marginal cost software, it’s not like it’s costing you for it to run. That’s running on the phone, on someone else’s power and whatever it might be. So what’s the solution going forward? What was the confidence that Unity was lacking that you think they can deliver, that they weren’t previously? MB: It is my view and I think it’s now really the whole company’s view, that there are other areas of massive upside in our business, and I’ll give you a few examples. The first and most important one is our advertising business. So, the real challenge coming into Unity, by the way, coming in, and this also goes back to the runtime fee, is nobody can figure out how the advertising business and the game creation business were connected to one another. Beyond the fact that you had the customer like, “Hey, click this button over here and sign up for ads”. MB: Yeah, but there were often different people inside the same customers. So, there’s the developer and then the person buying an ad, maybe they’re not even the same person. And by the way, we had done an acquisition and so we had two different groups of people doing this and when I first started, the investors would always ask me, “Shouldn’t you just split these things up? What do they even have to do with one another?”, and in many ways that core question was also one of the drivers of the runtime idea, because the idea was no, no, no, the connection is going to be in the business model, not in the product. So because what we’re going to do is we’re going to raise prices so substantially, but we’re going to say, “Hey, you don’t have to pay that if you buy advertising from us”. Yep, that’s right. MB: So, actually we’re like, “Hey, we’re going to make sense of this acquisition we’ve done”, we’re going to make sense of these business units that aren’t integrated, by creating a business model which unites them. But also, sadly flies in the face of what customers want and articulates no additional product value so that’s just a bundling, which just feels like you’re jamming something down my throat that I don’t want. If our advertising product was more effective and more efficient, people would use it on their own. What did AppLovin get right? Because this is sort of the period ATT comes along, Unity laughs at it, “Not a big deal, doesn’t impact our business” — turns out it did impact your business. Meanwhile, AppLovin comes along, acquires MoPub, just starts really crushing it, obliterating you in particular. What did they figure out that you didn’t? And how are you going to compete with them going forward? MB: Yeah, we missed a cycle of technology investment. While we were integrating acquisitions, while we were thinking about business models and ways of getting folks to buy more advertising by bundling products, they were building a completely new machine learning stack that was fundamentally more effective and efficient than the one we were operating on. We were on a old style algorithmic ML, really not even a deeply ML stack. Much more deterministic. MB: Yeah, they were moving to neural nets and into the future. Which was the way you had to deal with ATT, was you were losing that deterministic signal, so you had to be in a probabilistic world. MB: That’s correct. And so, that’s the thing you should be up all day and night thinking about. Not movie effects. MB: Yeah, exactly. Or, “Hey, my ad business is fundamentally uncompetitive, how can I strong arm customers into buying more of it?”, the answer is the product’s uncompetitive. How do I make the product competitive? Once the product is competitive, we have all sorts of opportunities and that’s what I mean about getting out of the box. And so, what we did was we built a modern self-learning neural net system from scratch, with some of the best engineers in the world, called Vector AI, and we launched it and it had an immediate, and market positive, impact in our business. You’ve talked about using Vector AI to basically incorporate gameplay into understanding the target. How does that work? MB: Yeah. So, the part one of this was, “Hey, let’s be fundamentally more competitive on our ad business”. Part two is, how do you think through, as you mentioned, those real connections between advertising, game creation, and the runtime? What actually connects those things? And what connects those things is the need to have a really deep and clear understanding of the gaming consumer, because that sits inside how you succeed in all phases of the game business. Whether you’re prototyping a new game, and want to understand how people are behaving in that game, and what’s engaging them, and what’s causing them to transact, and what’s causing them to quit, and what’s causing them to make friends, all that is a data challenge. How do I interpret data which we can have access to through the runtime, as a way of better understanding how to build a game? And then when I’m operating in live service, how do I use that same connection to consumer understanding to optimize my live service?

Investor Letters

All letters

Deep Dive: $ASML

ASML’s €230 million EUV machines underpin advanced AI chip production, with demand booked through 2028 after nearly $9 billion and two…

ASML $610B
Rebound Capital 9 clicks

Margin Call

Petrobras’s Morpho-1 well revived hopes that Brazil’s Equatorial Margin can become a major new oil frontier after hydrocarbons were…

PBR $81B
Doomberg 13 clicks

Our Views on CoreWeave

CoreWeave's financial statements point to a business model familiar to the authors after a closer review of the AI infrastructure company.

CRWV $34B
WHAT ARE WE DOING? “Contrarians At The Gate” 13 clicks
TCEHY · long

Tencent Holdings LTD

Tencent, valued near $500 billion, is a long idea built on its Chinese gaming, social-network and advertising franchises.

TCEHY $444B
Value Investors Club 1 click

Capital Wars

CoreWeave’s $35 billion to $39 billion 2026 spending plan relies on debt as unsecured yields reach 15%, making its GPU cloud expansion a…

CRWV $34B
Net Interest 12 clicks

Putting bear cases to rest

Medpace authorized $1 billion of buybacks through 2028 after Q3 revenue rose 8.5% to $716 million; service revenue grew roughly 12% as…

MEDP $18B
Best Anchor Stocks 15 clicks

Publications we follow

Every publication on this site, hand-picked.

129 publications

Newsletters 109

YouTube 4

Blogs 15

Seeking Alpha 69 of 272

Nominate a publication

X accounts we follow

Every account on this site, hand-picked.

91 accounts
Nominate an account

Investor letters we follow

250 firms

Sectors

Ordered by how often our editor reaches for them.