Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

Tuesday, August 25

Tuesday, August 25, 2026

Everything we published on this day.

37 stories — the front page that day →

@convequity 2 clicks

Convequity on X

Cerebras is a taxi. NVIDIA runs the buses. Almost every confusion about $CBRS begins with reviewing the wrong vehicle. We just published Part 1 of our Cerebras deep dive: → Why wafer-scale SRAM excels at low-batch decode → Why $NVDA becomes the cheaper token factory at high concurrency → Why Cerebras was built for training—but found its future in inference → Why the two architectures are increasingly complementary, not substitutes Our conclusion: Cerebras is mostly a buy over the next 2–3 years—but demand for premium-speed inference will decide the case. The full high-level report is live. 👇 https://t.co/awUwxPUSHI
@pernasresearch 1 click

Pernas Research on X

I come across a class of company every now and then that I think of as a Schrödinger company. It’s a business where the market opportunity is partly endogenous to execution. In most businesses, you can reasonably separate the opportunity from the operator: you know there’s an attractive market, and execution determines how much of it the company captures. With a Schrödinger company, the two are entangled. Only exceptional execution may prove that an attractive business opportunity exists. If it fails, you may never know whether the opportunity was flawed or the execution simply wasn’t good enough. Two examples come to mind: $sfix and $tdup. In both cases there could definitely be something there but you have to prove it. And it’s very difficult to prove it.
@DeepSailCapital

Deep Sail Capital on X

$RUM is by far the most "chasing the cool thing right now" company of any out there, and they just use massive share dilution to get it done. First they tried to get into Crypto with the Tether deal a few years ago (100m share dilution). And now they are trying to get into datacenters, with Northern Data Acquisition (120m share dilution). Today they announced they are offering a counterparty 50m shares via warrants with a strike price of $0.01 just for the commercial deal to buy their GPU services. How can anyone trust this management?
@majgeoinvesting 2 clicks

Maj Soueidan on X

$SNT q2 call will be interesting. If there was a Hall of Fame list for Mgmt. that are cryptic and lack transparency on calls, they’d get a gold jacket. Still, I want to find out if the new LIDAR business can carry the weak/lumpy legacy perimeter security business. They teased us with transparency on the Q1 call, but weren’t sure they’d keep breaking out LIDAR moving forward. BTW.. $TPCS would also be on this list.
@BrokenMoats

Broken Moats on X

Deliveries from a teetering on insolvency Lucid is the savior for $UBER? ill take the under 0, 100 or 500 lucid AVs for Uber will not change the incumbency problem for UBER. They will now have to share the market of ride share, and ultimately last mile delivery that they currently dominate with Waymo, Tesla, Zoom (Amazon) in addition to many other players. These are well financed players that can use their mobility solution as part of a broader consumer bundle (not uberone, but incorporated into Prime). Removing the cost of drivers in the equation will dramatically lower the cost and likely the gross profit dollars available to Uber in a now hyper competitive market with companies that dwarf their capital base to run them out with subsidies, bundling and marketing (the same tactics uber used a decade ago to squeeze out smaller players). Their advantage of mindshare currently will be eroded by price and Uber does not control their own tech (bad for their economic model v competitors), is behind and will remain behind in terms of tech due to their inability to fund AI/AV at the same level as their new competitors. The concept Uber will win from fragmentation of AVs and being the app consolidator is going to be very difficult to play out. The regulatory hurdles, and costs of being 6 months behind and almost as safe at a higher price is not going to be a collection of vehicles attractive to users overtime, nor will their privellaged position as the app you open with little consideration to book a ride. And in an agent world, you wont pull up an app you'll ask book me the cheapest ride to the airport (and Uber wont win that box as discussed, app mindshare and learned behavior is meaningless) Uber won't be able to invest or subsidize their rides to stay relevant with Waymo/Tesla/Amazon for very long. Uber and their management team missed the boat (ironically by winning short term profits and street fanfare in cutting all of their AV investments several years ago) and the risk is not they stay in second and you melt the ice cube of cash flows here, this is existential like the sony walkman when the iPod arrived. So keep hoping for Lucid and the Saudis to come, it won't matter in the end. There is a reason it has underperformed and now trades in line with Software (recent move lines up with squeeze in SMH).