Tuesday, November 10, 2026

@StableBread Oct 31

Fajasy on X

Continue reading here for why $TOYO’s space pitch rests on cells it doesn’t make yet:

@StableBread Oct 31

Fajasy on X

The ~$120M of debt TOYO Co., Ltd $TOYO expects to use for its planned $357M Texas cell plant is still in “very early stage” talks. $TOYO plans a 1.5 GW heterojunction (HJT) cell plant co-located with the Humble, Texas module plant. HJT is a different cell design from the TOPCon cells $TOYO makes today, and $TOYO later wants to add a layer of perovskite (a second light-absorbing material) on top of its HJT cells to get more power per cell. More importantly, the plant would let $TOYO make most of the cells for its Texas panels in the U.S. instead of importing them from Ethiopia, and $TOYO’s Section 232 term sheet (its request for relief from the new 15% U.S. tariff on imported cells) is tied to the plant. $TOYO announced the plant on June 8 at a projected $357M, with pilot production “within 20 months.” By its October 6 Analyst Day, its domestic design was complete, permit applications were ongoing, civil construction was set to start only after all permits are obtained, and pilot production was targeted for Q1 2028. $TOYO plans to fund most of the plant with 45X tax credits (earned on each unit of solar equipment made in the U.S.) and debt. The 45X piece assumes the cell plant runs (cells earn $0.04/W), and the debt is, in the slide’s words, “in the planning stages; no definitive documentation has been signed yet.” CFO Yasunari Harada told analysts on October 6 that talks with Japanese and U.S. lenders were a “very early stage discussion.” As for timing, Rhone Resch, $TOYO’s Chief Strategy Officer, told a July 29 webinar audience $TOYO was “under construction” on the plant, and the August 19 results of its Japanese parent, Abalance, later put the construction start in April 2027 (planned), consistent with the Analyst Day plan to wait for all permits. So the Texas module plant is running and contracted into 2027, but the cell plant that would make $TOYO’s U.S. supply chain its own is still awaiting permits, with its $120M of debt in early talks.

@convequity Oct 31

Convequity on X

Further thoughts on the $CBRS short-seller report I've already quote-tweeted this short-seller report, but I think it's quite funny. These days, a lot of tech short sellers seem to have a poor understanding of the technology itself. Perhaps they're better at timing market liquidity and identifying moments when sentiment is most vulnerable. Going back to the report's central claim, I think it's arguing that Cerebras hasn't adequately addressed the challenges of using a giant substrate to support an entire wafer, or of maintaining reliable connections across such a large chip. It's similar to the challenges facing $NVDA's Rubin Ultra. The idea is to connect four compute chips, with a combined size roughly eight times the reticle limit. At that scale, thermal expansion becomes a major issue, as does maintaining consistent communication between the chips. For Cerebras, these are certainly significant challenges. That's why it took so many years to turn the original idea into something that actually works. I would say the third-generation Wafer-Scale Engine, WSE-3, was the first generation to become commercially attractive. Earlier generations had lots of problems to work through, much like other chips, for example Google's TPU from v1 to v3. Even WSE-3 hasn't solved everything. One of the biggest remaining issues is parametric yield. The simplest way to explain it is that, when a chip spans an entire wafer, you can't guarantee that every part will operate reliably at the same high frequency. (Frequency is how many cycles per second the chip runs, so a higher frequency means more compute from the same silicon.) You therefore have to lower the operating frequency of the whole system. A normal chipmaker would cut the wafer into many separate dies and set each one's speed on its own. The argument is that those dies could then run at frequencies 50% higher, or perhaps even twice as high. Cerebras keeps the whole wafer as one chip instead, so it gives up some of the compute the same silicon could otherwise deliver. In return, it gets an enormous amount of compute and on-chip memory connected as one wafer-scale chip. Data moves between the cores over wiring on the wafer itself, rather than through slower, more power-hungry links between separate chips. That is what lets Cerebras serve AI models so quickly, especially for inference. That's the trade-off, and parametric yield has been a constraint for Cerebras. The workaround has been to lower the specifications and operating frequency, leaving enough headroom for the system to work reliably. But that also means leaving some of the wafer's potential unused, despite manufacturing it on an expensive, supply-constrained TSMC N5 process, and potentially on more advanced nodes in the future. That's why I find the latest CS-4 rack-level system, using the WSE-3T, interesting. My understanding is that WSE-3T uses the same wafer design as WSE-3 but delivers more than 50% higher performance. They've apparently found new ways to address the parametric-yield issue, allowing them to unlock frequency headroom they previously couldn't use reliably. It looks as though they're actively addressing the problem and making good progress. The earlier implementation may have been suboptimal, but even that version was already demonstrating commercial competitiveness. This is a common problem with short sellers today: they build a thesis around old information while the technology is evolving rapidly. They point to a technical limitation that may still exist, but fail to distinguish between a problem that makes a product unworkable and one that merely reduces its performance. Between roughly 2015 and 2021 or 2022, these issues may genuinely have been showstoppers. By late 2026, I don't think that's necessarily true anymore. They may still hold performance back somewhat, but they don't prevent the product from working. Meanwhile, Cerebras has the advantage of being the only company to have made a wafer-scale chip work. In my view, that benefit outweighs the remaining limitations. The short seller doesn't seem to recognize that distinction. I think this also explains why Oracle, AWS, OpenAI, and other companies have begun testing Cerebras chips this year. The technology seems to be reaching a point where these problems are largely, though not entirely, resolved. Previously, perhaps they were only halfway to a solution, and the product wasn't sufficiently practical. Now that they're much closer, large-scale commercial deployment becomes more attractive to customers. Of course, it remains to be seen how this plays out. They still haven't crossed the deployment threshold I have in mind, and once they do, we may see new problems emerge. For now, though, my sense is that these are well-known issues among people familiar with the technology. The short seller is largely repeating things that those people already understand. The more interesting question is the direction of progress: how quickly are the problems being resolved, and what does that mean for the product's competitiveness? The report doesn't really address that. The rest of the report seems to cover the usual short-seller material, raising fear, uncertainty, and doubt about adoption and other peripheral issues. But the central technical thesis strikes me as quite weak. It reminds me of previous short reports on companies such as $SMCI.

@mavix_leon Oct 31 2 clicks

mavix on X

Portfolio end of October '26: Shelly $SLYG Evolution $EVO Secunet Security Networks $YSN BioRem $BRM.V Parigroup $CEV.HM Leatt $LEAT Toyokumo $4058 -undisclosed- Enogia $ALENO.PA Lindbergh $LDB.MI ErreDue $RDUE -undisclosed- -undisclosed-

@stockgutter Oct 31

SG on X

So you got an LLM to spin out a fancy DCF model for you, or maybe you hacked one together in Excel. How do you account for changes in outstanding shares? Most use a linear assumption. That might work for a company that throws off steady FCF, has limited SBC, and consistently buys back shares. For these, share count often shrinks predictably. But for money losing companies, its usually just a poorly reasoned throw-shit-at-the-wall guess. That's a problem. Dilution is the greatest threat to these names, and Wall Street's DCF models never give it serious contemplation. Example: show me an analyst who modeled in 2024 that $SPCE would 7x its outstanding share count. There are none.

@convequity Oct 31

Convequity on X

Thoughts on Elon's recent neocloud comments "Neocloud" has always been a weak label. These businesses aren't really clouds, and they aren't particularly new. The more accurate term is GPU VPS. Most of them rent bare-metal GPUs. They generally don't offer redundancy across three availability zones, and they don't sell hundreds of platform services on top of the hardware the way a true cloud does. That puts them closer to VPS providers than to cloud platforms. CPU history shows the difference. CPU cloud concentrated among three to five players because platform services lock customers in and drive a demand flywheel. Customers use more of the stack over time, rather than only renting the hardware. That is how AWS-style clouds kept high margins and strong moats. CPU VPS went the other way: many providers selling essentially the same product, in a brutally competitive, commoditized market. If GPU supply stops being constrained by chips or energy, GPU VPS starts to look like CPU VPS, with gross margins of something like 50%. Better marketing and operations can lift margins and grow revenue, but they won't push gross margins to 75-80%. That is the long-term risk Musk pointed to when he predicted most neoclouds would go out of business. Most of these names are VPS businesses earning cloud-like profits and valuations from a temporary imbalance between supply and demand. $CRWV is part of that group. CoreWeave and $NBIS are also the two furthest from a pure GPU VPS, because both already sell platform services on top of the GPUs. CoreWeave is still the most exposed on financing. It pushed hard to maximize margins on each contract and took on interest-rate risk against fixed contracted revenue. Its counterparties, OpenAI, $MSFT and $META, may have played those deals more intelligently. So CRWV is the most vulnerable name on its balance sheet, yet one of the least commodity-like on product. That gap is where the return potential sits for investors willing to stick with CRWV.

@John_Hempton Oct 31 1 click

John_Hempton on X

Some things deserve a blog post because they are so silly. This one is on Arbor Realty ($ABR). Most the credit on this name should go to @viceroyresearch and @Gabe_Bernarde.

@ACapitalLP Oct 30

A Capital on X

Look at $FDXF. A month of underperformance based on “channel checks,” followed by a ~15% move and multiple analyst upgrades after a single session (catalyst day).