Monday, November 9, 2026

@finphysnerd Oct 16

Myles on X

@ArgonathCapital $IWM covers the junk and also covers market breadth which I'm bullish on.

@ParadisLabs Oct 16

Paradis on X

Michael Dell on the memory bottleneck: “It’s gonna get worse.” $DELL

@convequity Oct 16

Convequity on X

Our least-favoured bet on next-generation nuclear: light-water SMRs. Smaller does not automatically mean cheaper. A 50–300 MW small modular reactor produces far less power than a 1 GW plant, but containment, site infrastructure and other costs may not shrink proportionately. Factory production and simpler safety systems can help, but those savings must outweigh the loss of scale. The result could be a higher cost per installed kilowatt, not a lower one. Advanced alternatives offer higher temperatures and more industrial uses. But we think many face a bigger obstacle than licensing: securing enough fuel. To understand the investment trade-offs, separate two questions: what cools the reactor, and what fuels it? Light water → Proven technology, established fuel supply. Uses ordinary water and fuel typically enriched below 5% uranium-235. Runs at around 300°C under pressure, producing electricity and lower-temperature heat. Its advantage is maturity; its limitation is that it cannot directly supply the hottest industrial processes. Shrinking it does not necessarily improve its economics. Sodium → Higher temperatures without high pressure. Transfers heat efficiently and operates near atmospheric pressure, reducing the need for heavy pressure-containing equipment. The catch: sodium burns in air and reacts violently with water, requiring strict separation and additional safeguards. Helium → High-temperature heat without a chemically reactive coolant. Can reach roughly 750°C or more, supporting more efficient electricity generation and industrial heat. Helium does not burn or react chemically, but the system operates under pressure and must stay leak-tight. Leading designs also require specialised TRISO fuel. Fluoride salts → High-temperature heat at low pressure, without sodium’s fire risk. The challenges are corrosion, suitable materials and preventing the salt from freezing. Some designs, such as Kairos Power’s, use salt to cool solid TRISO fuel. Others dissolve the fuel in the salt itself. These are different designs with different fuel requirements. Then come two separate fuel bottlenecks: HALEU is about uranium enrichment. It contains more than 5% but less than 20% uranium-235. Many leading sodium-, helium- and fluoride-salt-cooled designs need it, although not every advanced reactor does. Western production remains limited. Centrus Energy ($LEU) has produced small quantities with demonstration equipment and targets new enrichment capacity in 2029. We do not expect sufficient supply for broad deployment before the late 2020s at the earliest. TRISO is about fuel manufacturing — not a type of coolant. Tiny uranium-bearing particles are wrapped in carbon and ceramic layers designed to retain radioactive fission products even at roughly 1,600°C. That is a major safety advantage, but manufacturing and quality control are demanding. US production already exists, but capacity needs to expand substantially. X-energy’s ( $XE) larger TX-1 facility is still under construction. Helium-cooled and some salt-cooled designs therefore face both bottlenecks: getting HALEU, then turning it into finished TRISO fuel. For investors, a reactor licence is not a fuel supply chain. And a smaller reactor is not automatically a better investment.

@StableBread Oct 16 1 click

Fajasy on X

Even if the AI model Nubeva Technologies $NBVAF (TSXV: $NBVA) is training to read malware binaries (compiled programs) works, six risks could keep it from paying off: → Labs closing the gap: OpenAI’s GPT-6 Astra nearly doubled GPT-5.6-sol’s score on SRE-Bench, a Columbia University-led test of AI models on reading binaries, within a month of SRE-Bench’s release, and Anthropic says its Claude Mythos Preview model can already reconstruct plausible source code from stripped binaries. The better the labs’ own models get at reading hard binaries, the less they need to pay for outside labels. → Big tech building in-house: Microsoft’s $MSFT Project Ire reverse engineers software files on its own to decide whether they’re malicious. Ire caught only about a quarter of the malware among nearly 4,000 hard-to-classify files in a test Microsoft described in August 2025. But in June 2026, Microsoft said Ire flagged a new malware sample that most major security products, including CrowdStrike’s $CRWD and SentinelOne’s $S, still missed. Google $GOOGL already owns VirusTotal and Mandiant, so the likelier customers are labs that don’t own a malware-scanning site or an incident-response firm already, like OpenAI and Anthropic. → Paid like a contractor: Data labeling companies (firms that turn experts’ knowledge into training data for AI labs) pay 60-70% of what they bill to their contractors, and Scale AI is already recruiting its own malware analysts. So a data labeling company could pay Nubeva by the hour or task, or hire analysts itself, instead of licensing the model or buying the company. → Selling labels teaches the buyer: Every binary Nubeva labels helps the buyer’s model learn the same skill, the reason CEO Randy Chou gave for not patenting how Nubeva reads binaries. So each sale could shrink the edge Nubeva is selling. → Cash and compute: Chou gave no cost for running “hundreds and thousands” of AI agents on a private model. For reference, Nubeva spent $714,772 on R&D in FY2026, equal to 28.7% of its $2.49M in cash as of April 30, 2026. → Missed targets: Nubeva’s AI decryptor builder missed four targets starting in August 2023, so “days and weeks” (Chou’s timeline for the model) is a target, not a result just yet. So even with a working model, Nubeva still has to (1) sign a contract before the labs can read the hardest binaries themselves, (2) get paid for its model and engineers instead of by the hour or task, and (3) fund the training without draining its cash.

@calvinblissett Oct 16 2 clicks

Calvin Blissett on X

I kind of love this, sets up the capital slosh effect that happened in 2000s with the great rotation from tech to value. I think something similar happens here eventually and offers uncorrelated returns which will be more valuable. I will be eating dirt on value until then, clipping my coupons and buybacks. $KSPI $BOC $SLM $HHS $ZIM $CCLD $WU $HY $AMR $CNR $METC $HCC $FISV $YB $BTU $CNRD

@UCompounders Oct 16 5 clicks

Undiscovered Compounders on X

To the surprise of absolutely no one, the Copart-ACV deal is closed. $CPRT $ACVA Get ready to hear the word “synergies” everywhere. Now it’s up to Copart to show they’ll be good ones.

@MultiplesCap Oct 16 3 clicks

Multiples Capital on X

Elon: "We just demoed a gravity deying car" The Market Today: "Thats cute, but look at the free cash flow on companies that sell cars with four wheels on the ground" 😄 $TSLA

@convequity Oct 16 3 clicks

Convequity on X

The market still prices $NVDA's moat on one idea: the GPU is special because it can do anything. But Nvidia has spent every generation trading that flexibility for raw AI horsepower. Its chips now look less like general-purpose GPUs and more like the AI-specific chips Google and Huawei build. That's great for performance, but it chips away at the very thing that set Nvidia apart. The closer its GPU gets to a TPU, the easier it becomes to replace. Deeper dive available at Convequity.