Asymmetric Bets

Decided to make a Substack (see bio). Will be posting free articles on market and paid articles on individual stocks First write up is on $brun, which I think will 3x in the next 12 months following a similar path as $NBIS $CRWV Write up covers founding origins, business model explained simply, $nvda Exemplar Cloud partnership, customers and partnerships, compares coreweave and nebius trajectories to BRUN, covers unit economics of BRUN's contracts including IRRs, insider lock up debunked, sizing the neocloud market and brun's opportunity within it, rebuttals to common questions, valuation framework Let me know what you think @JonahLupton

@JonahLupton @LIWEI_TWCapital

Lots of ai infrastructure stocks just pulled back 30-60% when the underlying theme is even more bullish now than it was when these stocks were trading at their highs… I think some of them rally back hard to new highs in the next couple months with Q2 earnings being a big catalyst.

Kimi K3 has received overwhelming demand since launch. In the past 48 hours, usage has pushed close to our current computing capacity limits. To protect the experience of existing subscribers, we’re temporarily pausing new subscriptions and prioritizing compute for current members. Are those AI infra stocks that tanked because of Kimi gonna bounce back tomorrow?

@TMTLongShort Republicnas sweeping in November is the least likely of these predictions

We’re decoupling from China. Escalation in Iran is inevitable. Tariffs are coming back. Anti distillation will work. Republicans will sweep midterms. Europe is going to commit seppuku. Canada will get annihilated on USMCA. Taiwan might be sacrificed. Hegemony will be secured.

With the delays at Juniper (Oracle’s 2.4GW DC that bloom energy provided fuel cells to) I suspect $BE will approach $125 in short order

The growth of the GPU debt market is pretty staggering, if there is a bubble, it will be here. Second largest credit market in the US behind mortgages

@ren_stocks there will be short covering into the hyperscale earnings July 22/29

The next person saying "we are so back" gets electrocuted by Pikachu ⚡️⚡️⚡️ Tell me that when $NBIS is close to the 300s. $AAOI close to the 200s. Not after a one-day rally.

@babyfolio At renewal, customer goes with someone cheaper maybe, not going to pay the premium anymore (coreweave historically charged 25% higher than say the iren/msft deal)... that pricing power seems gone now to me idk

@babyfolio It makes it easier to become a neocloud, no? Nvidia is open sourcing it. Coreweave and NBIS first got the pref allocations because of their expertise that Nvidia is now open sourcing for free so any software premium is gone, no?

For those who are worried, it'll be fine. I'm extremely confident in $NBIS's ability to execute. This drawdown isn't company-specific, it's happening across the broader market. I haven't seen anything that changes the long-term thesis, so I'm not worried in the slightest. If you want to sleep better during volatility, do the work. The more you understand the business, the less you'll be shaken by price swings.

Great post! You mentioned open source benefits the neos and I agree — but does NVIDIA’s move with DSX hurt its largest neo customers $nbis $cwrv $shaz $brun etc. I am getting neocloud whiplash the more I dig into DSX/RunAi. Nvidia making it much easier to become a neo it seems @jaminball

Kimi K3 may be an important inflection point for AI. Potentially negative for Anthropic and OpenAI while being net positive for essentially every other company in the world. I mean that very literally. Although the real “Sputnik moment” would be an open-source frontier model that was also token efficient unlike Kimi K3 which is 50-70% more expensive to run than GPT 5.6 per Artificial Analysis. Rationale:   A world where there are only 2-3 dominant frontier labs with 90% inference margins is net negative for every other layer while being awesome for those 2-3 labs. Those labs would become monopsonies for power, data centers, semiconductors and hyperscalers and would obviously vertically integrate over time into all those layers while also completely subsuming the application/software layers.    Anything that lowers margins and increases competition at the model layer is good for every other AI layer: power, semiconductors, hyperscalers, neoclouds and yes even software.   This is why Jensen is so supportive of open-source. An open-source model requires the *exact* same amount of compute to run as a closed frontier model of similar size and architecture. Kimi K3 is roughly the same price as GPT 5.6 Terra on a per token basis, which actually suggests that it is less computationally efficient as I am sure that GPT 5.6 is priced to a higher margin than K3. And given that K3 is a token wastrel, i.e. token inefficient, it is significantly more expensive per task than GPT 5.6 and Grok 4.5, which are much more token efficient. Cost per token and token efficiency (i.e. intelligence density per token) are the drivers of intelligence per unit of cost. The winning AI companies will be those that offer the most intelligence per $ over time.   Lower margin % at the model layer = more margin $ at every part of the infrastructure layer and is a godsend for software. This can happen either through open-source models like K3 at the frontier *or* having a vertically integrated model company like Meta, SpaceX or Google at the frontier. Both outcomes result in a lower margin % at the model layer as vertically integrated model companies don’t really care where the margin $ come from. This is why it was so painful for OpenAI and Anthropic when Google was right there with them from a model competitiveness perspective and why Grok 4.5 and Muse 1.1 were just as important as Kimi K3. 
The reason Kimi K3 is only *potentially* negative for Anthropic and OpenAI is 1) the @ericvishria point that the Claude and ChatGPT products and harnesses may be more important than their models today and 2) the hypothesis that they have much more advanced model checkpoints internally that are already being used for RSI. In the latter scenario, reaching RSI even a few months ahead of other labs might be enough to cement a permanent lead. Time will tell on both points. And likely fairly quickly. Caveat would be that since Kimi K3 is not token efficient and thereby actually more expensive than ChatGPT 5.6, we may need to see a more token efficient open-source model at the frontier or see Grok 5/Composer 4/Muse 2 at multiple points on the Pareto frontier for this potential risk to Anthropic and OpenAI to play out. And I am sure they will both vertically integrate as quickly as possible while continuing the product/harness strength they have shown over the last 8 months.

$hwm is an obvious short - the valuation is insane - " they make turbine blades in an oligopoly, Im the only one that knows this"... lol you are not. I am short. Valuation is nuts relative to $MSFT $META $NVDA $CRM etc. 2.7% FCF yield on 2030 in a bull case for a cyclical industrial - GTFO

@UFCHedgeFund It is indeed

@UncleAlpha007 HWM is not a momentum stock

How did $HWM survive that momentum blow out? Yes they make turbine blades and only three companies can do that at scale.... but pretty sure at 100x FCF the market knows this 😂 $HWM is in the same spot as $BE was a couple weeks ago when I posted the valuation makes no sense

@theakxis @signulll Wouldn't rule it out - setup is actually quite similar

@UncleAlpha007 @signulll He could buy your bitch 😆

kimi’s founder & ceo got his phd at cmu. why didn’t he stay in america?!

@rubicon59 Nemotron is coming

Why can’t $NVDA produce a leading open source AI model?

@matthew_sigel Yes agreed on the price action I’m talking more about the long term competitive dynamics when chips come off lease — doesn’t DSX pressure pricing

Appreciate the thesis but don't think it's a big factor. DSX was "open-sourced" May 31 and Mirantis was named as a day-one ecosystem partner...Every Linux needs a Red Hat eg. One week later NVIDIA put $2.1B into IREN. I think the sector weakness is more straightforward: big liabilities, minimal cash flows, probably bad earnings (delays, weak bitcoin price, higher costs). Unwind.

New X android update Fun to see (some) of the DMs I missed 🤝

To me it seems like Nvidia DSX/RunAi is massively commoditizing the Neos bc the reason they originally got the pref allocations was their software — Nvidia did not roll out DSX because they wanted to neos to exist because they wanted customer diversification from the hypers who are trying to create their own chips. The rollout of DSX l, which I guess happened late May (I didn’t follow it closely enough), now has me a bit concerned about the neos roll. Nvidia is giving away DSX for free

@UncleAlpha007 What's the question?

@SkinnyfatTony Brutal

@UncleAlpha007 blatant theft. hidden in plain sight. i did not catch simple things, i blame myself for this. some things were hidden in payroll and bank transfers that i also did not catch. hard to talk about.

Do you use an ERP system? One of the BIGGEST mistakes I made in my business was switching to one before I understood what I was getting into. I went from Quickbooks Desktop to a full blown ERP and this is where things started to go down the wrong road, for many reasons. First - I was "Engineer to Order" - and traditional ERP's are terrible at that. They're fantastic for "Manufacture to Order", but fall on their face when you try to force their workflow to adapt to custom machine building. I eventually switched to another ERP after my 3 yr term was up with the first, and it made matters worse due to my small company size and lack of 30 person IT department, new implementation, data transfer nightmares and so on. I hired people who ultimately found ways to embezzle due to my lack of understanding of these systems. Lesson learned - do more research before buying a massive ERP, understand what will work and what won't for your workflow. There's a cool software out called Total ETO - I really wish I had heard of it before going with those other two systems. I'd imagine with today's AI vibe coding many are building their own systems. If I were to do it over again, I'd definitely build my own, maybe still interface with an accounting software, but I would NEVER buy another ERP again. What's your experience with them?

@P_Bonnet What’re your thoughts on DSX/RunAI?

H100s are four years old, and their rental cost is up 30% since the beginning of the year. Not exhibiting weakness recently. B200s are over twice as expensive and the trend is up. Hard to see anything more positive for neoclouds who have capacity in the market currently?

@jaysyoon @RealJimChanos what do you think about RunAi/DSX as it pertains to your NBIS position. I got some neo whiplash from it now

A reminder that @RealJimChanos Jim Chanos has turned into a senile grifter who routinely lies about his analysis. Attached is an analysis that Jim shared a few months ago for the $IREN $MSFT deal where he calculated an unlevered IRR of 10% assuming a 5-yr GPU useful life. Now, he is claiming that $NBIS earns less than a 10% ROIC even using a 10-yr GPU useful life, despite NBIS getting a 2x greater prepayment than IREN on its deal with MSFT and getting a significantly higher $ / MW as well. If you run those numbers through Jim's spreadsheet using a 5-yr useful life, the ROIC is well north of 10%. If you assume a 10-yr useful life, the ROIC is even higher. In short, he is a clown. Disclosure: Long $NBIS. No position in $IREN