$UMAC Unusual Machines appears on planning to use OSC capital to parallelize their 100,000 motor / month automated line that they plan on having online by year end (4Q26). Last week, CFO Brian Hoff suggested UMAC seeks 6-8 additional lines which could produce 8-10M motors per year with extra flux capacity if needed (10M motors / 7 lines = 1.4M per line / 12 months = ~119,000 motors per month per line). Brian also suggested these additional 6-8 production lines would fulfill Group 2 and Group 3 drones - which are larger in size - therefore potentially dragging ASP (average selling price) higher on components. Not financial advice - but I took a swing at modeling out what the Defense ONLY portion of Unusual Machines' motor business could look like with an OSC capital infusion. By Q3 2028, this model assumes UMAC has 8 automated motor equipment lines up and running, kicking 9.6M motors per year at an ASP of $57. We learned from Allan that in 2Q26 UMAC made and shipped just under 50,000 motors in the quarter. Further, 13% of 2Q26 revenue was attributable to their largest product. Assuming that's motors, you arrive at an ASP of ~$43.47 ((0.13 x $16.7M) / 50,000). That's why this model starts at ~ $43 ASP in 2Q26 before being dragged up ~33% in totality by the end of 3Q28 on the assumption of additional lines being used for Group 2 and 3 drones. To state the obvious, there's a lot of guesswork here. On 9.6M motors at a $57 ASP, you land at ~$550M run rate on motors alone in 3Q28. Gross Margins are assumed at 38% throughout - the midpoint between 2Q26's ~35% and UMAC's 40% LT target. That yields ~$208M of annualized gross profit in 3Q28. CapEx for the tangible, physical equipment of each line is estimated to be ~$4M, with the rest (overhead, salaries, raw materials) assumed to be financed by the OSC. The model is far from perfect - but Cumulative Gross Profit less CapEx is treated as cash generation and added back to implied Enterprise Value along with 2Q26 Net Cash. OpEx as a % of Revenue is assumed to be ~19% for this production line - a majority of UMAC's overhead runs through COGS (which improves as employees become more efficient). D&A assumed at 1% of revenue. That leaves us with an implied 20% EBITDA Margin on the motor business (38% Gross Margin - 19% OpEx as a % of revenue + 1% D&A add back). Applying an assumed 25x EV/EBITDA multiple on the implied 3Q28E EBITDA (with 8 automated lines up and running) spits ~$2.7B in implied Enterprise Value. 25x in my view is justified as by this time UMAC will be growing very fast - with drone delivery and potentially robotics as additional growth tailwinds. After adding back Net Cash, the Cumulative Gross Profit less Capex and subtracting Working Capital OpEx (estimated a 1x annualized revenue) leaves ~$2.67B in implied equity value in the Defense Motor business alone. That suggests UMAC's Defense ONLY Motor business may be worth $38.30/share. Of course - there are SO many assumptions going into this (plus the $ figure isn't discounted) and the math almost certainly doesn't reconcile 100%. But - I think this analysis is directionally correct. In my opinion, the drone industry is at the beginning of the S-Curve and Unusual Machines is ready to ride it with the help of some OSC capital.

want to hear a joke? the efficient market hypothesis

A while back we thought $TSLA had more near-term catalysts than $SPCX. Our thoughts have changed.

I'll share something somewhat obvious... Generally we don't sell low unless we think something is wrong. Sell low isn't in our playbook. When we trim or sell it's usually after a rally. We've missed a few trim opportunities this year.

US and Iran discuss phased deal to reopen Hormuz and end US blockade, sources say

The 10-year Treasury note yield rose 0.147 percentage point on Wednesday to 5.113%, the highest level since July 2007. It was the largest one-day increase since April 9, 2025, when Trump announced he would pause the Liberation Day tariffs.

$UMAC $PUSA $XTND Haven’t seen anyone post this article yet, this drone deal is something to follow, specifically because I remember @TheDroneWarrior & Co said this will lead to them acquiring some interesting tech… cbsnews.com/news/ukraine-us-…

Commonwealth Credit Union Expands Partnership with Upstart into HELOC and Indirect Auto Lending $UPST

One year ago, BTIG sparked mass panic surrounding Upstart $UPST arguing delinquencies (DQs) were materially rising. They weren’t - and they haven’t. Two days later, the analyst walked back on his comments but the damage to the stock was already done. Why bring this up? @paulxgu I don’t think the sell-side is on your team. UPST lacks vocal institutional support, and there are many of these ‘mini events’ that explain the perception gap of your stock - it’s not just retail holders upset with price action. Of course that’s just my opinion, but think about how many of the analysts said you were toast a few years ago. I get that’s the business but it doesn’t sit right with me.

In our data driven culture, the internet will likely convince you that if you're not trading in and out of stocks, using margin or using derivatives that you're leaving something on the table Let me say this with the ultimate conviction - this could NOT be any further from the truth. It is still a completely viable stategy to continually buy stock in high quality companies over 20 years and do nothing but wait (assuming that company continues to innovate and execute) I have nothing against anyone using the aforementioned tactics, but don't ever let yourself feel like you're missing out by not doing more than DCAing. It's still effective to be "boring" and just buy and hold. And yes, this includes through periods where $TSLA stock moves sideways for five years Could you have made more elsewhere? Maybe. But will you also still end up doing great if you just kept buying during that time and you hold those shares for another 5-10 years? Absolutely It's still cool to think about investing in terms of decades, not months. Your mental health will also likely be much better taking the simple, proven approach to building wealth where you're not a slave to every news cycle

There's been a lot of chatter about Upstart $UPST President, Capital & Enterprise (former CFO) Sanjay Datta and his recent 'insider selling' - but the truth is more nuanced. When Upstart went public, the S-1 (registration statement) revealed the outstanding equity awards for the named executive officers. Focusing on Sanjay, you can see that he was granted two option awards on December 28, 2016 (image one). The first was the right to acquire up to 650,907 shares (474,619 + 176,288) once vested at a price of $1.35, and the second was the right to acquire 100,139 shares (60,500 + 39,639) once vested at $1.35 - with both options set to expire on December 28, 2026 (this year). The first vested 25% on January 1, 2018, the second on July 1, 2018, both monthly thereafter. There are two kinds of stock options. Incentive Stock Options (ISOs) and Non-Qualified Stock Options (NSOs). The two have big differences in tax treatment. ISOs - you do not pay regular income tax when you exercise (buy the shares). If you hold them for the required time, your entire profit is taxed at a lower capital gains rate. NSOs - you pay regular income tax on your profit the exact moment you buy the shares (the spread between the current price and the grant price). Once exercised, your cost basis then equals fair market value at exercise (strike price paid plus income recognized). I am going to assume that at least the larger of the two of Sanjay's grants was almost certainly an NSO, meaning he owes tax on the difference immediately upon exercise (his expiration date is December of this year, exercised in September). The tax code caps ISOs at $100,000 of stock (valued at the grant date) becoming exercisable in any one calendar year. Sanjay's two grants vested much more than that in 2018 alone, so by law most of them have to be NSOs (at least that's my thinking). On September 8th, 2026 - Sanjay exercised a portion of these options representing 300,000 shares of the 650,907 granted (46.1%) acquiring the shares at $1.35 for ~$405,000. Sanjay then subsequently sold 116,000 shares (of the 300,000, 38.7%) on the open market for ~$27.29, a total sale of ~$3.17M. Exercising isn't free. Sanjay had to pay the $405,000 to acquire the shares, and with it likely being an NSO, he owes tax on the spread between $27.29 and $1.35 = $25.94 x 300,000 shares = $7.78M of taxable ordinary income. That $7.78M is going to be taxed as ordinary income at his marginal rate. Assuming he's already in the highest Federal Tax Bracket (37% - impressive Sanjay), and assuming he lives in California, another 13.3% (image two). So, in total, Sanjay will owe ~50.3% of tax against the $7.78M in ordinary income, or ~$3.9M. Add in the $405K it cost to exercise the options, and total out-of-pocket cash for Sanjay I estimate at ~$4.3M. The 116,000 shares Sanjay sold at $27.29/share brought in ~$3.17M. That leaves a ~$1.13M potential cash deficit that Sanjay potentially paid out of pocket (or will come tax season). He may have been able to fund this with previous planned sales earlier this year. Considering I am not his CPA, I'm not entirely sure lol. Sanjay had the ability to sell more than the 116,000 shares to satisfy these tax obligations - and he chose not to do so (at least yet). If he wanted out, the obvious move would have been to sell all the 300,000 shares. It's also important to note that if Sanjay did not exercise the options, they would expire worthless. Now, Sanjay has been a frequent seller of UPST in the past - mostly through pre-scheduled 10b5-1 plans (though not all). Throughout 2023 to 2024, Sanjay was selling 1,000-3,000 shares or so weekly. Shares of UPST more than doubled from the summer of 2024 into early 2025 - so the selling wasn't indicative of lackluster business results. Narrative always follows price. That's not to say UPST is immune from criticism (they certainly aren't) - but it's simply not true that Sanjay's recent Form 4 activity is indicative of an executive racing for the exits. I'm not claiming my math here is 100% correct, but I think it is accurate directionally. I of course probably have bias to look for the positives in UPST as both I, and my firm, are long the stock - so please keep that in mind - but I am not concerned about Sanjay's Form 4. The insider buying from Gu and Dave is more of the signal imo.

This fall - we will be busy

$UPST - a few updated thoughts post FOMC: 25bps isn't going to do much given the bond market has front-run the tightening. Look at both the short (2yr) and long (10yr) end of the curve - both are rocketing higher. Higher borrowing costs = higher cost of capital = higher APRs = lower conversion (all else equal). What's more important is the direction of the Fed - and Chair Warsh (imo) was modestly hawkish today. The median FOMC participant sees the FFR 25bps higher by year-end (so another 25bps raise). Ultimately, I do think UPST will hit the guidance this year, but macro is eating a lot of the cushion. Doesn't impact the long-term thesis, but a frustrating bump in the road for investors that have been hanging in there for a long time. UMI (macro risk) is ultimately a measure of the consumer's financial stress and how that's affecting their ability to repay their loans. The higher UMI goes, the tighter the credit box (obviously). Upstart's FY26 guidance assumes UMI of 1.4-1.5. As of September 3, 2026 - UMI was 1.50. It does seem unfortunately probable in my view that UMI drifts a tad higher from here as the Iran War continues but no one knows for sure. So, rates are going to be say another 25bps higher by year end according to the median SEP. Raising the cost to borrow constricts economic activity and it can cause a recession (2022) - which in turn would drive UMI higher (less ability for consumers to repay loans). Or, if rates go higher and that tames inflation w/o a recession, it could improve aggregate consumer health (less stressed, better repayments). All to say the macro has materially worsened in the last few months which is a clear negative for Upstart the market is pricing that in to an extent. The macro-overhang is in large part driven by the war in the Middle East. Upstart is not going to be immune from the macro, but I don't think they will be as vulnerable as in years past with a more diversified product set and more committed capital. Plus, I am more confident in Paul as an operator. This was a really good pod w/ Paul the other day, who discussed a lot of topics relevant to Upstart and macro - highly recommend a listen: nitter.net/petkevichdan/status/20… TLDR; long-term bullish, macro worsening short-term, confident in Gu

Great discussion with a great @Rebellionaire team! 🦁 Very broad-based collection of viewpoints & possible #AI Artificial Superintelligence (ASI) outcomes.

'The Superintelligence Panic: Rebellionaire Panel Discussion' AI CEOs say we're facing extinction-level risk. Are they genuinely scared, or is this regulatory capture? The Rebellionaire Investment Committee meets once per week and this is what those conversations actually sound like. Monday morning, no prep, off-the-cuff, behind the scenes. *Timestamps* 00:00 ASI Panic Weekend 01:44 Is Insider Alarm Legit? 04:56 The Answer Key Problem 05:35 Henry's Three Scenarios 09:49 The Hugging Face Incident 11:49 Open vs Closed Models 13:14 Understanding AI Behavior 17:06 AI Efficiency Improvements 19:00 Investment Ban Risks 20:22 Compute Demand Outlook 24:46 The AI Cold War and Simplifying the Problem 27:46 Valuations and Sentiment 29:50 AI-flation and Rates 33:33 Wrap and Team Intro @bradsferguson @nerdalert @HenryInvests @ArranGray9

New on Cars and Capitalists: Paul Gu (@paulxgu), CEO and Co-Founder of @Upstart. @Upstart is an AI lender worth $2.5B today. They did $1B in revenue in 2025, up 64%, grew originations 86%, and were profitable. Paul took over as CEO from co-founder Dave Girouard (@davegirouard) in May. Before that, Paul was Upstart's CTO and built its first credit models. He's a Thiel Fellow and Yale dropout who came out of quant finance. In this episode of Cars and Capitalists, we discuss: 0:00 - "The model's gonna suck": why you can't copy Upstart 2:02 - Three sheep, two breeds, one A/B test 8:13 - The core business: 50 million Americans the prime market won't serve 21:20 - Inflating away the debt vs. the hard choices nobody will make 29:58 - Cartoon math: $100 of loans, $5 of contribution, a 20% fund 41:33 - The real moat: Jane Street techniques pointed at consumer credit 58:03 - The golf trip problem, and paying dealers to pick you 1:06:03 - "No one wants to fund a new product. It's like nuclear waste" 1:16:51 - Killing auto refi, the product they were best at 1:24:42 - 2040: relevant to economic history, and unambiguously pro-human Thanks to our sponsor @rootvc.

I'm torn because obviously AI should be regulated, but should we really let the government control all the keys of the castle?

🚨 Insider Buy: Upstart CEO Paul Gu @paulxgu today purchased 50,000 shares of $UPST at $25.55 for $1.27M 👀 💪

*ANTHROPIC SIGNED A $13.7 BILLION, SIX-YEAR COMPUTE DEAL WITH RUM GROUP $RUM

Behind-The-Meter (BTM) Power Generation is growing extremely fast. @SemiAnalysis_ claims there are 3GW of operational US datacenter IT capacity powered BTM today, with 75GW of firm, binding orders in place (wow!) of which ~20GW alone was ordered in 2Q26. Idle GPUs are far more expensive than paying up for power. If SemiAnalysis is right that inference on a 1GW data center prints $100B of revenue a year at 90% gross margins - the cost of power really does not matter. If you're going BTM - you are essentially going around the grid (extremely long connection times) by building your own gas turbines. From here, you are either operating on commission or bridge capacity. Commission = using BTM to test the DC before it opens (3-6 months). Bridging = using BTM as your primary power source once DC is live until you can connect to the grid (2-3 years). No matter how you slice it - BTM gas turbines need natural gas. And getting natural gas to a data center has its own queue much like the grid. Running a new pipe to your DC site takes permits, land deals and construction - usually 12 to 36 months. So, the turbines show up in ~9 months (if you're lucky), and the natural gas shows up in call it ~2 years. Liquefied natural gas (LNG) fills the gap. Natural gas is chilled into a liquid (LNG) so that its size shrinks, enabling a lot to fit into a trailer that is hauled to the site, and warmed back into gas before hitting the turbine. Demand for trucked LNG can rise as fast as those 75GW get built. Supply can't. There are only a handful of plants in the US that can make LNG for trucks, and building a new one takes as long as the pipeline it's meant to replace. Therefore, if you're running a data center on BTM power, it may not be a bad idea to lock in a significant allocation of LNG delivery before everyone else in that 75GW queue figures this out - many already have!