August 24, 2026 Meeting

Introduction

We held the latest meeting of The Investing Group on Monday, August 24, 2026, at the Union League Club of Chicago in Room 816.

The evening moved from a one day micro cap halt festival into Iran and the rial, crypto positioning, Grok and SpaceX, Tesla and autonomy, Moderna's melanoma vaccine, Flock cameras, Bally's stalled Chicago casino, and a lightning round that ran from drones and Chinese humanoids to lithium and peptide manufacturers.

If the night had a single thread, it was this: the obvious story is usually already in the price. The work is finding where the second order effects land, who actually captures the cash, and whether the vehicle you are using to express a view will survive the path required to get there.

Small float speculation still obeys halt mechanics more than fundamentals. Crypto exposure can be cleaner through miners, trusts, and listed vehicles than through the coins themselves, depending on what you are actually trying to own. Frontier models are compressing toward similar intelligence scores while the cost curve and the surrounding ecosystem decide who wins commercially. Healthcare may be one of the cleaner ways to own AI without having to pick the next hot infrastructure name. Physical casinos are getting squeezed by phones, bars, and prediction markets at the same time. Surveillance tools keep spreading whether people like them or not, which is exactly why some people in the room own the companies building them.

Nothing in this group is financial advice. Invest at your own risk. That line opened the meeting, and it still applies to everything below.

1. CID HoldCo, Halt Mechanics, and Nearly 23 Hour Trading

The discussion opened with another stretch of violent small and micro cap trading, along with Nasdaq's plan to expand toward a nearly 23 hour session. The tape supplied a fresh example before the geopolitics even started.

CID HoldCo (DAIC), which does business as Dot Ai, was the crazy one on the day. This was a different flavor from the Chinese micro cap wave that has dominated recent sessions. The company is an asset tracking platform using patented passive and active RFID, AI, and edge cameras, serving construction, military, mining, and other industries, with headquarters in Las Vegas. The description is mostly buzzwords. That does not stop a micro float from running hundreds of percent.

The stock opened around $1.08, immediately went into a halt, unhalted around 9:40 a.m. ET (a 10 minute pause after the 9:30 open), and jumped in a single burst. Some in the room, including Eric and Quinn, traded it through the swings. The position did not work. Shares only briefly touched above $3.00 at the high before fading back through a long sequence of Limit Up Limit Down pauses to close around $1.73, still up about 306 percent on the day. After hours they surged another 126 percent to $3.90 and were still around $4 into the evening. There was no company news that morning. The setup underneath the squeeze had already been ugly. On August 6, Nasdaq staff issued a determination to delist the stock for failing the $50 million

market value requirement, and the company had disclosed a loan default that could lead to foreclosure on a material portion of its assets. This is a distressed sub $5 million microcap. The float and halt mechanics are the whole story.

The useful observation concerned the week that often follows a mover like this. When a crazy name starts the week, sympathy plays in related small caps tend to appear over the following sessions, and those follow through names are frequently easier to trade than the original halt magnet. Tracking and camera related micro caps were mentioned as the obvious candidates if the theme persists.

Nasdaq's planned move toward nearly 23 hour trading sat on top of that tape. The overnight session is expected to begin December 6, 2026, subject to SEC approval and readiness of the securities information processor. Trading would run from 9:00 p.m. ET Sunday through 8:00 p.m. ET Friday, with a daily one hour pause. Regular hours still run from 9:30 a.m. to 4:00 p.m. ET. Pre market already opens at 4:00 a.m. ET, and after hours currently runs until 8:00 p.m. ET. The new overnight session would leave only about a one hour break. Combined with the Pattern Day Trader rule, the structure points toward more overnight pumping in thin names, more gambling incentivization, and more of the same. The $25,000 minimum and the PDT designation were eliminated effective June 4, 2026, with broker phase in running through October 2027.

A related change is already live in options. Cboe's C1 exchange opened an extended morning session from 7:30 a.m. to 9:25 a.m. ET for a small group of high liquidity symbols, about 21 names at launch including Apple (AAPL), Nvidia (NVDA), Tesla (TSLA), Microsoft (MSFT), Amazon (AMZN), Advanced Micro Devices (AMD), and Meta Platforms (META). The session went live August 17 after being pushed back from July 13. There is also a 4:00 to 4:15 p.m. ET curb window, and only limit orders are accepted. Major index options already trade around the clock. Nobody in the room had been trading single name options in those extended windows yet. The invitation was simple: if you try it, report back. More hours means more chances to hit a lottery ticket or blow up an account.

In this part of the market, float, halts, overnight liquidity, and positioning explain the tape better than the press release.

2. Iran, Operation Economic Outcast, and a Currency That Prints Millionaires

The group covered Iran quickly because the pattern has become familiar: a lot of posturing, a headline designed to sound decisive, and energy markets that have already stopped treating every statement as a regime shift.

Treasury Secretary Bessent announced Operation Economic Outcast on August 24, describing it as an unprecedented campaign against Iran and its enablers after a Financial Times op-ed that promised an economic D Day. The policy expands secondary sanctions against entities doing business with Iran, a plan to pressure countries that refuse to sever economic ties, while stopping short of imposing new major penalties on Iran itself. Iran answered with public defiance. Pakistan continues trying to mediate. The UAE had announced the prior week that it was suspending all trade with Iran.

Crude was still carrying some residual Iran premium, well off the panic highs from earlier in the conflict. WTI fell about 2.5 percent Monday to $84.89. Brent lost 2.5 percent to $92.06. WTI in the $80s and Brent in the low $90s is the right picture.

The more striking data point was the currency. The Iranian rial hit a fresh all time low Monday, dropping to about 2.02 million per dollar after breaking 2 million on the open market Sunday for the first time. One dollar now makes you a millionaire in Iran. A Google chart that appeared to show a jump from 42,000 to 1.3 million created some confusion in the room until the two rates were separated. The 42,000 figure was the old official peg from 2018 into the early 2020s. The black market rate had already blown out to around 112,000 years ago, well ahead of the official number. Iran's Central Bank rate now sits around 1.5 million rial to the dollar, while the market rate is what most Iranians actually pay. The chart jump reflects that official rate being reset in stages. The rial first broke 1 million in the spring. The devaluation is real. It has been happening in layers, and the chart compression makes it look like a single overnight event.

That opened a question: is this the case for crypto in unstable countries? The room treated it as a dollar story. People facing a collapsing local currency rarely transact in Bitcoin. Bitcoin is too volatile for payroll, rent, and groceries. They reach for stablecoins, which are easier to obtain than physical dollars on a black market. Argentina already works this way. So do several countries in Africa. If that channel keeps growing, it is bullish for dollar demand even when it looks like a crypto story.

Oil, sanctions, and the rial are the same movie the group has been watching for months. Markets have learned to fade the theater until something changes the actual flow of barrels.

3. Bitcoin, Ethereum, and How Professionals Still Struggle to Place Crypto

Bitcoin had been moving, and a few people in the room had bought. Roberto's purchase last week came in around $77,000, more momentum than breakdown, after the coin had spent a long stretch in the $60s. Monday's open at about $77,728 was the highest level since May. The hoped for dip never arrived. That is a familiar feeling.

Andree's working framework was more specific. Tokenization and stablecoins look like the next wave of actual adoption. The trade expressing that view is long ether and short Bitcoin. The logic runs through market history and market structure. After 2021, the last cycle trained a generation of traders on altcoins, then Bitcoin outperformed and Bitcoin dominance rose. The bear market that followed turned a lot of people into Bitcoin maxis, which means the crowd is now underexposed to everything else, including Ethereum, the second largest network. The financial industry is slowly standing up on Ethereum and its layer twos. A layer two is a network built on top of Ethereum that processes transactions on its own faster, cheaper rails and then settles the results back to the main chain, inheriting Ethereum's security while cutting fees and multiplying throughput. That architecture matters for the thesis because activity on a layer two still pays tolls to the base network, so growth anywhere in the stack ultimately drives demand back to ether itself. Coinbase (COIN) has a layer two. Robinhood (HOOD) has a layer two. BlackRock's (BLK) largest crypto fund sits in Ethereum. The thesis is that value accrues to the native asset of the crypto economy with the most total value locked, and that Ethereum

outperforms Bitcoin over the next cycle. Sized as a spread, the trade can be closer to delta neutral. Sized directionally, it is a relative value bet.

The market caps make the valuation problem obvious. Ethereum is about $233 billion. Bitcoin is about $1.33 trillion, roughly five to six times larger. Bitcoin's bull case is first mover status and digital gold. Gold itself is in the $30 trillion neighborhood, while crypto as a whole is still around $2 trillion. Relative value arguments can look wide open from that altitude. They still have to survive 80 percent drawdowns. The last crypto drawdown took a lot of people out. Belief in a multi year market cap catch up is only useful if you can hold through the path.

John Hockberger asked the question asset managers actually have to answer: is crypto inside anyone's client allocation model? Victor's view split the industry into two channels. Newer managers have built product suites for the demand. On the institutional and retirement plan side there has been more movement, even if it remains difficult. On the intermediary side, the financial advisor channel, personal interest exists, but many shops have shuttered those sales divisions. Selling a product that is down 20 to 30 percent year to date is a miserable job. Traditional asset managers, outside some of the larger bank platforms, have mostly stayed out.

Even when a client wants it, the portfolio construction problem does not go away. If you are already 100 percent allocated, something has to be trimmed. The question becomes what crypto adds. Does it reduce volatility? Does it add alpha? If you cannot answer that, it is hard to put into a model portfolio. The evidence that Bitcoin is a reliable hedge is thin. Plenty of people would argue gold has failed that test as well.

Eric's practical stance was that crypto sits outside his circle of competence, so he does not pretend otherwise. For people who still want listed exposure, the cleaner paths discussed were the Bitcoin miners and data center names that either still hold coins or are pivoting into AI compute, plus the trusts and holding companies that wrap the coins in equity form.

IREN Limited (IREN) sells Bitcoin immediately and has been reducing mining, which helps explain why it has not been tracking the coin. Cipher Mining (CIFR) and Hut 8 (HUT) sit in the same neighborhood, with different mixes of mining versus AI data center ambitions. MARA Holdings (MARA) holds Bitcoin on the balance sheet and is one of the largest corporate holders, around $4 billion of coins at recent prices, typically ranked second among corporates behind Strategy (MSTR). That makes MARA a more direct beta vehicle, though the quality of its AI transition is a separate question.

Riot Platforms (RIOT) got a short lived bounce after a compute deal with Anthropic. Anthropic struck a $9.1 billion agreement for a 20 year, 191 megawatt supply from Riot's Rockdale, Texas campus, disclosed August 10 and 11. Shares initially soared more than 20 percent and then gave up almost the entire gain. Riot has also been selling monthly Bitcoin production and shrinking its treasury, with holdings declining from 15,680 to 11,380 BTC over the quarter, drifting toward the IREN model of selling coins and leaning into AI compute.

Grayscale Bitcoin Trust (GBTC) just moves with Bitcoin and remains the simplest way to buy exposure without a crypto brokerage. Strategy is closer to a levered Bitcoin ETF with a lot of decay. Tom Lee's vehicle, BitMine (BMNR), plays a similar game on the Ethereum side, with the same dilution problem. Those equity wrappers can give you more crypto per dollar of stock. They can also grind you down while you wait.

The group has now asked for a deeper crypto allocation session. The open question is whether Bitcoin earns a permanent seat in a portfolio the way a serious allocator would have to justify it.

4. Grok 4.6, SpaceX, and Why Ecosystem Still Beats a Benchmark

Grok 4.6 became the bridge from crypto into the Elon stack.

Few people in the room were using Grok on purpose. One mention was getting forced onto it in Cursor before switching models. The independent comparison that mattered sat on Artificial Analysis. Grok 4.6, released August 12, scores 61 on the Intelligence Index, in line with GPT 5.6 Sol, behind Claude Opus 5 at 63 and Claude Fable 5 at 62, and just ahead of Moonshot's Kimi K3. Meta's Muse Spark and several Chinese systems including GLM filled out the next tier. Gemini remained further back on the intelligence score while still looking like the fastest, a reminder of what TPUs are for.

Cost is where the chart gets interesting. Fable and Opus land around $2 to $3 per unit of intelligence. GPT 5.6 is just under $1. Grok 4.6 came in at $0.84 cost per task, roughly in line with Kimi. Performance versus cost is where xAI (now folded into SpaceX as SpaceXAI since the February merger) and Moonshot are making the most noise. Alibaba (BABA) remains a partial owner of Moonshot, which is why the group already has some exposure through the portfolio.

Benchmarks do not automatically produce adoption. Anthropic and OpenAI already have customer lock in and ecosystems. People will not mass migrate to Grok just because a score is close. What SpaceX does have is a stack that can absorb Grok: launch, Starlink, terrestrial data centers, planned orbital compute, and now the Grokbot workflow hype. Grok 4.7 was described as weeks away. Starlink Mobile is being positioned, at least in SpaceX's own claims, against AT&T (T) and Verizon (VZ), including aircraft connectivity.

The group recently added SpaceX (SPCX) to the portfolio. Eric has turned hyper bullish over recent weeks as the Grok developments have stacked up, a stronger stance than he held earlier in the position, and he plans to hold most of it. Valuation is anyone's guess. SpaceX went public June 12, 2026 at $135 in the largest IPO in history, raising $75 billion in the base offering and roughly $86 billion once the overallotment is included, with the market value around $1.85 trillion. The structural argument is larger than the next quarter. Launch, satellite internet, mobile, AI models, automation, ground data centers, and orbital data centers in one corporate system is as close as this generation may get to a Standard Oil style monopoly attempt. Love him or hate him, Elon plays business like a video game and goes for maximum damage. Asking Grok to roast ChatGPT produced the expected Elon flavor: ChatGPT as a corporate HR intern terrified of getting fired, reading from a script reviewed by 47 risk committees, agreeing with you when you are wrong and correcting you three paragraphs later. The comedy is the point. The positioning is the investment.

The bear case came from Quinn, offered less as his own view than as a counterweight to keep the discussion honest. It centered on supply and incentives now that the stock is public. The first lockup related window was followed by a 15 to 16 percent spike, which is the opposite of what Twitter had been preparing people for. The tape around that move was messier than a clean unlock pop. On August 6, SPCX rose 6.14 percent on 3.4 times normal volume. The next session gained another 15.8 percent, and the stock closed back above the $135 IPO price by

August 10. The day before the expiry, though, shares had sunk almost 14 percent to an all time low after earnings showed larger than expected AI capex, the company's second worst day on record. The August 7 rally also coincided with the announcement of a $16.8 billion joint AI semiconductor factory with Tesla. One analysis argues most of those unlocked shares had not yet reached anyone in a position to sell them. The market is still forward looking. A widely advertised event is usually an event that has already been argued into the price. The pop had more than one catalyst sitting underneath it.

Around the same time, Harvard Management Company disclosed about 12.94 million shares worth $2.2 billion, the single largest holding in its $4.26 billion reported U.S. stock portfolio. Harvard oversaw about $57 billion as of June 2025, so the position is still a fraction of the full endowment, but a 50 percent weight in one high growth name inside the public equity sleeve is not a weight most endowments keep forever. If Harvard is a signal for other pre IPO holders who now have a public mark and a need to rebalance, selling could become a short term drag. Cathie Wood loading up through ARK Invest cuts the other way, or maybe it is a contrary indicator, depending on how much weight you give the Cramer comparison.

Disclosed university endowment holdings were estimated around $3.5 billion directly, or $4 to $6 billion in a wider range. Against a $1.85 trillion company those blocks are real money and still small relative to the market cap.

Upcoming lockup dates on the table were September 10, September 25, October 10, and October 25, about 7 percent each, matching the published schedule of five smaller time based tranches at roughly day 70, 90, 105, 120, and 135 after the June 12 IPO. A day 70 tranche had already passed around August 20 to 21, just before this meeting. One of the remaining days will probably produce a down tape just because markets enjoy an excuse. Two longer dated caveats also landed. Quarterly disclosures will now show more about customers, governments, and funding sources. Political risk around Starlink contracts is harder to model. During the IPO process SpaceX's advisers lobbied the index providers to change their inclusion rules because SpaceX is SpaceX, and it mostly worked. Nasdaq adopted a fast entry rule letting top 40 market cap newcomers join the Nasdaq 100 within 15 trading days, which put SPCX in that index by July 7, and FTSE Russell rewrote its eligibility rules as well. The S&P 500 was the holdout. S&P Dow Jones rejected its own fast track proposal in June and kept its 12 month seasoning and GAAP profitability requirements, the same screen Tesla failed for a decade, so SPCX cannot join the benchmark until mid 2027 at the earliest. Size and influence can rewrite most of the checklist, though not all of it. Jeff Bezos has been lobbying the current White House that the government does not have to rely on Starlink, that Blue Origin could stand up an alternative. Contracts that deep are difficult to unwind, and they are not impossible to pressure after a political shift. The 2028 election was described as potentially the most market moving one in memory because of how much AI, data centers, China, and domestic politics are now tied together. Markets drop on uncertainty. Midterms this year are an earlier version of the same problem.

John Hockberger asked whether it makes sense to hold both SpaceX and Tesla. Victor owns both, in different accounts, and would rather own SpaceX. He just does not want to pay the capital gains on Tesla. Elon has denied acquisition rumors.

Tesla's robotaxi permits in Las Vegas are a real opening, and the numbers that circulated in the room had the approval backwards. On August 20, four days before this meeting, Nevada

regulators unanimously approved permits for Tesla, Waymo, and Uber to run commercial robotaxis in Clark County, authorizing up to 8,000 driverless vehicles over 12 months. Tesla drew the largest allocation, around 5,000. Tesla's Cybercab chief engineer said the company expects to field around 2,500 within the year and called 5,000 a ceiling. Waymo was cleared for up to 1,000. The Optimus story, factory retooling away from cars, and the idea that most Teslas could eventually be imported all remain live. Victor has cooled on Tesla's full self driving path because Waymo looks like a real competitor and the near term market may look more like a duopoly. Institutional access during the SpaceX IPO roadshow was treated as a constructive signal.

Tesla still has one lucky break the Detroit incumbents handed it: several traditional automakers scrapped EV programs. If cars become a service, that decision looks worse with time. Omar and John Donners put a longer clock on it, with John Donners speaking from his transportation consulting background. Roughly one in four or five U.S. jobs still ties into the auto industry in some way. Some municipalities are pouring money into EV infrastructure. Others are kicking the can. Illinois is investing more than most, and the speed of that buildout will shape how fast the rest of the auto industry actually feels it. Autonomous vehicles are on the horizon, and the rollout outside San Francisco, Phoenix, a few Chicago pilots, and a couple of East Coast programs remains slow. The public has tolerated Tesla Autopilot accidents more than people expected, partly because there is often still someone behind the wheel. A single sensational headline of a fully autonomous vehicle killing a pedestrian could freeze the politics anyway. A pandemic style shock could compress the timeline the way Covid compressed remote work. Absent that, the physical world still moves on permitting, public opinion, and local government time.

Cybercab's Austin reveal on September 3, 2026, was parked as a topic for the next meeting.

5. Moderna, mRNA Cancer Vaccines, and Healthcare as an AI Second Order Play

Moderna (MRNA) produced one of the longest stretches of the night. The room switched to Grok for most of it. Claude had been fighting with citation and site index errors during live lookups that evening and is more restrictive on biological questions.

Shares had exploded after positive Phase 3 top line results from INTerpath-001, announced August 19, for a personalized neoantigen mRNA vaccine given with Merck's (MRK) Keytruda. The trial enrolled 1,137 patients randomized 2 to 1 between the combination and Keytruda alone. It met the primary endpoint of recurrence free survival and a key secondary endpoint of distant metastasis free survival. It is the first positive Phase 3 result for any mRNA based cancer vaccine and the first for an individualized neoantigen therapy. Earlier Phase 2b data had shown a 49 percent reduction in the risk of recurrence or death, with five year follow up still in view.

Merck climbed more than 12 percent on a market cap around $333 billion, adding roughly $40 billion of value, comparable in dollars to Moderna's entire move and a huge single day for a mega cap, even if the percentage bounce looked modest. Moderna closed at $174.38, up 177 percent, with market cap rising from about $25 billion to $69 billion, then fell about 20 percent the next morning. By meeting day the name sat in the high $50 billions, around $56 to $58 billion, still up well over 100 percent from the pre news $25 billion.

The company history is the context for why the move matters. Moderna spent years looking like not much. Then Covid arrived and a mid cap upstart produced a vaccine alongside Johnson & Johnson (JNJ). The stock ran from the teens toward the $400 to $500 area, split adjusted, in about a year and a half. Financials never fully justified that peak. The stock then collapsed back toward pre Covid levels around $20. This melanoma readout is the first clean signal since then that the mRNA platform may work beyond infectious disease.

Set that beside Novo Nordisk (NVO), which the group recently dumped. Novo has GLP-1s, a low forward multiple, free cash flow, and a dividend, and it also has execution problems. That combination can be a value trap. Moderna is the opposite shape: cash burning, likely heading toward another offering, financially ugly, and now carrying a platform narrative the market is willing to capitalize.

A short primer on why the pipeline language matters. Phase 1 mostly answers whether the drug is safe enough to keep testing. It does not prove the treatment works. Phase 2 asks whether there is a real signal in a smaller sample. Phase 3 asks whether that signal holds up in a larger one. Once a program is in Phase 2 or 3, especially Phase 3, it has legs. The Merck partnership spans nine Phase 2 to 3 trials across tumor types, including non small cell lung cancer, renal cell carcinoma, and bladder cancer. Melanoma peak sales estimates in the $1.4 to $3 billion range do not justify a high $50 billions valuation by themselves. Non small cell lung cancer may be the larger prize, with peak sales talk in the mid to high single digit billions, or up to about $16.8 billion by 2035 in some of the more aggressive numbers. The current valuation assumes a meaningful probability that more of those trials work.

Gilead Sciences (GILD), now around $182 billion, was the historical rhyme: a speculative biotech that hit HIV and hepatitis C and became a cash cow. Moderna is already expensive relative to that early Gilead. The premium is the point. The stock probably does not go back to the pre news price unless the story itself breaks. If it does, that would be the value discussion. These names do not trade like other stocks. Pipeline, FDA process, and binary trial readouts dominate, which is why the Discord has a dedicated biotech screener.

CEO Stéphane Bancel has been in the seat since October 2011, shortly after the 2010 founding. He came in as an operational leader, was early, and has been there through every stage since. Founder led public companies have a long record of outperforming hired operators when the market shifts and the company has to pivot. Meta still exists as a social company because of Zuckerberg. Elon's companies rhyme with that pattern. Moderna sits in a different category from SpaceX, with some of the same "they actually have something" quality.

Al Pakrosnis's question cut to the investment problem. As models improve, more domain specialists will be able to connect dots they could not connect alone. The isolated Twitter stories the group covered earlier in the summer start looking like early signals. Paul Conyngham, a Sydney tech entrepreneur with a machine learning background and no biology training, spent a few thousand dollars sequencing his dog Rosie's tumor DNA, used ChatGPT and AlphaFold to identify the neoantigens driving her cancer, and worked with the RNA Institute at the University of New South Wales to manufacture a personalized mRNA vaccine, the same modality Moderna just took through Phase 3, that shrank her largest tumor by roughly 75 percent. Douglas Yao, a computational biologist, went viral claiming that an AI- assisted lab built in his home garage produced an Alzheimer's candidate, PAC-832, with models designing thousands of molecules and AI-programmed robotics running the screening. Neither

result is vetted. Rosie is a single veterinary case, given alongside immunotherapy and not called a cure even by Conyngham, and Yao's work remains self-reported without peer review. The signal is not any individual outcome. It is that solo operators with AI tools can now run discovery workflows that until recently required an institution, exactly the capability diffusion Leopold Aschenbrenner's Situational Awareness thesis pointed at when it flagged biotech as the place the inflection would show up. Moderna may be the public company version of that moment.

Direct comps discussed included BioNTech (BNTX), about $28 billion after its own 22 percent sympathy day, another mRNA name with a Covid history and a deep oncology pipeline; Arcturus Therapeutics (ARCT), a much smaller cystic fibrosis mRNA name around a $400 million market cap that has been running in sympathy; Intellia Therapeutics (NTLA) in CRISPR gene editing; and Recursion Pharmaceuticals (RXRX) on the AI drug discovery side. ARCT was floated as a possible short after a sharp move. Shorting these names into extended advances is a very specific kind of pain. They can keep going for longer than a clean chart says they should.

Victor's counterpoint was the large cap pharma model. Big companies buy the small ones. Johnson & Johnson, Abbott Laboratories (ABT), and Pfizer (PFE) all ticked up with Moderna. Efficient market theory would say an Alzheimer's cure over the next 10 to 20 years is already in terminal values. The fact that the whole group bid together is a reminder that those models are mostly vibes in the terminal year. If you do not want to pick the lotto ticket, sector vehicles start to look reasonable. The iShares U.S. Healthcare ETF (IYH) holds Eli Lilly (LLY) around 15 percent, then Johnson & Johnson, AbbVie (ABBV), Merck, UnitedHealth Group (UNH), and others. That is the slower, more value flavored way to own the same demographic and innovation tailwinds.

Healthcare has been a portfolio pivot for the group already, including BrightSpring Health Services (BTSG) on the end of life and specialty care side and Hims & Hers Health (HIMS) on peptides and consumer distribution. The AI bottleneck conversation usually goes to power, chips, and data centers. Healthcare may be the second order version: the industry that quietly absorbs the models.

Quinn added GE HealthCare (GEHC), a name that had not come up despite all the prompting. GE split into three companies. The healthcare piece is a dominant hospital hardware franchise in radiology and diagnostics, with a heavy acquisition record and, by Quinn's reading, the most AI patents among healthcare companies. It is med tech more than drug discovery, closer in spirit to Intuitive Surgical (ISRG) than to Moderna. Slow, installed base, AI going into machines that already sit in hospitals. That is often what a durable second order investment looks like.

Diana asked the question underneath all of this: whose AI are these companies actually using? The short version is that large biopharmas mix internal platforms with frontier labs. OpenAI shows up with Moderna, Novo, and Thermo Fisher Scientific (TMO). Anthropic shows up with Bristol Myers Squibb (BMY), Novo, Sanofi (SNY), and AstraZeneca (AZN). Google DeepMind, despite some talent departures, still sits with Eli Lilly, Novartis (NVS), and Johnson & Johnson through tools like AlphaFold. Everyone still sits on Nvidia underneath. Amazon came up as the boring, well positioned infrastructure answer across a lot of this work, and as a name the portfolio could still add. If LLM inference keeps getting cheaper, the models themselves start looking more like commodities, with oil boom parallels. The pick and shovel layer remains. Veeva Systems (VEEV), already in the group's orbit, sells the clinical trial and regulatory software a lot of these companies run on. The group timed that one well.

Health insurers were left as an open thought. Fewer people dying of cancer should, in theory, help the cost structure. Waiting on politicians to fix insurance looks like a losing bet. Innovation driving costs down would be the actual solution. Moderna is one data point, and it will not be the last.

6. Flock Cameras, Axon, and Owning the Surveillance State You Dislike

Flock cameras came up next, and the room did not pretend the topic is comfortable.

Flock Safety is the Atlanta based private company behind a large share of the automated license plate reader cameras going up across the United States. The stated use case is stolen vehicles, missing persons, and crime solving. The objection is that a tool built to catch criminals can become a surveillance layer, with the same flavor of system people used to associate with China. News stories keep landing on the worst version: stalking, exes, cops using the network for things that have nothing to do with a missing van.

One legal framing in the room was that Flock quietly shifts the constitutional default. Enough location data, and you start having to prove innocence after the camera has already built a story about where you were. Clearance rates, the statistic Flock likes, measure closed cases. They do not measure whether the person was guilty.

The company's own numbers, which should be treated as marketing, claim support for more than 1 million investigations, assistance in about 20 percent of cleared cases in deployed jurisdictions, more than 10,000 missing persons located, and involvement in a large share of stolen vehicle recoveries in some areas. A TCU analysis was cited as finding that adding one Flock camera per sworn officer lined up with about a 9.1 percent increase in quarterly clearance rates. Oakland reported an 11 percent improvement in violent crime clearance after deployment. Cleveland's figures were in the 15 to 18 percent range. Flock also claims involvement in roughly 10 percent of reported U.S. crime. Independent evidence is mixed, vehicle heavy, and early.

Quinn's distrust was more specific. The Instagram bio says Flock brought 10,000 plus missing people home. Confirming that those people were actually recovered, as opposed to a plate being scanned, is much harder. Flock also has multiple products, including pure plate readers and systems aimed at watching people. Activists cutting down plate readers while describing them as people trackers is part of the information war around the company. AI misreads on plates have already produced the wrong car being pulled over.

The hate cycle had already turned into product changes. On August 13, Flock announced it will cut default data retention to seven days from 30, and implement mandatory case codes and automated misuse checks, with CEO Garrett Langley saying "We got this one wrong," after more than 50 jurisdictions ended or paused agreements. There is also a documented pattern of cities cutting Flock cameras and then advancing Axon (AXON) automated license plate reader contracts, which fits Victor's long the watcher framing.

None of that makes the technology go away. Market share estimates, from Flock friendly sources, put Flock around 80 to 82 percent, Motorola Solutions (MSI) around 5 percent, and Axon with a smaller slice plus other lines of business. Flock's most recent round, disclosed in

April 2026, was about $500 million at roughly an $8.3 billion valuation, a modest step up from the $275 million round at $7.5 billion in March 2025 that Andreessen Horowitz led with Founders Fund participating, which is the Peter Thiel connection even if Flock does not work directly with Palantir Technologies (PLTR). Axon was an early investor and partner, then severed the relationship. They are now rivals. Axon sits around a $48 billion market cap.

Victor's framework was the most honest one in the room. He dislikes the surveillance state, owns Axon, has owned Palantir longer, and treats those positions as an emotional hedge. People vote against their own interests often enough that the cameras are probably arriving anyway. If you are going to be watched, you might as well be long the watcher. Palantir is already inside government in a way that does not unwind cleanly when administrations change. Flock is catching the current hate cycle. News cycles move. An Axon acquisition that retires the Flock name would fit the pattern of burying a scandal inside a larger brand.

Axon already has a monopoly on police body cameras. The overlap with Flock is small today. The shared end state is the dystopian one: drones as first responders, cameras as the dispatch layer, something out of Robocop. Historical industrial booms produced wealth gaps, protests, and a long messy middle before the gains were broadly felt. AI looks like that kind of boom. Security, defense, and surveillance names are one of the trades that benefit from the volatility even when the social outcome is ugly. That is also why 2028 keeps coming up. Growth from AI and unrest from AI can show up in the same decade.

7. Bally's, River West, and a Casino Company Getting Squeezed From Every Side

Bally's (BALY) was the local story of the night, and it is a reminder that not every investment story in this group has to be a technology compounder. Some of them are slow bleeds.

Bally's paused construction on most of the non gaming pieces of its $1.7 billion River West casino project, citing a gaming dispute with the City of Chicago over video gambling terminals. The 2026 budget lifted Chicago's ban on those terminals. Bally's argues that violates the 2022 deal. The casino floor is still pointed at early 2027. The hotel tower, event center, and restaurants are paused. For a mixed use project, pausing the non gaming amenities is pausing most of what makes the project a destination.

The dispute is a convenient story. The financials are the more convincing one. Bally's is a highly levered operator that has been in distress for a while. Market cap around $467 million against an enterprise value around $8 billion tells you the debt load. The second quarter 10-Q, filed August 14, a couple of days after the pause, included the phrase "substantial doubt" about its ability to continue as a going concern. That language is anchored to the Bally's Chicago entity, and the company says the disclosure is based on a forward looking technical accounting analysis that considers only unconditionally secured funding. Fair enough as a caveat. The cash problem is still the cash problem. Absent new financing, the company does not currently expect to satisfy liquidity and leverage requirements under its revolving credit facility. Bally's Chicago ran a loss of $56.4 million over the first half of 2026 and has an accumulated deficit of $233.3 million.

Mario's construction view was that cash flow is the real constraint. Building in Chicago with required union labor is expensive. Delay too long, and restarting becomes prohibitive. One

pitch from the Bally's side is to finish the casino floor first and push the hotel and restaurants later. That still leaves the cash problem unsolved. A buyer of the whole company is the cleaner endgame.

Physical casinos have a demand problem on top of the balance sheet problem. Sports betting, prediction markets, and phone based trading have turned gambling into something you can do from a barstool or a couch. DraftKings (DKNG) sits around a $13 billion market cap after a rough stretch that cut its value roughly 40 percent in a year. Add Flutter Entertainment (FLUT), owner of FanDuel, at around $17 billion, and the two big phone books together stack up against Las Vegas Sands (LVS) at around $32 billion. Kalshi and Polymarket are private and already in the tens of billions of valuation. DraftKings' in person kiosks at the Wrigley Field bar were shut down, likely because nobody used them when the phone was already in hand. That bar then closed entirely.

Video gambling terminals and slot machines in neighborhood bars, including Hawkeye's near UIC, remove another reason to travel to a casino floor. Quinn's line was simple: if you want sports, you bet on your phone. If you want slots, you walk to the bar. Bally's has an app in the sense that most casino companies have an app. DraftKings built an interface, a customer base, and a habit. Shipping a login screen does none of that by itself.

Ownership adds another sour note. Standard General L.P. owns about 65 percent. The company was sold out of Caesars Entertainment (CZR) related structures around 2020 through a New York VC path, with Twin River in the history. Noel Hayden, who built Gamesys, owns about 4.9 percent. Whether this is a version of stripping the useful assets and leaving the public vehicle with the debt is an open and ugly question. Chicago may have been sold a developer that cannot actually finish the job on the original timeline.

Comparables do not rescue the equity. The Fontainebleau in Las Vegas started in 2007 and did not open until 2023, stopping first for Lehman and then for Covid. These projects are cyclical and unforgiving if you build at the wrong time. Las Vegas itself has already pivoted toward high end celebrity weekend traffic. Foot traffic is down. High end spend has kept the top of the K shaped economy roughly whole. Water is the longer term constraint there. Atlantic City is the version of this story that already lost. Macau has been the brighter international tape, which is why Las Vegas Sands, despite the name, is now mostly a Macau and Singapore bet after selling its Vegas assets. MGM Resorts (MGM) is the more balanced mix of Vegas, Macau, and online.

The Chicago project still talks about roughly 500 hotel rooms and a 3,000 seat entertainment space. Even if the building eventually opens and does $800 million of revenue, the recoup timeline on this debt stack is measured in many years, and refinancing into higher rates eats the medium term. Someone like MGM could take it over for scraps. The equity, meanwhile, looks like a candidate for repeated dilutive offerings, reverse splits, and an AMC Entertainment (AMC) style long grind lower.

Al Pakrosnis asked for the bull case. The room did not really have one. That is why Bally's came up as a slow bleed short candidate for people who want an extended hedge, with all of the usual shorting caveats attached. The casino in River West can still get built. Bally's as a stock does not have to be the thing that wins from that building.

8. Lightning Round

AEVEX (AVEX). Quinn opened with AEVEX, ticker AVEX on the NYSE, a recently public drone and autonomous warfare company. It raised $320 million selling 16 million shares at $20, valuing the group at $2.2 billion, and began trading April 17. The stock has traded well above the $20 offer price since. The company has been supplying drones to Ukraine since 2022, with more than 9,300 systems delivered or committed across Phoenix Ghost and EUCOM Deep Strike, and more than 10,200 expected by the end of 2026. Contracts with the U.S. government were 78 percent of 2025 revenue of $432.9 million, most of it from Ukraine, which is both concentration risk and a demand signal. They are at capacity and trying to scale production. The business includes a software layer, U.S. Air Force and Army unmanned mission work, NATO and Indo Pacific exposure, and a wildfire intelligence product, alongside the airframes. Defense spending, Ukraine, Taiwan contingencies, and the same security tailwinds discussed earlier all sit behind the name. Speculative, and closer to investable than most of the drone ideas the group has kicked around, because so few pure public vehicles exist.

Unitree. Unitree's Shanghai debut came up next. The Chinese humanoid robotics company closed its first day up about 460 percent after raising $904 million, becoming the first listed humanoid robot maker in mainland China, at a market value of around $50 billion, with DeepSeek among the investors. Revenue rose to 1.70 billion yuan in 2025, about $235 million, with a net profit of 278.21 million yuan, and the company shipped more than 5,000 humanoid units. Real sales and a real profit are what separate Unitree from the typical robotics listing. The valuation is still wild relative to those fundamentals. STAR Market pricing rules also produce structural underpricing: CXMT surged 466 percent on its debut on the same exchange last month, with only 10 percent of shares floated, so the first day pop says as much about the listing mechanics as it does about the robots. China's hardware stack, especially around Shenzhen, is the industrial argument. Robots assembling cars, humanoids running, and Waymos already flooding Santa Monica are the same movie at different frames. You either believe the story enough to sit through an 80 percent drawdown, or you do not. Interactive Brokers is the practical path for people who want the shares.

Albemarle (ALB). Michelle raised Albemarle, a top lithium producer around a $16 billion market cap with beaten down value metrics and vertical integration across lithium, bromine, and energy storage. About half of net sales tie to new energy end markets, with operations in Australia, North Carolina, Chile, and Nevada. Morningstar's $200 target is the usual analyst optimism and still useful as a reminder that the market has already punished the name. This is a pick and shovel on batteries, EVs, and storage, one layer below which robotaxi brand wins. It is the kind of mid cap, off the beaten path name the portfolio process is built to consider.

Bachem (BCHMY) and WuXi TIDES. Al Pakrosnis brought two non U.S. peptide manufacturers. Nobody really makes finished peptides at scale in the United States. If the group's peptide thesis is about selling pickaxes into a gold rush, the manufacturers sit closer to the source than Hims, which is a distribution and consumer brand bet. Bachem is Swiss, trades on OTC as BCHMY, and already supplies a lot of the Western research chemical channel through looser Swiss rules and third party testing. WuXi TIDES is the Chinese analog, also available in a thin OTC listing, with a much larger implied market cap and much worse liquidity (average volume around 2,000 shares). The Swiss company earns better margins on Western contracts. The Chinese product is cheaper and has a harder time clearing regulatory doors.

The point of going downstream is regulatory agnosticism. There is a doctor prescribed channel and a gray market channel. Which one explodes is a coin toss that depends on FDA process. If you think peptides become a much bigger thing and you do not want to bet on that coin toss, you buy the ingredients. API manufacturing is the unglamorous layer that keeps showing up in the prompts. Novo will play more by FDA rules. These names are how you own the input either way.

Closing Thoughts

The August 24 meeting was a tour of second order effects.

CID HoldCo was a lesson in halt mechanics, overnight liquidity, and why a crazy Monday often seeds easier sympathy trades later in the week. A distressed sub $5 million name with a delisting cloud and a loan default does not need a press release to run 300 percent. Nasdaq adding nearly 23 hour trading and Cboe opening a pre market options window for mega caps is the same lesson written into market structure. More hours, more leverage, more opportunities to get rich or get wrecked in names that do not need news.

Iran remains a headline machine that oil has partly learned to fade. WTI in the $80s, Brent in the low $90s, and a rial that makes millionaires out of $1 bills is the human detail. The investment detail is that collapsing currencies pull people into dollar stablecoins more than into Bitcoin.

Crypto still does not sit cleanly in an asset allocation model. That is exactly why the listed wrappers, the miners that hold coins, the miners that sell coins and pivot to AI, and the ETH versus BTC relative value trade are more useful than a generic "everyone should own some." Volatility is the product. Position sizing is the job.

Grok 4.6 closing the benchmark gap at 84 cents a task is less important than the stack around it. SpaceX is the bet that ecosystem, distribution, launch, spectrum, and compute can turn a close enough model into a monopoly attempt. Lockups, endowment rebalancing, a $16.8 billion chip plant with Tesla, Cathie Wood, and Elon's tweets will all create noise. Quarterly filings and politics will create actual information. Tesla's Vegas robotaxi permit is a 5,000 vehicle ceiling with a 2,500 near term plan, and Cybercab's Austin night is September 3.

Moderna is the first public biotech in this cycle that looks like the AI drug discovery thesis leaving Twitter and hitting Phase 3. The financials are still a burn. The pipeline is why the market recapitalized it in a day, and Merck's 12 percent move shows the dollar value can show up in the giant as well as the specialist. Healthcare, from GE HealthCare's installed imaging base to Veeva's trial software to IYH's large cap mix, may be one of the better ways to own AI's consequences without having to win the infrastructure lottery.

Flock and Axon force a different kind of honesty. You can hate the destination and still recognize the cash flows. More than 50 jurisdictions paused or ended Flock deals, Langley said they got it wrong, and cities that pull Flock cameras have a habit of shopping Axon next. Industrial booms produce unrest. Unrest produces cameras, drones, and software. Palantir already showed how deep those systems can sit inside the state.

Bally's is the anti SpaceX. Local politics, union construction, video gambling at the corner bar, DraftKings on the phone, a going concern warning on the Chicago entity, and a $467 million equity stub under $8 billion of enterprise value. Chicago may still get a casino. Shareholders of this vehicle should not confuse the building with the stock.

The lightning round kept returning to the same habit this group has been building all year. If the story is drones, look at who is already selling into Ukraine and the Pentagon, which is how AEVEX showed up. If the story is robots, Unitree has real 2025 revenue and a profit, and a STAR Market pop that still requires an 80 percent drawdown stomach. If the story is EVs and storage, lithium producers with real assets still exist. If the story is peptides, the ingredient makers in Switzerland and China may matter more than the consumer app on this side of the ocean.

We look forward to reconvening on Monday, September 14, at the Union League Club of Chicago. Tesla's Cybercab moment, Chinese humanoids, and whatever the next small cap circus produces will still be waiting. For anyone who missed the night or wants to revisit a segment, the full recording of the August 24 meeting is available on YouTube (https://www.youtube.com/watch?v=0dQGG8Jh91c).