Left wing media has convinced Democrat voters our country is awful. As a result the percentage who are proud to be American has collapsed to record lows over the past 25 years. A country that isn’t proud of its history won’t long endure.

“Michelle Obama is a man” shouted on the White House lawn in a ring sponsored by Bud Light only available on Larry Ellison’s Paramount Plus. What a way to celebrate America 250 and the twilight of liberal democracy.

The "I f***ing love Science" crowd turning on the man who popularized the electric car in America and out-rocket-launches NASA by orders of magnitude because of politics is a funny twist. Maybe they don't really f***ing love science that much

BREAKING: Elon Musk has become the first trillionaire

The S&P 500, MAG7 and NASDAQ Fell sharply on Tuesday and it got me thinking about SpaceX, OpenAI and Anthropic's IPOs - On Tuesday the US market had a scary puke. In three hours the S&P 500 erased $2.1 trillion in market cap per @KobeissiLetter. The markets rallied toward close to reduce some of these losses - This comes after Friday’s selloff when the S&P 500 erased roughly $1.8T, Nasdaq had its biggest point drop ever, and chip stocks lost over $1T. - That is what got me thinking about SpaceX / OpenAI / Anthropic IPOs. Everyone assumes these things go straight up because they are the only public ways to own AGI but it's curious given the market's weakness/volatility recently. - These IPOs are enormous. If we assume ~$1.75T for SpaceX, ~$1T for OpenAI, and ~$1T for Anthropic, that is ~$3.75T of new public market cap. Roughly 17% of the Mag 7 and 5.6% of the S&P 500. - The float setup probably supports these stocks doing well at first. SpaceX is offering 555.6M shares against roughly 13.1B Class A B shares, so only about 4.25% economic float, or around 4.9% with the greenshoe. - That is tiny versus prior IPO crazes. Google was roughly 7%, Arm roughly 10%, Snowflake roughly 10% to 12%, Facebook closer to 20%. OpenAI and Anthropic have not disclosed terms yet, so I would not claim exact floats, but if they raise $50B to $75B around ~$1T valuations, that is basically 5% to 7.5% float. - These floats are historically small so it's easy to move the prices. Think low float/high FDV (but solid businesses). - Given how low the floats are as a percentage and the names, I think it's fair to assume these go up and to the right to start and do well out of the gate as long as the economy doesn’t fully turn over. - My question is what happens next - My main concern is that I think the highest conviction buyers hyper entered these names privately. I've spent years watching folks chase 2nd/3rd/4th order SPVs to enter these companies so the most ambitious buyers have already bought - This is hard to quantify, but the sentiment is obvious: OpenAI and Anthropic have both warned investors about unauthorized equity exposure / SPVs, and Anthropic secondary demand reportedly got so insane that one platform saw 50 institutional inquiries and claimed aggregate demand above $1T in three months. There was/is clearly an absurd market to get into these deals historically - So the real question is: after the first scarcity squeeze, who is left to buy when the float expands? - Retail and institutions who never had a clever broker, a fiance who is chief of staff at anthropic or secondary platform access can finally enter, and that demand against a ~5% float is what drives the initial pop. This is obviously a massive market (index funds, retail) and that's why I think there is an initial run higher. - As a retail trader you’re not going to not own AGI stocks out of the gate. Imagine admitting that to your friends who will judge and ban you from the group chat. - But the same low float math cuts the other way. If only ~5% trades at IPO, roughly 95% unlocks later, mostly on a standard 180 day calendar. On my numbers above that is potentially $3T of paper hitting eligibility in 2027 (assuming no change in valuations really). The lockup window is the concern and what happens as this gets priced in ahead of time. - Remember funds who invested years ago want to produce DPI to crystallize carry, clout, host victory lap podcasts and raise new funds. So I would not expect major old investors to sit on TVPI when they can move on and lock in DPI. - I asked AI to estimate investor only ownership pre IPO: OpenAI ~55%, Anthropic 70-75% (estimate until the S-1 prints), SpaceX ~46%, on roughly $180B, $132B and $11.9B raised respectively. - Now the question is what is the propensity for holders to sell? The percentage of each cap table sitting on at least a 10x return: ~65% of Anthropic (~$650B of paper gains), ~41% of OpenAI excluding the nonprofit Foundation (~$400B), and 75-80% of SpaceX (~$1.4T, the biggest of the three since the oldest money bought the cheapest). I thank my AI companions for these estimates. - But not all 10x paper is in the sellable category. Of the roughly $2.5T across the three names, about a third, call it $900B, sits with funds and employees who are more likely to sell. The rest is founders, strategics and foundations (Musk alone is ~$735B of the SpaceX figure, Microsoft is most of OpenAI's) who mostly won't sell at unlock. - And that $900B is not equally trapped. SpaceX has been letting funds and employees out through tenders since 2022 and OpenAI has run three, including one at a $500B valuation. Anthropic has never run one. So the largest pool of fund and employee paper that has never had an exit sits at Anthropic, and its first exit ever is the IPO. - Are these folks auto-selling? No. But a percentage of this $900B will and I think it's worthwhile to paint a picture to offset those who are only looking at passive indexes and new inflows. There are offsets. - And by the time those unlocks hit, public investors will have quarterly numbers, gross margins, capex, dilution schedules, payback periods and guidance to model for the first time. The mystique of no financial information I just want access premium that powered the SPV frenzy may not hold up as well vs real analysis. - Investors will face normal questions like do returns satisfy more investment since this entire flywheel currently hinges on more investment dollars to fund buildouts, models and to subsidize inference costs - I’ve covered my concerns on costs here: x.com/Shaughnessy119/status/… - As a recap to my above tweet I think enterprises are hitting a wall on spend and realizing they can switch to open source alternatives for 1/10-1/100 the price (or 10-100x more usage for same money) since these open source models have gotten really good, aggregators make it easy and they are offering increasing privacy methods which helps to marginally neutralizes the privacy and data retention edge the web2 AI labs used to claim. - Also new information on my cost thesis is that Apple announced a new version of Siri which is less impactful for enterprise/API AI spend since its consumer but it nonetheless impacts since the AI is free and has unique context (screen, iMessage, photos) the AI labs don’t have. This can lead to less consumer signups. - If you argue we will get these huge revenue drivers sooner (new meds, technologies, unstoppable API revenue which flow to OpenAI/Anthropic’s upside), this entire thesis is wrong because that drives stock prices and funding higher and keeps the flywheel going. If you argue it may take longer then these companies have to rationalize returns to new classes of investors and that timing mismatch (returns vs funding) is where these stocks can really cool off, and may be a great buying opportunity. We’ll see. - To linger here, I think AGI can offer the world of the future that will be worth obscene amounts of money. I am not arguing that won't happen. I am arguing that it could take more time than we expect and in the meantime the AI labs are dealing with large substitution issues given cheap open source AI models are 1/10-1/100 the price and that impacts margins near term which impacts their ability to raise capital and could cause near term price disruptions. - This worse case scenario can happen faster if enterprise customers cap spend, route to cheaper models. - So my net view is that these IPOs all likely do well at first. The area of concern for me is a few things but boils down to the timing between continued investment pre AGI during a time of intense competition and substitution where users can switch to open source AI models. - I have less specific views on SpaceX given I spend more time on the AI side so OpenAI/Anthropic are the main areas of my discussion. SpaceX is included since its one of the three mega IPOs this summer. - Given I’m in both worlds, I think $BTC is actually a beneficiary of the madness; it's just impossible to time. If AI trade slows or has a near/medium term hiccup I think money flows to it as a different asset. If the AI trade doesn’t slow down I think you have an insane amount of wealth creation later on (think Anthropic employees chasing $20m SF houses) and UBI will eventually cause max money printing which benefits it. Near term if these IPOs are forever up and to the right they will suck money out of the market and no asset is safe. - Crypto usually front runs globally liquidity/economic issues and I feel it clearly has given the sell off and flows to AI. - I'm bullish on both AI and Crypto so I don't want to be painted as either or I just think given I am in both worlds I can offer a new viewpoint. - Net I think the IPOs pop on small floats and sell off hard. If we get insane revenue to OpenAI/Anthropic sooner (new drugs, technologies, AGI) then this is all up and to the right. If we don’t and investors start questioning the numbers in an era of easy open source substitution I think they will face pressure.

Our LLM Token Expenditure Index should really have been named the “Token Expenditure Price Index” bc it’s an expenditure or usage-weighted average token price index. It tells you how much currently the entire market AI is paying for a million LLM tokens irrespective of models. The naming might’ve led to some misinterpretations as some seem to have interpreted the index as either the total volume of token used or the average price of tokens. In reality, the index captures something more subtle than either interpretation: it tells us the marginal willingness to pay for LLM models. Over the course of the year, while model token prices haven’t moved that much, the usage patterns have moved dramatically leading to the token index movement down and then up sharply as AI users moved en masse into using cheap open weight models and then en masse to the much more expensive frontier closed source models. From consumers to enterprises, everyone is Claude-maxxing! More recently, as can be seen in the chart below, the token index has stagnated, which suggests that usage migration towards frontier models has slowed. Time will tell whether this is just a pause or an inflection in the trend as users move back towards open weights models. In a sense our token index could be roughly interpreted as a “quality premium” of frontier models over the much cheaper open source models (if we assume users and prices are both “rational”). For more details on what we offer beyond the few indices we’ve listed on the Bloomberg Terminal, check us out at silicondata.com and give us a holler! 😊

You know Drs are dumb right?