Before buying the Anthropic IPO, learn the valuation lessons from SpaceX, Peter Lynch, and the rapidly changing AI landscape.
KEY TAKEAWAYS
Peter Lynch's famous advice to "buy what you know" begins with product familiarity, but it never ends there. Knowing a company should lead investors toward research–not directly into buying shares.
SpaceX's recent IPO demonstrates that excellent businesses can still produce poor investment outcomes when purchased at excessive valuations. Company quality and investment returns are not always the same thing.
AI competition is accelerating rapidly as open-weight models narrow the performance gap with proprietary systems. Consumers benefit from this trend, while investors must reconsider long-term profit assumptions.
AI infrastructure increasingly resembles a capital-intensive utility business where competitive advantage depends more on financing and scale than proprietary technology. That changes how investors should think about future returns.
Anthropic appears to be one of the strongest companies in artificial intelligence, but its valuation already reflects extraordinary expectations. Successful investing requires evaluating both the business and the price being paid.
MY HOT TAKES
The next phase of AI investing will be driven less by technological excitement and more by valuation discipline. Exceptional businesses can still disappoint investors if expectations become excessive.
AI's competitive moat is shrinking faster than many investors appreciate. The industry remains attractive, but the economics are evolving as intelligence becomes increasingly abundant.
Infrastructure spending across AI is beginning to resemble previous capital-intensive investment cycles. Scale alone does not guarantee superior long-term shareholder returns.
Retail investors continue to confuse product adoption with investment opportunity. The hardest question is rarely whether the company is good–it is whether the stock already reflects that reality.
The biggest winners in AI may not be determined solely by innovation, but by capital allocation, pricing power, and sustainable free cash flow. Those factors ultimately matter more than excitement.
You can quote me: "You don't make money by knowing the company–you make money by knowing the price."
What do you know? I am not sure if it was before or after my mother-in-law explained to me how she was able to find a two-for-one special on Klondike ice cream desserts (that’s a real sale) or her proposal for fixing Medicare completely, that she informed me about how Claude was going public. I responded, “Claude? As in Anthropic, the company that makes Claude?” I am pretty sure that I knew the answer–of course she was referring to Anthropic.
My Mother-in-law is highly informed and mostly accurate, as she is still capable of formulating her own opinions, which seems a lost art in most of the investment community these days. Well, there it was–she soon asked me whether she should "get in on the Claude IPO." She hears about Claude constantly and has been asking her grandkids to show her how to use “him”. And that, right there, is the beginning of the problem. In fact, the problem goes well beyond our spirited coffee talk–most people use Claude or one of its competitors daily!
Legendary stock guru Peter Lynch told a generation of investors to buy what they know, and Wall Street has been mangling that advice ever since. Somewhere along the line, "know the company" became "love the product," and "do the homework" became "get in line," became “don’t miss out–FOMO!” Five weeks ago, a record number of investors got in line for SpaceX. Let's sidestep for a moment and talk about how that's going.
SpaceX priced its IPO at $135 a share on Thursday night, June 11th, and by the time SPCX opened for trading the next morning at $150, the frenzy was already historic. Citadel Securities, the largest retail wholesaler on the street, reported the highest IPO auction order activity ever recorded among individual investors. Net retail buying ran at more than three and a half times the level of NVIDIA–yes NVIDIA!–which held second place. The offering itself ran four times oversubscribed, and more than 500 million shares changed hands on day one, a figure that approached Facebook's storied 2012 debut (storied for so many reasons
Which brings us back to AI, where the ground is shifting even faster than it is in aerospace. Last Friday, Beijing-based Moonshot AI unveiled Kimi K3, which it bills as the largest open-weight model ever released–2.8 trillion parameters–with the full weights scheduled to be handed out, free of charge, before the end of this month. By Moonshot's own admission, K3 still trails the best proprietary models from Anthropic and OpenAI. But that admission is doing an awful lot of work in the sentence. The gap between the frontier and the fast followers is now measured in months, not years, and the fast followers are giving their work away for NOTHING. Semiconductor stocks sold off on the news for the same reason they sold off when DeepSeek pulled a similar stunt: if frontier-adjacent intelligence becomes abundant and cheap, the returns on the hundreds of billions of dollars being poured into AI infrastructure become considerably harder to defend. To be really clear (and don’t miss this), viewed through the consumer's lens, this is genuinely wonderful news. Competition is doing exactly what free markets promise it will do–better products, at lower prices, delivered faster. My mother-in-law's grandkids will spend their careers with smarter, cheaper AI, and that is a win for everyone. An equity investor, however, has to look through a different lens and wrestle with a harder question: what does a moat that thins this quickly do to the terminal value baked into a near-trillion-dollar valuation? To be clear, the model providers are not yet a commodity–the frontier still commands a premium, and the leaders have built real enterprise franchises. The direction of travel, however, is unmistakable, and markets price direction long before destinations are reached.
Go one layer down the stack and the commoditization is already largely complete. The hyperscale data centers that power the frontier models are differentiated by exactly one thing: how much capital their owners can shovel into the ground. It is not proprietary technology–everyone is buying the same chips, the same cooling, and the same turbines, whenever they can get their hands on them. When a moat is measured in gigawatts and access to cheap financing rather than in intellectual property, that is not a technology business; that is a utility wearing a growth multiple. And utilities, as my fixed income roots compel me to point out, earn returns that gravitate toward their cost of capital over time. Here is a shadow data point that ties this whole story together with a bow. According to SpaceX's own prospectus, Anthropic has agreed to pay SpaceX $1.25 billion per month–per month–through May of 2029 for compute capacity at the Colossus 1 data center in Memphis. Yes, you read that correctly: the cautionary case study in this morning's blogpost is the compute landlord of the company about to go public. The marginal dollar of AI infrastructure is increasingly funded by debt, sovereign wealth funds, and circular deals among the players themselves rather than by boring old free cash flow. That is not automatically a crisis. It is, however, absolutely a question, and questions are precisely what disciplined investors are supposed to bring to an S-1 (that’s the IPO document filed with the SEC).
So, what do we actually know about Anthropic, the company my mother-in-law wants a piece of? We know it filed a confidential draft S-1 with the SEC on June 1st, hot on the heels of a $65 billion funding round that valued the company at $965 billion, with a listing expected as early as October–potentially the first AI company to debut with a trillion-dollar handle. We know that its growth is genuinely breathtaking: annualized revenue leapt from roughly $9 billion at the end of last year to more than $44 billion by May. I have spent nearly four decades on Wall Street, and I have never witnessed a revenue ramp like that one. It deserves respect–full stop–and it explains why the enthusiasm is not irrational on its face. We also know that the company lost $5.6 billion in 2024, spends around $19 billion a year on compute, and does not project positive free cash flow until 2028. And we know that OpenAI filed its own confidential prospectus just days later, which means the two biggest names in the industry will be courting the same investor dollars within months of each other (OpenAI reportedly delayed its public debut until possibly early 2027), while Moonshot hands out frontier-adjacent models like Halloween candy. That is a lot of supply–of paper and of intelligence–arriving at the market in a tight time window.
None of that says the IPO should be avoided, so let me be direct: this is not a bearish blogpost about Anthropic. A case can be made that it is among the best-positioned companies in the entire ecosystem, particularly in the enterprise, where the real money lives. This is a piece about discipline. The questions worth asking before getting in line are the very ones the SpaceX crowd skipped on their way to $225. How much growth is already embedded in a $965 billion valuation? What margin structure survives a world of thinning moats and commoditized infrastructure? And what price makes this remarkable company a remarkable investment–and is the IPO print anywhere near it?
Loving the product and loving the price are two different loves, and only one of them pays. My mother-in-law, to her enormous credit, asked before she bought. That alone puts her ahead of the crowd that chased SPCX north of $200–and it puts her exactly where Peter Lynch wanted her: at the starting line of the homework, not the end of it. When I hesitated to answer her question about the IPO, she said simply: “let’s ask Claude,” so I did, and Claude (Fable 5) to his credit answered simply: “You know the company–but do you know the price?” That was a wise answer.
FRIDAY’S MARKETS
Stocks closed lower on Friday, with the S&P 500 falling 1.01%, the Dow Jones Industrial Average losing 406 points, or 0.77%, to 52,146, and the Nasdaq Composite dropping 1.40%, as semiconductor shares sold off following Moonshot AI's Kimi K3 release and the PHLX Semiconductor Index entered bear-market territory. For the week, the S&P 500 lost 1.6%, the Nasdaq fell 2.9%, and the Dow declined 0.9%. The 10-year Treasury yield eased to 4.54%. WTI crude jumped roughly 4.5% to settle near $82 a barrel amid renewed U.S.-Iran hostilities.
NEXT UP
Conference Board’s Leading Economic Index (Jun) may show a decline of -0.1% after an increase of 0.1% in May.
This week is light on the economic releases but still brings us Flash PMIs, housing numbers, and Durable Goods Orders. The earnings parade picks up momentum this week with the first of the Mag-7 megacaps stepping up to the mic, amongst many others. You better check back often if you want to know what’s up and where to get the best deals on your favorite desserts, compliments of my mother-in-law.