Who are the Category Kings of AI Going To Be? | The Pirate Street Journal

Who are the Category Kings of AI Going To Be? | The Pirate Street Journal

This is a free preview of a paid episode. To hear more, visit www.categorypirates.news [https://www.categorypirates.news?utm_medium=podcast&utm_campaign=CTA_7] On June 12, SpaceX is going public at $135 per share and a $1.75 trillion valuation. 4% of total shares are being offered. Of those, 30% have been allocated for retail investors to buy directly at the $135 IPO price. The standard retail allocation in a mega-cap IPO is 5 to 10% of shares, with the overwhelming majority reserved for large institutional buyers like pension funds and mutual fund managers. SpaceX is tripling that, which is unusual. At a $75 billion raise, that is roughly $22.5 billion in shares flowing directly to retail. More than many entire IPOs. One of SpaceX’s lead underwriters told Reuters they had never seen anything like the expected retail demand. Both Anthropic and OpenAI [https://www.wsj.com/tech/ai/openai-kicks-off-ipo-process-in-test-of-investor-appetite-for-top-ai-labs-eb7bebe1?mod=djemalertNEWS] have also filed to go public sometime this fall. It is a mad rush to raise capital to fund the AI infrastructure buildout. Valuations defy gravity and become a moving target as ARRs change every month and cash-burning businesses like xAI flip to cash-flowing with the stroke of a single deal. IPOs Are Dangerous, Right? Truist analyst Keith Lerner pulled the data on 30 major IPOs from the last 15 years. The names you have heard of. Facebook. Uber. Palantir. Snowflake. CrowdStrike. Even the biggest winners experienced massive take-downs in their first year. The median Year 1 max drawdown across all 30 IPOs was 54%. The average was 55%. Robinhood dropped 90% from its peak. Rivian dropped 88%. Lyft dropped 79%. Uber dropped 68%. That is just the drawdown. Jay Ritter at the University of Florida has tracked every U.S. IPO since 1980. Over three years, the average IPO underperforms the market by 23.4%. It is a feeding frenzy for the sharpest sharks on Wall Street. IPOs get bid up by retail enthusiasm, then shorted on the way down. The average Joe and Jane gets whacked. So if you are sitting at home watching the SpaceX IPO ads roll across your screen, the safest move is to: remember the data. Except. It turns out most of America will eventually invest in SpaceX (and likely Anthropic and OpenAI) whether they realize it or not. 62% of Americans own stock as of 2024. The vast majority of that ownership is indirect, sitting inside 401(k)s, IRAs, S&P 500 index funds, and total market funds. The S&P 500 alone is tracked by tens of trillions of dollars in passive money. To get into the S&P 500, a company is supposed to make money. The sum of its four quarters of earnings has to be positive on a GAAP basis, and so does its most recent quarter. That rule is decades old. It is the reason Tesla sat outside of the index until the end of 2020, years after it had become one of the most valuable companies on earth. That rule is about to be broken on purpose for some of the indices. The other major indexes have already moved. Nasdaq’s Fast Entry rule, effective May 1, 2026, cut the Nasdaq-100 waiting window from roughly three months to 15 trading days. CRSP, the index behind Vanguard’s funds, introduced an alternative path that could place SpaceX in the Russell 1000 within five trading days of the IPO. The S&P 500 refused to change the rules. So is this bad? Not necessarily. There will be many Category Kings of the AI era. SpaceX is currently one of them. So who will be the Category Kings of AI? And how does the AI game board look through the category lens? Let’s look at who the Category Kings were across the value stack in the PC era, the Dot-com era, and the Social/Mobile era. The pattern tells us what to look for now. Remember, we’re not financial advisors, and this is not personal financial advice. We’re presenting you with data through the category lens to give you a different POV on how categories rise and fall. And in this case, how the mega tech stack categories evolve over time. The Pirates 6 Layer Rum Cake Jensen Huang sells the AI economy as a five-layer cake. Energy at the bottom. Chips next. Cloud infrastructure above that. Models on top of cloud. Applications at the very top. Stack them. Slice them. Invest in them. It is a useful frame. It is also incomplete. The Pirates expanded the cake to six layers and simplified the language. Power instead of Energy. Internal Hardware instead of Chips. Infrastructure stays. Operating System instead of Models, because the OS is what every era’s category designer has always called the layer that makes everything else work. We added a layer Jensen left out. End-User Hardware. The device the customer touches. The PC, the phone, the device, the App Store toll booth. Applications stays at the top. Six layers. Power, Internal Hardware, Infrastructure, Operating System, End-User Hardware, Applications. The reason Jensen’s cake matters less than the Pirates Cake is that Jensen’s cake describes one era. The Pirates Cake describes every era. PC. Dot-com. Social/Mobile. AI. Same six layers, different category leaders. Before digging in, three rules guide the analysis. Rule 1. Pay close attention to multi-era category kings. Microsoft won the PC era. Microsoft won the Dot-com era. Microsoft is in the top three of the AI era opening act, four decades after going public. A $10,000 investment in Microsoft at its 1986 IPO, held through every crash and every doubt, was worth approximately $80 million by December 31, 2025. Roughly 8,000x. A 26% compound annual return for forty years. A $10,000 investment in Apple at its 1980 IPO was worth approximately $25 million by the same date. Roughly 2,500x. A 19% compound annual return for forty-five years. A $10,000 investment in Apple the day the iPhone launched in June 2007 was worth approximately $570,000 by December 31, 2025. Roughly 57x in eighteen years. What’s our point? There will be multiple Category Kings. You can still do great, even if you miss it early. Multi-era winners compound through platform shifts, recessions, market crashes, leadership changes, and competitor onslaughts. Rule 2. Hardware wins first. Then software. Every era opens with a hardware-led leader. IBM in PC. Intel in Dot-com. Microsoft in Social/Mobile, which is the exception that proves the rule because Microsoft was the prior era’s vertical integrator extending its run. Now Nvidia in AI. The opening years of every era belong to whoever ships the picks and shovels. Always. The closing years belong to whoever owns the operating system and the layers closest to the customer. Rule 3. Multi-era winners own the most valuable areas of the stack. Not the most layers. The most valuable ones. Operating Systems are always one of them. End-User Hardware or Distribution into the customer is usually another. Applications on top are the third. The App Store alone takes 30% of every transaction every developer makes on the platform forever. That is the most valuable layer in the matrix, and Apple owns it outright. The closer to the customer, the higher the multiple. Internal Hardware is bigger in revenue. End-User Hardware and Apps are bigger in compounding value. Now walk the matrix. The Era Matrix Companies as rows. Layers as columns. Each filled dot is a real owned business in that layer. The two market cap columns show the average across the first three years of an era and the last three years. Teal marks the first-window leader. Pink marks the last-window winner. They are never the same company. PC era · 1985 to 1999 The first-window leader was IBM at 79.9% share of named-player market cap. IBM owned three layers. Internal Hardware in mainframes and servers. Infrastructure in enterprise services. End-User Hardware in the original IBM PC and the ThinkPad. IBM won the opening for one reason. The IBM PC defined the category. Every other PC was an IBM clone. The hardware that ran the era belonged to IBM. The last-window winner was Microsoft at 41.9% share. Microsoft owned three layers. Operating System in Windows. Distribution through OEM bundling deals that put Windows on every PC sold. Applications in Office. Microsoft did not make hardware. Microsoft made the thing every piece of hardware needed to be useful. By 1999, Windows ran on 95% of PCs sold. Office had no real competition. Microsoft was, briefly, the most valuable company in the world. IBM dropped to 17.6% by the close. The hardware category gets the party started, and Operating Systems take the stage later. Dot-com era · 1995 to 2002 The first-window leader was Intel at 25.1% share. One layer. Internal Hardware. Every server, every workstation, every desktop running the web ran on Intel chips. The web was a hardware buildout before it was anything else. Cisco rode the same wave at 11.4%. IBM held on at 16.8%. The last-window winner was Microsoft at 24.7% share. Same Microsoft. Same three layers. Same Operating System. Same Distribution. Same Applications. This is the most important data point in the matrix. Microsoft is the only multi-era, multi-category winner in modern technology history. Won PC. Won Dot-com. Did it by holding the Operating System layer through the platform transition. Windows ran the local PC. Internet Explorer bundled into Windows became the way most people got to the web. Office moved from the desktop to the web. Same playbook. New surface. Same compounding. Intel finished Dot-com at 20.1%. Cisco at 16.8%. Both are still huge. Both are about to fall away in the next era. The hardware leader of one era is rarely the hardware leader of the next. Social/Mobile era · 2004 to 2020 The first-window Category King was Microsoft at 47.0% share. The same Microsoft. Two consecutive era wins and into the category lead of the third era’s opening act. This is what Rule 1 looks like on a chart. The last-window winner was Apple at 25.3% share. Apple owned four layers. Internal Hardware in Apple Silicon. Operating System in iOS. End-User Hardware in the iPhone and the App Store. Applications in Music, Maps, Messages, Camera, Photos. The most complete vertical integration in technology history. Apple owned the most valuable layers of the stack. Operating System. End-User Hardware. Apps. The three layers closest to the customer. Intel collapsed from 21.9% opening share to 3.9% closing share. The hardware leader of the prior era lost an entire era’s worth of share. Cisco never recovered from Dot-com. Microsoft, the vertical integrator, did. Same pattern as the IBM-to-Microsoft handoff. Hardware leaders fade when the layers they do not own start to compound. Nvidia opened Social/Mobile at 1.1% and closed at 3.2%. Few people saw what they were going to do next. AI era · 2024 to 2026 Nvidia is the first-window Category King at 22.3% share. One bingo point in Internal Hardware. Same role Intel played in Dot-com. Same role IBM played in PC. Behind Nvidia: Apple at 20.3% with four layers. Microsoft at 19.6% with three layers. Alphabet at 12.9% with five layers. The vertical integrators are already in position. The last-window Category King has not declared itself yet. The matrix says it will not be Nvidia, if they don’t add more points on the Bingo card. The matrix says it will be one of the companies with the most valuable and/or the most points on the Bingo Card. Ideally, it will own the Operating System plus the layers closest to the customer. The question is which one. Head to Head in the AI Era We have a few hypotheses about what might happen next. Hypothesis 1. Low-regulation states can be the Category King in US power. China is adding the equivalent of the entire US grid every few years. In 2025, China added 543 gigawatts of new capacity across all technologies. That single year of additions is 12% more than all the power plants combined in India. China’s total power generation capacity is now 3.75 terawatts. The US sits at 1.3 terawatts. Roughly three times the US. China is investing more than $500 billion per year in energy buildout. The US grid grew by a fraction of that. Regarding Power, the US federal government is being slow and stupid. Poorer, but prescient states can leapfrog their legacy. The regulation, the permitting, the environmental review cycles, the local NIMBY opposition. None of it is calibrated for an AI era that needs power yesterday. Microsoft had to revive Three Mile Island because building a new nuclear plant from scratch is essentially illegal. So xAI built Colossus 1 in Memphis, Tennessee. You would think they would be in the pole position for Colossus 2. Tennessee regulators slow-walked the gas turbine permits. Mississippi said yes. Mississippi’s governor announced xAI’s $20 billion investment as the largest single investment in Mississippi history. Tennessee just took Starbucks and In-N-Out headquarters. The states willing to build will win the AI era. California’s regulatory structure makes it effectively impossible to build the hyperscale facilities (500 MW+) that are now the standard for AI infrastructure. Developers in California top out around 50 MW. In Virginia and Texas, projects routinely run ten times that size. Every transmission line, backup generator, and power connection triggers CEQA environmental review, adding 18–36 months to any project timeline. The result: Silicon Valley has 489 MW of total data center capacity. Northern Virginia has 4,039 MW. That’s an 8x gap. And Virginia added over 1 GW of new capacity in 2025 alone. California added roughly 20 MW. Texas is booming with abundant land and no state income tax. Google committed $40 billion to Texas data center infrastructure through 2027. Microsoft is scrambling to secure natural gas-backed sites in Texas and West Virginia. No comparable investment is flowing into California. You can see smaller, speedier governments winning in Asia. Korea is the only country in the world other than the US with more than one trillion-dollar market cap company. Samsung Electronics crossed $1 trillion in May 2026. SK Hynix crossed three weeks later. Both built on the DRAM and HBM memory categories that feed the AI hardware buildout. Two trillion-dollar companies in a country of 51 million people, mostly because they own a critical category in the AI value stack. Bloomberg Economics forecasts Samsung and SK Hynix combined performance bonuses alone will grow from 4 trillion won in 2026 to 30 trillion won by 2028. South Korea’s finance minister is publicly debating how to use the tax windfall, with talk of creating a sovereign wealth fund to absorb it. Korea Herald is reporting Samsung and SK Hynix combined tax revenue over the next three years is projected to roughly equal half of South Korea’s national government debt. Two companies. Three years. Half the country’s debt. And Korea’s debt-to-GDP ratio is already only 49%, less than half the US level. The AI tax windfall is hitting a country that did not need it. That is what owning a single category in the AI value stack can do for an entire economy. Europe and high-regulation US states lose. California cannot build a 170-foot bridge in under three years. California is not going to build power plants and data centers in the AI era. Europe has not had a meaningful new technology company at scale in two decades. Both will be left behind. Western Europe, sadly, is becoming a place locked in the past, where tourists from the future visit. Hypothesis 2. Today, Alphabet and Muskonomy appear poised to win. Both are going for full bingo. Both own or are acquiring more layers than the rest of the field. Alphabet has the cash flow to solve the layers they do not yet own. Musk has the ability to raise immense amounts of capital just on the cult of his personality. Hypothesis 3. Expect weird M&A and JVs. The enemy of my enemy is my friend. Microsoft and OpenAI was the masterstroke of the era’s opening years. And now reportedly fraying. Musk had public beef with Anthropic, but his beef with OpenAI was bigger, so he cut a $1.25 billion per month deal with Anthropic for Colossus capacity [https://techcrunch.com/2026/05/20/anthropic-will-pay-xai-1-25-billion-per-month-for-compute/]. Google just cut a deal to pay xAI over $900 million per month for compute. This seems odd given they are both going for Bingo, but then again Google owns a healthy chunk of SpaceX. Solo players will shrivel. Standing alone with one or two layers is not a path to winning. In the mega categories. It is a path to being acquired, merged, or marginalized. With these hypotheses in mind, here is what must be true for each player to be the AI era Category King. Nvidia Nvidia has one spot on the bingo card. Proudly Internal Hardware. But their timing is awesome right now. Here’s what must be true for Nvidia to win. Nvidia must hope that the Internal Hardware phase is the longest of any era we’ve seen. This is the bet they are making, which is investing within the Internal Hardware part of the stack. Nvidia’s taken their monster cash flows and invested $2 billion into Lumentum, which is Optics and Components. They invested $3 billion into Corning for fiber optic cables. They invested $2 billion into Synopsys, a leading electronic design automation firm. And a $2 billion investment into Coreweave, to help build out clouds. If Internal Hardware’s reign lasts a long time, Nvidia will be fine. While they aren’t personally moving into other areas of the bingo card, Nvidia has committed $53 billion in about 170 deals across the value stack. * Power: $2.1 billion into Iris Energy * Operating Systems: $100 billion commitment to OpenAI, $10 billion into Anthropic and $2 billion into xAI * End User hardware: They participated in a $1 billion Series C into Wayve, a UK autonomous vehicle company. They also invested in a $675 million dollar round into Figure AI robotics. * End User applications: $50 million into Recursion Pharmaceuticals, $50 million into Kore.ai [http://kore.ai], a conversational AI for enterprise, participated in a $141 million Series B for Hippocratic.ai [http://hippocratic.ai], a medical-grade LLM. Equity stakes are great, but they are not businesses they own. Nvidia’s bet is that hardware demand stays insatiable for years. The playbook from here is the Apple playbook of the 2010s. Stock buybacks. Dividends. Returning cash to shareholders. Growing the stock through financial engineering rather than category expansion. Nvidia is going to be enormous. And unlikely to be the era mega-category winner. OpenAI OpenAI has two spots on the bingo card. Operating System in GPT. Applications in ChatGPT and enterprise products. Here’s what must be true for OpenAI to win. The rumored chip program would add Internal Hardware. Stargate would add Infrastructure. ChatGPT has consumer mindshare but does not own end-user hardware. Jony Ive left Apple to join OpenAI, and he’s building a consumer AI device [https://www.businessinsider.com/openai-cfo-ai-device-jony-ive-2026-6]. The Apple deal would add Distribution into End-User Hardware. That’s a lot to do, while trying to out-innovate Anthropic. And the rumored trust issues with Sam Altman make partnerships tricky. If the capital markets tighten, OpenAI does not have Alphabet’s cash flow to buy their way in. Anthropic Like OpenAI, Anthropic has two spots on the bingo card. Operating System in Claude. Applications in Claude.ai, Claude Code, and enterprise products. Unlike OpenAI, Anthropic had a rumored profitable quarter and ARR that seems to grow each month. Here’s what must be true for Anthropic to win. Anthropic has a lot of the bingo card to fill, with cash flow that is promising but not yet present. This means partnerships are the default. The Colossus deal with Musk gives Anthropic Infrastructure access without owning it. The enterprise embedding strategy through Claude Code, Claude in Excel, and Claude in Chrome is a credible path to being inside every workflow rather than owning a consumer surface. The math of building Power, Internal Hardware, Infrastructure, End User Hardware and Applications organically is brutal. The math of joining one of the bingo contenders is easier. Meta Meta has three spots on the bingo card. Infrastructure in their own data centers. End-User Hardware in Ray-Ban Meta and Quest. Applications in Facebook, Instagram, and WhatsApp. Here’s what must be true for Meta to win. Meta must solve for power, but as important, Meta would need to build or buy a real Operating System. Llama is a model, not an OS layer the way iOS or Gemini is. Meta would need Internal Hardware beyond MTIA. This is where all the cash they burned on AR/VR would have come in handy. These are a lot of spots on the bingo card to invest or buy their way into. Facebook needs a friend. A deep partnership with someone who has the layers Meta does not have. Standing alone, Meta does not have a credible path to bingo. Microsoft Microsoft has three spots on the bingo card. Infrastructure in Azure. Operating System in Windows and Copilot, as well as their partnership with OpenAI. Applications in M365 and Copilot apps. Here’s what must be true for Microsoft. Microsoft has to solve for Power and Internal Hardware. Power is the same constraint everyone faces, but Microsoft has the cash flow and the data center footprint to move first. Internal Hardware is the harder one. Microsoft has never been good at end-user hardware. Windows Mobile failed. Zune failed. Surface is a rounding error. And Microsoft lost over one billion on their retail stores. Can they win in Internal Hardware? The realistic path is acquisition. AMD has been rumored for a decade. Or can Microsoft time it right when hardware’s importance fades? This will be the biggest bet of Nadella’s career. Apple Apple already has four spots on the bingo card. Internal Hardware in Apple Silicon. Operating System in iOS, macOS, and Apple Intelligence. End-User Hardware in every device the customer touches. Applications in the Services bundle and Apple Intelligence features. Here’s what must be true for Apple. Apple has to solve for Power, Infrastructure, and AI user experience. Power and infrastructure are heavy capex businesses Apple has historically avoided, so Apple buys power on the grid. Apple leases compute from AWS and Azure. Apple has to own the inputs they currently rent. The biggest risk to Apple is in AI experience; Apple sucks. Think about the difference between Siri and ChatGPT or Claude. It’s a joke. And experience is what Apple has historically been awesome at. On Monday, June 8, they finally announced Siri AI [https://www.apple.com/newsroom/2026/06/apple-introduces-siri-ai-a-profoundly-more-capable-and-personal-assistant/] as part of the iOS 27 September software update. Their saving grace might be their hard stance on privacy, which is a credit to Tim Cook. His missional stance on privacy has built massive trust with consumers. If and when AI models can shrink enough to use less power and work on end-user hardware, Apple might be able to dodge their missing spots on the bingo card. Alphabet Alphabet has five spots on the bingo card. Internal Hardware in TPUs. Infrastructure in GCP. Operating System in Gemini and Android. End-User Hardware in Pixel and Nest. Applications in Gmail, Workspace, YouTube, Maps, Search. Here’s what must be true for Alphabet to win. Alphabet has to solve for Power, like the others. They also have to prove Internal Hardware is more than just a blip. Gemini has to integrate deeply into Gmail, Workspace, Search, Maps, and Android faster than AI cannibalizes the ad business that funds everything else. But they have what it takes. They have the cash flow to solve Power. They have the user base for the layers closest to the customer. They have the data to train the best models. The risk is internal. Alphabet’s organizational physics make speed hard. Muskonomy Musk theoretically has all six spots on the bingo card. Power in Tesla Energy via solar panels, batteries, and mega-pack batteries. Internal Hardware via AI4 and AI5 chips for Teslas and eventually Optimus. It is doubling down in Internal Hardware via its upcoming Terrafab investment as they build their own chips. Infrastructure in Colossus. Operating System in Grok and FSD. End-User Hardware in Tesla cars, Optimus, Starlink satellites, and Starlink terminals. Applications in FSD, Optimus, and the Macrohard concept. The only entity in tech history covering all six layers. Here’s what must be true for Musk to win. The clock is ticking as Musk has to execute M&A under a favorable White House. Tesla and SpaceX are strongly rumored to merge. Musk wanting to put all of his companies together is a major driver of the SpaceX IPO. The three companies have to stay coordinated under one strategic umbrella, even though they have different cap tables and different boards. While Tesla is a $100 billion revenue company and SpaceX has been cash flow positive for a decade, none of it is throwing off the free cash flow of an Alphabet or Nvidia. And Musk has to survive himself. This is a massive keyman risk, given his penchant to say what he thinks regardless of controversy. The Ultimate Game of Thrones The AI era is the ultimate mega category game of thrones. Eight players. Six layers. Five principles from history and one open layer at the bottom of the stack. One throne. Here is what to do about it. The Career Decision Everyone is fighting for layers two through six. Almost no one in the United States is fighting for Power. That is where the wide-open category space is. Whoever builds the company that figures out how to deliver gigawatts of clean, fast, permittable power for AI data centers becomes the category kingmaker for every other player in the matrix. This applies to careers across the stack. Take the job at the company that has a credible plan to solve Power, whether that company is a state economic development office in Tennessee or Mississippi or Texas, a nuclear startup, a distributed energy network, a grid operator, or one of the bingo contenders investing in their own generation. The talent shortage in Power is enormous. The category potential is enormous. The competition is thin. Run away from anything that has one layer and a great story. Intel was the best-performing tech stock of 1995. Nvidia is the best-performing tech stock of 2024 and 2025. One of those statements aged badly. The other one is sitting on a pattern that says it will too. And remember. All of this analysis could be wrong. We’re providing the category lens to a discussion about the potential for who will be the biggest winners, in a business media landscape that is ignorant of how market categories work. The Creator Capitalist Move Learning how to invest (money, time, and intellectual capital [https://www.categorypirates.news/p/intellectual-capital-5-ways-to-turn?utm_source=publication-search]) wisely in this Game of Thrones can extend your financial capital [https://www.categorypirates.news/p/runway-not-retirement-the-creator]. And open the aperture of your thinking about the future. Solving problems across the value stack is worth its weight in gold. Look at the gaps in the matrix. Power is wide open. Infrastructure outside the hyperscalers is contested. Operating Systems belong to a handful of giants, but the categories that ride on top of those operating systems are wide open. The Creator Capitalist white space is in the categories Wall Street has not named yet. Build a category that requires AI but is not defined by it. Do not build an AI app. Build a category that an AI application is one expression of. Whoever languages [https://www.categorypirates.news/p/category-design-tip-use-languaging?utm_source=publication-search] a new category that lives in the matrix gaps gets to be the next Microsoft. The compounding from getting that right is bigger than anything else you will do with your career or your capital. The Speculative Capital Question If you are thoughtful about picking the next Category King, you can live like a king. If you are already invested in the S&P 500, you are already invested in AI Category Kings. The matrix proves the point. Apple, Microsoft, Alphabet, Meta, Nvidia, and the rest of the top names sit inside the S&P 500. They are most of the index. The S&P 500 is the most boring and reliable way to participate in the AI era. Buy. Hold. Forty years from now your $10,000 looks like Microsoft money. Probably not Microsoft money. But it looks like a winning lottery ticket someone left on your desk. Don’t forget. The S&P 500 has a median return of 13.1% over the last 50 years. If you want a little more juice, QQQ works. The Nasdaq 100 overweights the technology names that win eras. Same companies. Higher concentration. Same buy-and-hold thesis. 19% is the median return of the QQQ since its 1999 launch. If you have speculative capital, use your category design brain to figure out the top three companies that will get to bingo. Look at the matrix. Find the companies with five or six layers, or a credible path to five or six layers. Make your call on which one will own Operating Systems plus the layers closest to the customer. Make your call on which weird M&A or JV moves first. Make your call on which solo player gets acquired and at what price. Make your call on whether Muskonomy stays coordinated under one strategic umbrella long enough to claim bingo. The Category Kings will be the players who fill out the most valuable rows. We’ll be back with another episode next week. Hey Ho, Let’s Go! Arrrrrrr, Category Pirates Eddie Yoon Christopher Lochhead P.S. Do you like our Deep Dive Reports? Why or why not? Please let us know on LinkedIn and Substack notes. P.P.S. Do you like the name Deep Dive Reports? Should we call it PSJ DDR? Dance Dance Revolution?

"Lowest Consumer Sentiment" Is Good News?

The University of Michigan consumer sentiment index [https://www.wsj.com/livecoverage/stock-market-today-dow-sp-500-nasdaq-05-22-2026/card/consumer-sentiment-drops-to-new-low-university-of-michigan-survey-finds-5jUreNorgRrKwatx1SC4?mod=hp_lead_pos2] just came in at 44.8. The lowest reading in the history of the survey. April was already the worst on record. We beat it, then beat it again a month later. Near-zero unemployment. Record-high stock market. GDP growing. Entrepreneurship at an all-time high. And the consumer says this is the worst they can remember. Something doesn’t add up. Here’s what we covered in this episode: 1. “Record low” is good news. The consumer isn’t behaving like the survey says. Nobody’s broke and curling up in a ball. They’re getting smart. Waiting longer to buy a car. Buying less packaged food. Trading stuff for experiences. Because the real issue is not the consumer ‘income statement’. The real issue is the consumer ‘balance sheet’ is bloated and designed for a nuclear family that is declining. * Single family homes that are too expensive and too much space for single people. * Big cars sold for families that aren’t being formed. * Groceries to cook in a DoorDash world. The old linear life script is dissolving. Get married, get the house, get the promotion, collect your Scooby Snacks of purpose along the way. When that script breaks, people go find meaning on their own terms. 2. The synthetic customer, and the race to beige. Bain published “Synthetic Customers Earn Their Stripes.” [https://www.bain.com/insights/synthetic-customers-earn-their-stripes/] AI-generated buyers, backtested against a real conjoint study, replicated about 90% of the outcomes. Which features drive choice. Which products to launch. Even early price sensitivity. Target and US Bank are already testing on synthetic audiences before anything ships. The technology is a huge unlock. Sadly most companies will use it the wrong way. Everybody builds one. Everybody aims it at the fat part of the bell curve. They optimize the average customer into the ground and call it insight. Synthetic customers trained on average consumers makes us all dumber. We have an opposite POV…check it out. 3. The triple wasn’t good enough. The QQQ is up around 600% over ten years, roughly 21% a year. The S&P did 13 to 14%. To get rich, you just had to sit still. Gen Z isn’t sitting still. A third have played prediction markets, a third hold crypto, and a quarter of their portfolios sit in non-traditional assets. The punchline: roughly 69% of Polymarket accounts have lost money since 2022. The FOMO flipped. A triple every year for a decade, and a whole generation said, not good enough, I want the fences. Why swing that hard? 9/11, then 2008, then COVID, each one before they could legally drink. When every safety net detonates that early, “wait 40 years” sounds naive. What’s coming up on Pirate Street Journal Three weeks a month, we drop the video. Three topics, thirty minutes, one cowbell. Once a month, we publish a written deep dive, the kind of category analysis you cannot get anywhere else. That one is for paying subscribers only, monthly and founding. Next week, we’ll be publishing a deep dive. Two ways to climb aboard now: Monthly subscriber: $20/month. You’ve done dumber things with $20. Founding subscriber: $375/year. For about a dollar a day, you get every mini-book we’ve ever written (300+), every audiobook (30+), digital copies of all seven of our Big Books, and unlimited access to The Pirate Eddie Bot and Pirate Christopher Bot, your 24/7 AI jamming partners for category building. Subscribe today, get the next deep dive the day it drops, and start jamming with the bots. Recorded Friday, May 29. Every number above is as of that morning. Piratey disclaimer: This is NOT financial advice. None of us have a Series 63, Series 7, Series 6, CPAs, CFAs, IUDs, IEDs, and hopefully not IBS (this makes DUDE Wipes sad). Stay tuned for next week’s episode. Hey Ho, Let’s Go! Arrrrrrr, Category Pirates Eddie Yoon Christopher Lochhead This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.categorypirates.news/subscribe [https://www.categorypirates.news/subscribe?utm_medium=podcast&utm_campaign=CTA_2]

How Jennifer Hall Thornton, An Agentic Mom, Agentified Her Own Kids

Jennifer Hall Thornton [https://www.linkedin.com/in/jenniferhallthornton/] has a PhD and a law degree. Her kids’ friends call her Dr. Doctor. She ran the everything-but-sales side of a digital company while raising two kids thirteen months apart, with an elderly mother nearby and a husband who lived on a plane. She was in the first Academy cohort, before anyone could tell her the timing was right. She told them the timing was right. She is also an Agentic Mom [https://www.categorypirates.news/p/motherhood-30-how-creator-capitalist]. Jennifer built a set of Claude skills that logs into her kids’ college Canvas, checks for anything new, downloads it, summarizes it, emails them, and then builds practice quizzes from the material. She built it for herself first. Then she shared it with her 18 and 19-year-olds. The agentic mom is agentifying her children. A mother who learned to delegate the donkey work of her own life to AI agents, looked at her kids drowning in logins and busywork, and decided they did not need to be good at the stuff she had to be good at to survive school. Pirate Jennifer did not raise her kids to be the best version of her. She is consulting them into the best version of them. Most parents try to clone themselves. They want the kid to do the thing they did, or the thing they wish they had done. Pirate Jennifer did the opposite, and she did it before Creator Capitalist gave her the words. She made her kids write thank-you notes. Three a year, starting in third grade, one for each year of school by graduation. Both kids wrote far more than required. Her son’s handwriting is unreadable, and his fourth-grade teacher still messages him two years out of high school because he wrote her a note. Her daughter emailed a high school neuroscience teacher as a college freshman because something he taught her showed up on a biology midterm her professor never covered. He wrote back inside the hour. That is relationship capital [https://www.categorypirates.news/p/relationship-capital-the-8-connections], built by a nine-year-old and ten-year-old, compounding for a decade. The dinner table was a debate where you had to keep up no matter your age. That is intellectual capital. Showing up late, turning in sloppy work, going to office hours because you are actually curious, all of it builds reputation capital. She taught the four capitals before her kids could spell them. Then Creator Capitalist [https://creatorcapitalist.ai/] came out, and Jennifer read it and thought, well, duh. She had been running the flywheel the whole time. She finally had the languaging [https://www.categorypirates.news/p/languaging-the-strategic-use-of-language] for it. She told us that Creator Capitalist starts thirty years earlier than we thought. We wrote Creator Capitalist for adults. Mid-career or later, people who have already built some of the four capitals and are looking for what is next. Jennifer read it and started talking about how the frameworks were for her teenagers. In her velvet-hammer way, she told the donkeys to look again. Creator Capitalist is for kids, too. She is right. The four capitals are a flywheel, the same compounding engine that Christopher teaches young people about money. Except it starts spinning at 19 instead of 50. Tell a 19-year-old they have no intellectual capital, and they will believe you. Pirate Jennifer calls BS. Intellectual capital [https://www.categorypirates.news/p/intellectual-capital-5-ways-to-turn] is not the diploma. It is how you approach a problem, and what your friends come to you for. Her son is a poli-sci major who taught himself to run every machine in the engineering lab, became the TA for a course only engineering students are allowed to teach, and got paid for it. He applied to twelve colleges and got into twelve, with no sports and no titles, because he was different and had a story to tell. That is a Creator Capitalist, two years out of high school. Her daughter Emily is studying biochemistry at UT Austin. She loads every note and slide into a Claude project and has Claude teach her, write quizzes, grade them, then build new quizzes targeting only what she is getting wrong. Before her last exam she asked Claude for the hardest questions it could write. When the professor found one Claude missed, she fed it back and told Claude it blew it. She is co-creating her education with AI, then taking it back into the traditional system and winning. Her friends say it is too complicated. They are going to regret it. Here’s how to navigate this conversation: * 06:01 – Parenting in the age of AI: Pirate Christopher opens by asking what it is actually like to raise an 18 and 19-year-old right now, and why Jennifer is a far better parent to young adults than she was to little kids. * 11:09 – The book whack: How Jennifer told Category Pirates that Creator Capitalist, the book they wrote for mid-career adults, is really a curriculum for teenagers, and why they did not see it coming. * 11:44 – The thank-you note machine: Three notes a year starting in third grade, the fourth-grade teacher who still writes back, and relationship capital built by a nine-year-old. * 14:34 – The poli-sci grease monkey: The son who got into all twelve colleges he applied to with no sports and no titles, then became the engineering lab TA he was technically not allowed to be. * 22:55 – The flywheel at 19: Why the four capitals are the same compounding engine Christopher teaches kids about money, except they start spinning thirty years earlier. * 33:07 – “I call BS”: Jennifer dismantles the myth that young people have no intellectual capital, and explains what intellectual capital actually is. * 34:44 – The agentic mom agentifies her kids: The Claude skill that logs into Canvas, summarizes new material, and builds quizzes, and the moment Jennifer realized she should hand it to her children. * 39:26 – AI in the classroom: Why banning AI is the wrong move, what counts as cheating versus co-creating, and the World Book Encyclopedia parallel. * 42:01 – Teaching Claude to teach you: Emily’s biochemistry workflow, the hardest-quiz challenge, and the Feynman technique applied to an AI. * 46:33 – Build a bot, sell it to your friends: Eddie on why a college kid should build an O-chem TA bot, charge classmates fifty bucks, and learn the material better in the process. * 53:15 – The Dickens exercise: If 18-year-old Jennifer were starting today, loving to learn and unafraid to fail, what would she build. * 01:04:19 – Start with relationship capital: Why young people should start the flywheel with the capital they already have, and how to find the intellectual capital you do not know you own. We were not wrong about Creator Capitalist. Jennifer just showed us it starts earlier than we thought. The framework does not care how old you are. It cares whether you get the flywheel spinning before everyone else does. Jennifer did it intuitively, over fifteen years, without the words. She raised two Creator Capitalists, then spent this conversation reverse-engineering how she pulled it off. Imagine what that compounding looks like for a kid who has the framework AND the next twenty years ahead of them, instead of behind them. Whether you are the parent or the kid, the work starts the same way. Get the agents running. Get the flywheel spinning. Stop doing the donkey work a machine can do, and spend the time you save building the four capitals nobody can take from you. Arrrrrr, Category Pirates 🏴‍☠️ Eddie Yoon Christopher Lochhead P.S. The fastest way to start agentifying your own life (or your kid’s). Pirate Jennifer built her agents alone, a step or two ahead of everyone around her. You do not have to. Become a Founding Subscriber and you get access to The Pirate Eddie Bot and the new Pirate Christopher Bot, the fastest way we know to put the four capitals to work in your life, your career, and your kids’ education, without waiting for permission from a school that still thinks AI is cheating. → [Become a Founding Subscriber to get access here.] [https://www.categorypirates.news/subscribe] This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.categorypirates.news/subscribe [https://www.categorypirates.news/subscribe?utm_medium=podcast&utm_campaign=CTA_2]

Category Queen vs. Category Queen: Is OpenAI About to Get Dethroned?

Anthropic is paying SpaceX $1.25 billion a month for compute. Every month through May of 2029. Roughly $45 billion total. That single contract is bigger than SpaceX’s entire annual revenue today. Software is paying hardware. Hardware is paying energy. Energy is paying space. Is it three card monte? Or is the pie getting massively bigger? This should be on every business news front page this week. Instead, the headlines are about Sam Altman’s house getting attacked, Eric Schmidt getting booed off a stage, and OpenAI quietly leaking that it might file for an IPO too. Here’s what we covered in this episode: 1. AI gets an F in marketing. 360,000 Americans are in Facebook groups organized against data centers. AI is polling less popular than ICE, less popular than Trump, less popular than politicians. The technology being protested is curing diseases and driving much of US GDP growth. So why are so many angry at AI? Pirate Eddie has a theory, and it does not involve AI at all. Marriage rates. Birth rates. Teenage drinking. Labor force participation among young men. All down. Life stages and zest for life as we know was crumbling before AI. 2. SpaceX is going public. The TAM is the size of the American economy. The S-1 is pitching a $28.5 trillion total addressable market. U.S. GDP is $32 trillion. SpaceX is asking public markets to fund a business roughly the size of the entire American economy, run by a CEO whose vesting schedule requires interplanetary colonization. Imagine being on the dock before the Nina, Pinta and Santa Maria set sail and you were offered a chance to invest in the new world. What was the ROI on America? What if you had a chance to invest in Space? But the history of IPOs is that retail investors get hurt. The history of Elon Musk is that betting against him is also expensive. The Pirates how category designers think about an IPO this size without losing your mind or your savings. 3. The category queen vs. the category queen. The fastest-growing company in the history of business is not OpenAI. It is Anthropic. Anthropic was founded by people who left OpenAI. They are now in talks at $900 billion. Higher valuation. Higher growth. First profitable quarter ever at $11 billion in revenue. Andrej Karpathy, founding member of OpenAI and the man who coined vibe coding, just joined Anthropic. Ross Nordeen of xAI joined Anthropic earlier this month. The defectors are recruiting the defectors. Category Design 101 says the queen takes 76% of the economics. Everyone else shares 13%. So which one is the queen? Pirate Christopher has a frame on the video that recasts the entire question. He thinks this is not OpenAI versus Anthropic at all. This is the (new) Pirate Street Journal. Every Wednesday, we pick three headlines worth paying attention to and break down the category underneath. Three Wednesdays a month, the video is free (for now). Once a month, we drop a written deep dive for paid subscribers, the kind of category analysis you cannot get anywhere else. Two ways to climb aboard: * Monthly subscriber: $20/month. You’ve done dumber things with $20. * Founding subscriber: $375/year. For about a dollar a day, you get every mini-book we’ve ever written (300+), every audiobook (30+), digital copies of all seven of our Big Books, and unlimited access to The Pirate Eddie Bot [https://categorypirates.com/pages/bots] and Pirate Christopher Bot [https://categorypirates.com/pages/bots], your 24/7 AI jamming partners for category building. To never miss a deep dive, become a subscriber today. [https://www.categorypirates.news/subscribe] What’s coming up on Pirate Street Journal A few of the threads from this week’s episode are running headlong into a much bigger story, which is the IPO season that is about to define the next decade of public markets. On June 10th, two days before SpaceX is estimated to start trading, we are publishing the next PSJ written deep dive. We will be working through how a category designer thinks about investing in an IPO of this scale, what the category math says about SpaceX, OpenAI, and Anthropic going public in the same window, and we don’t know what the headlines will do between now and then, so there will be more. Piratey disclaimer: This is NOT financial advice. None of us have Series 63, Series 7, Series 6 7, CPAs, CFAs, IUDs, IEDs, and hopefully not IBS (this makes DUDE Wipes sad). Stay tuned for next week’s episode! Hey Ho, Let’s Go! Arrrrrrr, Category Pirates Eddie Yoon [https://eddiewouldgrow.com/] Christopher Lochhead [https://lochhead.com/] P.S. - Founding subscribers get every mini-book we have ever written (300+), every audiobook (30+), digital copies of all seven Big Books, and unlimited access to The Pirate Eddie Bot and Pirate Christopher Bot [https://categorypirates.com/pages/bots], your 24/7 AI jamming partners for category building for about a dollar a day. You have done dumber things with a dollar a day. 👉 Become a founding subscriber. [https://www.categorypirates.news/subscribe] This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.categorypirates.news/subscribe [https://www.categorypirates.news/subscribe?utm_medium=podcast&utm_campaign=CTA_2]

The Great Re-Rating: Is the SaaSpocalypse Real?

This is a free preview of a paid episode. To hear more, visit www.categorypirates.news [https://www.categorypirates.news?utm_medium=podcast&utm_campaign=CTA_7] Last week, we recorded the very first episode of the Pirate Street Journal. The Pirate Street Journal is for leaders with a different mind. A different take on business news. Through the category lens. Our mini-books are timeless. PSJs are timely. Our mini-books are long stories longer. PSJs have 30-minute seat belts. Our mini-books are thinker’s high. PSJs try to help you think before you act. But, but, but, but… Piratey disclaimer: This is NOT financial advice. None of us have Series 63, Series 7, Series 6 7, CPAs, CFAs, IUDs, IEDs, and hopefully not IBS (this makes DUDE Wipes sad). Think of this like professional wrestling. It’s entertainment. Don’t be so smart, you’re stupid and suplex your safety net savings. Hey Ho, Let’s Go! PSJ is the new weekly thing. The video’s free. The deep-dive written analysis is paywalled. Watching makes you informed. Reading makes you different. Two ways to climb aboard: * Monthly subscriber: $20/month. You’ve done dumber things with $20. * Founding subscriber: $375/year. For about a dollar a day, you get every mini-book we’ve ever written (300+), every audiobook (30+), digital copies of all seven of our Big Books, and unlimited access to The Pirate Eddie Bot [https://categorypirates.com/pages/bots] and Pirate Christopher Bot [https://categorypirates.com/pages/bots], your 24/7 AI jamming partners for category building. To read this week’s deep dive, become a subscriber today. [https://www.categorypirates.news/subscribe] 1. Why is Salesforce down? Why is Micron up? The Mag 7 reported earnings, and they were great overall. But here’s some weird data. Salesforce (one of the Category Kings of SaaS) lost about a third of its market capitalization in the last 12 months, despite strong revenue and operating income. Forward P/E down 28% in twelve months. Benioff just announced a $50 billion stock buyback, one of the largest in corporate history. Micron (memory for AI) saw over a 6x increase in its stock price in the last 12 months, also with incredible revenue and operating income. Forward P/E sat at roughly 3x a year ago. Today it is over 7x. The stock more than tripled in that window, but earnings grew faster than the multiple did. In the columns, we have the Mag 7, plus SpaceX, which is soon to go public, as well as Micron and Salesforce. The rows are what matter. * Top row, Potential investors. Forward P/E above roughly 27, which is about +5 above the S&P 500 average PE multiple. * Middle row, market-average band. Forward P/E is roughly 17 to 27. The S&P 500 lives here at around 22. * Bottom row, Performance investors. Forward P/E below roughly 17, which is about -5 below the S&P 500 average PE multiple. The actual PEs are merely a placeholder, as there’s nothing magic about plus or minus 5 from the S&P 500 average. We want to discuss the fact that there are two types of investors. Performance investors. They invest in companies because of their current and near-term performance. Their performance is predictable, reliable, and steady. Sometimes slow, but never surprising. These are usually Category Kings today. These companies are valued at lower multiples, whether it is price to earnings, enterprise value to EBITDA, or price to sales. And there are Potential investors. They invest in companies regardless of their performance now or in the near term, but in their long-term future potential. Usually, these are companies that can become future Category Kings that no one else really sees. These companies are valued at much higher multiples, usually because earnings or sales are emerging and expected to accelerate. When Potential investors start buying a stock, they lift the forward PE multiple as they are willing to pay a premium for potential. They think the category size of prize [https://www.categorypirates.news/p/sizing-the-category-prize] is growing and has huge upside. They think the category is on the good side of the S-curve. All boats rise with the tide. When Potential investors sell a stock to Performance investors, it depresses the forward PE multiple because they aren’t willing to pay a premium for potential. They think the category size of prize is static and has limited upside. 2. Are you on the good or bad side of the s-curve Performance vs. Potential investors are fundamentally debating one fundamental question. Is the category and company on the good or bad side of the S-curve? You don’t have to be right on the precise number and date. It’s not like picking black 17 on the roulette table. It’s just picking black or white. Using data and Category Design. And thinking about thinking. You don’t have to predict timing. You don’t have to predict a number. You should, but don’t have to, do fancy analysis. Left or right of the S-curve is the question. If you are right, and everyone agrees with you, it can be a profitable bet. If you are right, and everyone disagrees with you, you can create generational wealth. But you have to be comfortable with the loneliness, name-calling and mockery that comes with rejecting the premise. When Pirate Eddie wrote in HBR that Netflix’s 80% stock drop in 2011 was Wall Street being dumb [https://hbr.org/2011/08/why-im-happy-netflix-raised-it], Wall Street called him dumb. When Pirate Eddie shared on CNBC about Tesla’s superconsumer [https://www.youtube.com/watch?v=sAxXwCC6hrc] being a new superconsumer who valued both functional and fun cars, Wall Street called him dumb again. When Pirate Eddie wrote in HBR that General Mills should sell its cereal business [https://hbr.org/2018/02/why-dominating-your-category-can-be-a-flawed-strategy], he made a lot of former clients/friends at General Mills angry. But the data at the time was undeniable. 12 years of category decline. And unless you believed carbs and sugar were ever coming back into vogue, General Mills’ cereal business would never be more valuable than it is today. And they should sell it. General Mills’ stock is down 38%, while Kellanova (old Kelloggs with cereal spun out) is up +4% since being acquired by Mars. General Mills’ PE ratio is 8x, and Kellanova’s PE is 23x. Sometimes being right doesn’t feel great at first. But the cost of being legendary is the willingness to be different. 3. Re-rating is a result of Category Design Re-rating is when Wall Street decides a company’s multiple should be higher or lower. Revenue, gross margins, and cash flow don’t change. The value of those economics does. Everything we value, we’ve been taught to value. Re-ratings are simply a redefinition of the Category. Did you know Domino’s Pizza was the 2nd best performing stock from 2010 through the end of 2019? Why? It transformed from a pizza delivery company to a tech company that happens to deliver pizzas. They invested heavily in their ‘pizza tracker’, apps, and frictionless mobile apps. It’s Category Design 101. And if you invested $1,000 into Domino’s at the beginning of 2010, you’d have $40,000 in 10 years. The best part is that re-ratings can happen slowly. You could have jumped on the Domino’s train any of the first 9 years of its run and done well. Wall Street is often blind to Category Design. Category Design is your unfair advantage. 4. The SaaSpocalypse is overstated The financial press has decided this is the death of software. Salesforce down $135 billion. ServiceNow down $100 billion. Workday down $50 billion. Hundreds of billions of dollars in enterprise software market cap gone in a year. It is the wrong frame. Software is not dying. On May 15, Marc Benioff sat down with the All-In Podcast and said, “… the software market’s rerated. It happens every now and then. There are cycles. You know, I’ve been doing Salesforce for 27 years, enterprise software for 40. And the market’s rerated.” — Marc Benioff The earnings are fine, but the multiples got cut. Salesforce guided to do $46 billion in revenue and $16 billion in cash flow this year. Performance is not the problem. Potential is. The market used to price these companies as Potential plays. Software is eating the world, every business needs a CRM SaaS, the seat count never stops growing. That story matured. The category got knowable. The TAM became visible. So the market quietly moved these names down a row. From Potential. To Neutral. Some all the way to Performance. Benioff is responding to this exactly the way a category designer should. He is doing three things in parallel. Buying back stock at compressed multiples because he believes the business is worth more than the market pays for it. Acquiring companies (Informatica), while, in his words, “everything’s a little cheaper.” And, most importantly, repositioning Salesforce out of the SaaS category entirely. AgentForce. Slack as the context engine. Humans, agents, and headless platforms interoperating. If that repositioning works, Salesforce gets re-rated up again under a new category label. Same business. Different multiple. Different shareholders. That is the move. 4. AI hardware is more valuable than AI software The content in this section is 100% created by AJ on X @alojoh [https://x.com/alojoh]. He’s a former Goldman Sachs investment banker, who built his own pirate ship that is a combination of investment research and trading advice with a rare alignment of incentives with his subscribers. The goal of equity research is to drive trading revenue for investment banking, not necessarily at the benefit of the reader of the research. There is a strong motivation to put out positive news and analysis for investment banking clients and even stronger reluctance to say anything negative about those same clients. It is not 100% trustworthy. The incentives for most traders/investors is to grow their own returns, even at the expense of subscribers/readers. They may tell you to buy a stock, but only after they bought it, and at times, they sell as they tell you to buy. Or their incentive is to grow their assets under management and charge you 2% of assets and 20% of carry for as long as possible. AJ is the odd combination of a top-tier investment researcher who uses it to trade for his own account. His basic subscription on X is only $7/month, but his hardcore channel is $500/month, which Pirate Eddie subscribes to and has already generated more than a 20x ROI on the cost of the annual hardcore subscription. Sign up if you like buried treasure. AJ is a pirate who routinely rejects the premise. This section is a synthesis of his insights and work and is shared with his permission. One of the most provocative quotes from AJ is, “Hardware is the endgame” for AI. For decades, software economics have always trumped hardware economics. In the age of AI, it’s no longer always true. Eighteen months ago, the DRAM memory industry was effectively left for dead. Post-COVID demand had snapped back. Customers had massively over-ordered during the shortages. The industry was working through an enormous inventory overhang. Prices collapsed. The major players took huge losses. Standard semiconductor cycle. Standard low-multiple memory business. Then the AI buildout happened underneath them. In two or three months, the entire setup will be inverted. Utilization went from roughly 20% to over 100%. Hyperscalers moved into “whatever it takes” pricing mode on memory. Long-term supply agreements got locked in across the three major players (Micron, Samsung, SK Hynix) who together produce roughly 95% of global DRAM. Memory used to be a commodity. Now it is AI infrastructure. The cleanest way to show you why is to put Micron next to the most darling AI software stock on the public markets right now. Palantir. * Revenue growth (year over year). Micron +196%. Palantir +85%. * Operating margin. Micron 67.6% (GAAP). Palantir 60% adjusted, 46% GAAP. Yes, you read that right. The memory chip company has a higher operating margin than the AI software company. * Rule of 40 (revenue growth percent plus operating margin percent). This is an awesome rule of thumb. If you want to know if a business is healthy or not, just take their revenue growth percent and add it up with their operating margin percent. If both are greater than 40%, you have a great business. Per Palantir’s own Q1 2026 investor deck, Micron sits at 265%. Palantir sits at 145%. Palantir is literally publishing slides that show Micron crushing them on the metric they brag about. * Forward P/E. Micron roughly 7x. Palantir roughly 95x. Palantir trades at a multiple thirteen times higher than Micron, while growing slower, earning lower margins, and scoring lower on its own framework of choice. This AI hardware company has the better business by every operating measure. The software company has the better multiple by every valuation measure. That is what a re-rating up looks like when it is not yet complete on one side and what a Potential premium looks like when it might be ahead of itself on the other. This does not mean Palantir is doomed. It means the gap between the two multiples is going to close. 5. The Mag-7 will be the Mag-10. Abundance will follow We are not telling you software is over. Anthropic and OpenAI will do extraordinarily well. Pirate Christopher made the call on the pod. These are the largest, fastest-growing companies ever created, sitting in private markets at scale that exceeds most of the Mag-7. They are the leadership of the next decade of AI software. When they go public, the math changes for everyone. But the SaaSpocalypse is more about the rise of AI hardware and hardware in general. This is why Micron is exploding. And why SpaceX is about to launch the largest IPO in history. The Mag-7 framing is already obsolete. Three names belong in the group that are not in it yet, and the math suggests all three of them get added inside the next twelve months. * OpenAI. Created the fastest-growing category in startup history. Reported to be exploring an IPO window. * Anthropic. Became the fastest-growing company in history. Already operating at a scale that exceeds most public Mag-7 members on the metric that matters most, category formation speed. * SpaceX. IPOing in June. Already cash-flow positive thanks to the Colossus lease deal with Anthropic, and reportedly being talked about at $1.75 to $2 trillion. And it does not stop at ten. Stripe could be next. The private IPOs of the last decade are starting to unwind. Every name that has been trapped in private markets at Potential multiples is about to become available to public investors who have been starved for Potential exposure. One more thing the matrix hints at. There is still no energy company in the Mag-anything bucket. Tesla is sometimes pitched as the answer because of its solar and battery business. The actual AI energy story is coming. There is a real chance we are entering one of the largest equity bull markets in modern history, and most professionals are still positioned for a recession that did not happen. Global GDP in 1926 was about $3 trillion. Today it is roughly $126 trillion. That is a ~50x expansion of human-created value in 100 years. The mechanism was not “stocks went up.” The mechanism was that humans kept designing new categories that created value from nothing. Refrigeration. Electrification. The semiconductor. The internet. Each of those was, in its moment, an abundance boom. Each one looked like a bubble in the middle innings. Each one was actually a re-rating of the entire economy onto a new S-curve. If AI is the next one, the abundance argument runs like this: * The S&P 500 goes up because the Mag-10 carries it. The top names are already a huge share of index weight. When they re-rate up, the index does too, almost mechanically. This is already happening. * The Nasdaq goes up more, because both the AI hardware layer and the AI software layer live inside it. Hardware re-rates up. Software re-sorts. The up names are bigger than the down names. Net positive. * Productivity finally shows up in the data. If even 30% of the AI productivity case is real, US GDP growth surprises to the upside for years. That is the macro backdrop bulls have been waiting for since the 1990s. * The Potential investor pool expands as more capital chases the abundance narrative. More Potential capital, chasing the same names, pushes multiples higher. The re-rating accelerates. If we are right about the abundance boom, the index investor does well. The Performance investor does fine. The Potential investor does great. But the category designer wins differently. The category designer wins by spotting re-ratings, in both directions, before the analyst class can model them. All of this boils down to two questions. * Is the category on the good or bad side of the s-curve? * Is the risk of staying the same greater than the risk of changing? If you can answer those questions honestly about your business, your career, your portfolio, your category, you could create a massive outcome and avoid sailing into rocky waters. Arrrrrrr, Category Pirates Eddie Yoon [https://eddiewouldgrow.com/] Christopher Lochhead [https://lochhead.com/]