Welcome to Postcards From the Edge of the World Don’t recognize this sender? Unsubscribe with one click Garrett Baldwin recently imported your email address from another platform to Substack. You'll now receive their posts via email or the Substack app. To set up your profile and discover more on Substack, click here.  'Cause You're Otis Knight. To Whom It May Concern (You): What’s the point of tonight’s letter… It’s that not all bottlenecks are chokepoints. Last week, I showed you how the denominator breaks. The Debasement Index... money supply divided by real GDP. Demographics attacks the bottom of that equation by reducing the workforce. That means fewer people, less output, and more printing. I named it Dependency Driven Debasement, and the math hasn’t changed since you read it. But I left something out. There’s a second force attacking the denominator right now. It’s coming from the opposite direction. Demographics removes bodies from the equation. This one removes the value of the remaining bodies. And it has happened before. The LadderIn 1790, 90% of the American labor force worked in agriculture. By 1900, it was 40%. By 2000, it was under three. The most dramatic collapse came between 1880 and 1930... a thirty-percentage-point drop in 50 years. Mechanization and railroads made the family farm obsolete. Millions of people lost the only work they’d ever known. And the economy didn’t just absorb them. It moved them up a rung. Off the farm and into the factory. The Industrial Revolution did the same thing in textiles. A single power loom increased output per worker by 40%. The Luddites smashed machines in 1811, and the British government hanged them for it. But within 60 years, the factory system had replaced the entire cottage industry, and the displaced workers climbed another rung. The statement goes this… “Out of the mill and into the office.” Put that on a stupid shirt… People moved into management, administration, and clerical work... the emerging class of people who thought for a living rather than lift for a living. Something big happened next… Ford built the assembly line. In 1913, it took 12.5 hours to assemble a car. By 1914, it took about 90 minutes. The skilled craftsmen were replaced by semi-skilled workers doing one repetitive task... and the economy moved people up again and into the professional class. They shifted to the knowledge economy that would dominate the next century. I know how hard it is to look at AI and think that all hell is gonna break loose. But… Each time, technology eliminated one category of work, and the economy adapted by pushing workers higher on the cognitive ladder. And each time, the transition got faster. Agricultural mechanization took 100 years. The Industrial Revolution took 60 years. Ford restructured manufacturing in 15. Computerization reshaped office work in the twentieth century. The pattern was accelerating, but the ladder always held. There was always a higher rung. Here’s what’s different this time. AI is not eliminating the bottom of the ladder. It’s eliminating the top. Every previous disruption pushed workers upward into cognitive labor... because machines couldn’t think. We had a model in America around college, degrees and an intelligence premium... The wage gap between a college graduate and a high school graduate... widened every decade from 1980 onward. The entire economic advice of the last 40 years was: climb the ladder, because the machines can’t follow you there. An MIT study found that AI can already replace nearly 12% of the U.S. workforce with current technology. And workers with bachelor’s degrees are five times more likely to be displaced by AI than workers with only a high school diploma. Yea… Researchers calculated the effective hourly rate of current AI models at roughly $0.40 per hour for cognitive tasks. A human software engineer bills at $120. A human financial analyst bills at $80. A human legal researcher bills at $65. It’s a structural repricing of what human cognition is worth in a world where cognition is no longer scarce. Remeber this math… Agricultural mechanization = 100 years Industrialization = 60 years Electrification = 20 years. AI adoption hit 100 million users in two months. Enterprise adoption went from 33% to 71% in a single year. The ladder held for 200 years. This is the first technology that breaks the rung you’re standing on. Act like it. The Bottlenecks That BreakThe evidence shows up across every industry that was built on the assumption that human cognition was expensive and scarce. Start with the software itself. About 46% of all new code is now AI-generated. Junior developer job postings have fallen 67%. Coding bootcamp enrollment dropped 40%, and programs are closing across the country. Microsoft cut 15,000 workers in 2025 while its CEO noted that GitHub Copilot was already writing 30% of the company’s new code. The premise of a $120-an-hour software engineer was that writing code was hard. AI made it easy,, and the human bottleneck opened. Legal services are next. I own a slab of that sector… I am taking a bath now… AI now completes a contract review in 26 seconds... a task that takes a human lawyer 92 minutes. The cost dropped from $6,900 to roughly $3. Balls… My friends and colleagues at Baker McKenzie... one of the largest law firms on earth... cut 600 to 1,000 support positions in February, citing AI efficiencies. The bottleneck was billable hours. The bottleneck opened. India is screwed… Translation is further along than most people realize. An Oxford study sayst 28,000 translator positions were never created between 2010 and 2023 due to machine translation. In finance, AI translation volume surged 700% in a single year while human-only translation fell 47%. Half of all freelance translators are considering leaving the profession. The bottleneck was linguistic fluency. The bottleneck opened. Accounting is following the same trajectory. Across the Big Four, graduate hiring fell 44% year-over-year in 2024. PwC cut 5,600 staff in the first half of 2025. EY has delayed graduate start dates for three straight years. The firms are adopting the “diamond model”... a thinner base of junior workers, a wider middle of specialists who supervise AI outputs. The bottleneck was someone who could read a spreadsheet and produce a compliant filing. The bottleneck widened. I wrote about India’s IT services industry earlier this week... the $283 billion outsourcing empire where entry-level hiring collapsed 50% to 60% in a single year, and the Bank for International Settlements confirmed the gap is widening. The demographic dividend was a cognitive arbitrage, and AI is closing the spread. Every one of these industries was a toll booth on human inefficiency. The entire $300 billion SaaS market... every $50-per-seat subscription for Salesforce, Slack, Workday, Jira... was a tax on the fact that humans are slow, forgetful, and need interfaces to stay organized. AI doesn’t need interfaces. The seat just disappears. We see that per-seat pricing has already dropped from 21% to 15% of SaaS models over the past 12 months. The industry is repricing itself around the possibility that its core customer may not be there much longer. The Avoid ListLet me be direct about what I’d reduce exposure to right now. This is based on insider buying activity and just straight-up common sense.
These are not short recommendations. I’m telling you to stop assuming they’re as safe as they were five years ago. What SurvivesAI closed the cognitive escape hatch. So where do you go? Down… as i’ve said.. into the physical. Last week, I introduced the idea that aging societies must fund certain chokepoints regardless of politics, the Fed, or AI. Have you considered… Healthcare beds. Energy infrastructure. Defense hardware. Regulated utilities. These survive because they require atoms, not bits. An AI can replace a financial analyst. It cannot replace a skilled nursing bed. It cannot replace a natural gas pipeline. It cannot replace a guided missile. And here’s the irony nobody in Silicon Valley wants to confront... radiology, the field that was supposed to be the poster child for AI displacement, is actually hiring more than ever. Projected shortage of 42,000 radiologists by 2033. Average income $520,000, up 48% in a decade. Why? Because the patient is physical. The liability is even more physical. The judgment call that sends someone to surgery is physical. AI assists the radiologist. It doesn’t replace the body in the room. That is the dividing line. For 200 years, the ladder held pretty well. Every time technology destroyed a category of work, the economy created a new one... higher up, more cognitive, more human. Each transition was faster, but the direction was always the same. Up. AI breaks the pattern. The cognitive work that was supposed to be automation-proof is the work most exposed. The ladder didn’t just lose a rung. It lost the top. Not all bottlenecks are chokepoints. The cognitive ones are opening. The physical ones are tightening. Own the ones that are tightening. The Sovereign MoveTonight, I’m going to talk about a stock that I previously mentioned. Tim Melvin and I are now having our Thanksgiving conversations remotely. He’s a deep value and quality investor… I’m just looking for yield, reversion, and what aligns in the liquidity cycle. We’re testing the great academic work of those who came before us… and building conviction on what really matters. I’ve been short KKR since $114 because of the studies we talk about. It’s at $84 now… And since he’s the bank master… I defer to the fact that the bulk of the names appearing on our screen are financial. But then, there is this one name… right in my wheel house. We talked about it tonight… I love the yield. Tim loves the smooth momentum and the higher highs, and higher lows… I love the narrative on AI. Tim said… the price-to-cash flow is 10.3. Could be better, but buy the dip if it comes. Continue reading this post for free in the Substack app
|
Subscribe to:
Post Comments (Atom)
0 Response to "8.5% Yield on a Cheap Energy Stock..."
Post a Comment