John McCormick

The only way to win is to become inevitable.

Mankind has always been at the mercy of the weather. No longer. Rainmaker is the first company in history to routinely, unambiguously, modify the weather. Last quarter, we produced >143MM gallons of unambiguously man-made precipitation. Here’s how we do it 🧵

@SarkaryShahvir Because the humanoid robot can amortize cost of batteries+actuators+sensors+compute across dozens of appliances and use cases rather than duplicating it for each one. It is the same reason humanoids will be a big deal as an autonomous interface to legacy weapons platforms.

My interview with Shawn Ryan is now live We sat down for a 3-hour deep dive into the future of humanoid robots, flying cars, weapon detection systems, and next generation AI interfaces Youtube Video: youtu.be/99pOdGEGu6s

Since ‘22, dozens have written about cost exchange. Fortunately, @PeteHegseth and Feinberg commenced the first meaningful unmanned budgetary pivot in early ‘25, and are demanding high commercial+low cost acceleration to flip our traditional script.

We are in a moment of massive manufacturing need and @IsembardGroup sits right at the intersection of 3 happening simultaneously and quickly: 1/ A massively and quickly increasing demand for fast, flexible, and local manufacturing around aerospace, defense, and robotics. 2/ Increasing labor displacement fueling a resurgence in an SMB economy and individuals' desires to own their own destinies/run their own businesses. 3/ Agentic operating systems allowing for fast set ups, networked management, better results, and demand distribution. @IsembardGroup's franchise model and capital efficient approach is allowing incredibly fast scaling, with 25 factories by the end of this year across the US and Europe. @AFitzgerald1992 and his team are well on the path to delivering components 10x faster and 50% cheaper than current suppliers. We are thrilled to lead their $50M Series A @usv as they create the dominant decentralized manufacturing network.

Today @IsembardGroup announces our $50m Series A. More factories, more engineers, more countries. If not now, when... 🇬🇧🇺🇸🇩🇪🇫🇷🇺🇦

@tmrohan Love this

Chris Sacca being asked about what he’s investing in at Lowercarbon and he sings Rainmaker’s praises. “Augustus is the real deal”

@lacostejonathan @TheSpaceVC @FinnMurphy12 @WizLikeWizard @NWischoff @lucasbagnocvaz 🙋‍♂️

Only 20.2% of new VC firms started in 2021 made it to Fund II SVC (@TheSpaceVC) is one of those firms what other emerging managers from 2021 are on Fund II/III & crushing it?! @FinnMurphy12 @WizLikeWizard @NWischoff @lucasbagnocvaz

@bznotes Different game in liquid markets

Fascinating. Some of my biggest successful investments happened this way.

GDP is a blunt instrument in the incoming AI world... If tech makes cancer cheap to prevent instead of expensive to treat, GDP drops. If robots monetize our chores, it goes up.  We're steering a 21st-century economy with 20th-century gauges!

Introducing Helix 02 It's our most powerful model to date - it's using the whole body to do dishes end-to-end and it's fully autonomous

@wolfejosh Over under for the word “bro” or “brah” on a car full of 10y old boys for 20 mins….is 500+

Made simple Dad app with Claude called UmLikeLike I pick which of my 3 kids is speaking. Hit record. It counts how many "likes" they use in a sentence (and extrapolates "likes"/min)+ tracks trendline over time green for falling, red for rising ;)

@aphysicist Raise your hand if you’ve been to Idra and hit a delicious Autogrill on the way…🙋‍♂️

the same capex dollars in a datacenter that’s obsolete in ~2 years produces ~35x less value than 10 of these machines over ~8 years. would you like to learn more?

@patrick_oshag @AntithesisHQ

Feels like this line (from a great discussion thread) may come to represent the coming era “[human] review capacity is super limited compared to output volume”

here's my holiday mega rant on why every economist who told you american manufacturing was strong should never be taken seriously again: GDP counts "manufacturing output" and economists nod approvingly when the number goes up. but that number tells you almost nothing about whether america can actually make things when it matters. a 62 year old CNC machinist in ohio who can hold ±0.00005" tolerances on titanium and trained three generations of apprentices is about to retire. his knowledge disappears from the economy entirely. it shows up nowhere in the statistics. we treat the erosion of manufacturing human capital the same way a strip mining operation treats topsoil: an externality not worth measuring. the fundamental problem is that economics treats manufacturing as a flow (inputs, outputs, productivity per hour) when the strategic question is actually about stock. what capabilities exist? who holds them? are they regenerating or depleting? it gets worse. the manufacturing output numbers aren't just measuring the wrong thing. they're also fake. susan houseman at the upjohn institute showed that the computer and electronics industry, which is a tiny slice of manufacturing, is responsible for the majority of measured output growth over the past few decades. here's how the trick works: when a computer chip gets twice as powerful at the same price, government statisticians use "hedonic quality adjustments" to count that as doubling real output. we didn't make more chips. we made the same number of chips. but the statistics say production doubled. strip out computers and US manufacturing output has been flat or declining for decades. the "productivity miracle" that justified letting the industrial base erode? it's mostly just moore's law run through a statistical blender. so we're not just measuring the harvest while the soil erodes. we're measuring a fake harvest. the military understands readiness better than economists do. a combat unit reports two scores: S-Rating (do you have the tanks?) and P-Rating (do you have trained crews to fight them?). current economics only measures the S-Rating. it assumes if the machine exists, the labor exists. but if america has 100,000 5-axis CNC machines and only 40,000 qualified programmers, our effective capacity is 40%. the remaining 60% is stranded capital. assets that appear on balance sheets but are functionally scrap metal. we need an entirely different measurement framework. manned industrial capacity: total installed capital multiplied by qualified operator ratio. this tells you what you can actually produce, not what your equipment theoretically allows. proficiency weighted workforce: stop counting "manufacturing jobs" and start counting proficiency units. weight workers by time to proficiency. an operator you can train in two weeks counts differently than a master machinist who took ten years to develop. if a factory replaces 10 retiring masters with 10 new hires, employment data says "no change." reality says capability just crashed. net skill replacement: years of experience retiring minus years of training entering. if 1,000 machinists with 30 years experience retire and we replace them with 1,000 trade school grads with 2 years training, we didn't break even. we lost 28,000 proficiency years. a negative NSR is a recession in capability even if GDP hits all time highs. we are burning the furniture to heat the house. replacement latency index: in a crisis, the question isn't "how cheap" but "how fast can we scale." measure the time required to double skilled output. a deep bench of apprentices ready to step up means three months. no bench means seven years to train new masters from scratch. the US is in a latency trap: high skill needs with no surge capacity. now here's where i push back with realism on the techno optimists who think AI and robotics make this conversation obsolete. they're wrong, but not for the reason most critics think. the real question isn't "will machines replace humans" but "how do we transfer knowledge in both directions." for a century, knowledge transfer in manufacturing meant master to apprentice. watch, practice, learn the feel of the material, develop intuition over decades. this is how tacit knowledge perpetuated itself. that model is breaking down. the masters are retiring faster than apprentices are being trained. the chain is snapping. but here's what most people miss: the knowledge doesn't have to die. it can be captured, encoded, transferred to machines, and then transferred back to a new generation of workers who learn differently than their predecessors did. this is the actual evolution of the trades. not "robots replace machinists" but "the machinist's knowledge gets embedded in systems that make the next generation of machinists more capable faster." AI can capture what the veteran knows about how some weird alloy from a supplier behaves differently on humid days. it can encode the pattern recognition that lets a toolmaker look at witness marks and know what went wrong. it can compress twenty years of trial and error into training data that accelerates the next cohort. but this only works if we do it now, while the knowledge holders are still here to teach both the humans and the machines. this is what we're building at atomic. not automation that replaces human expertise but systems that capture it, augment it, and transfer it. AI that learns from the best toolmakers and then helps newer toolmakers perform at higher levels faster. the human stays in the loop but the loop includes machine intelligence that holds institutional knowledge. the factory of the future isn't lights out. it's a partnership where tacit knowledge flows from humans to machines and back to humans. where a retiring master's expertise doesn't disappear but becomes part of the system that trains their successors. and here's the crucial point: the more powerful these AI tools become, the more valuable deep manufacturing expertise gets. because someone has to teach the machines. someone has to catch the errors when AI optimizes toward a local maximum that looks good mathematically but a human expert would recognize as fragile. someone has to know what questions to ask. atomic is currently training AI on the accumulated knowledge of a generation that is retiring. what happens when that knowledge base stops refreshing? we'll have systems optimizing based on stale understanding, with fewer humans who can catch the drift. so the "robots will save us" crowd has it exactly backward. AI makes the measurement problem more urgent, not less. we need to know: do we have enough humans with deep knowledge to teach the machines? are we capturing expertise fast enough before it retires? are we building systems that transfer knowledge back to new workers? none of this shows up in current economic statistics. while our industrial base was hollowing out, here's what the experts were looking at: GDP said manufacturing output was growing. industrial production index said capacity utilization was healthy. productivity statistics said american workers were more efficient than ever. the numbers looked fine. the consultants wrote reports. the economists gave speeches. then reality hit. COVID came and we couldn't make masks, gowns, ventilators, basic PPE. not because we lacked factories but because we lacked the people and knowledge to spin up production fast. we had to beg other countries for medical supplies. the metrics said manufacturing was fine. we couldn't make cotton swabs. china decoupling revealed we can't make the chips that run everything. not just advanced semiconductors. basic microcontrollers that go in cars and appliances. the metrics said our technology sector was world leading. we can't fabricate the components. ukraine exposed that we can't produce artillery shells fast enough to sustain one ally in one conventional war. 155mm rounds. not stealth fighters. not nuclear submarines. shells. the arsenal of democracy cannot feed a single front. the metrics said our defense industrial base was strong. the crises already happened. we already failed the test. every metric we track said we were fine and we could not make the things we needed. we're measuring a fake harvest while the soil erodes underneath us.

@tkexpress11 One of core themes - but mainly through the lens of infrastructure capital / private credit et al helping deep tech / hardware companies scale production and more.

In 2026, Venture Capital will eat Private Equity It used to be that venture capital and private equity lived on two separate planets: VC = San Francisco PE = New York They targeted completely different universes of companies: --> PE - people heavy biz services, stable/low growth, predictable cashflows --> VC - tech-forward, high growth, high risk, massive TAM What was the playbook for B2B VC backed startups? --> Grow to unicorn scale by selling to other early adopter tech companies, then Fortune 500s XX> SMB and mid-market services - think field services, IT staffing, accounting, construction, recruiting - were always tough to sell into for startups Why? -->Thin margins, high labor costs, and small IT budgets >> But as AI eats labor, these businesses are in play << There are 3 ways where VC and PE are colliding: 1/ Private Equity funds will become channel partners for startups. PE funds are focused on financial engineering and cost optimization. Startups building AI products and services can sell across their portfolio to automate the backoffice and uplevel sales and marketing. PE funds have made AI their #1 strategic priority and have hired central leaders to oversee their portfolio adoption efforts 2/ PE portfolio pages are a startup idea menu Private equity will often buyout vertical software companies whose TAM didn’t allow venture scaled returns. As software evolves from data storage and collaboration to agents taking action and completing work, AI should massively expand the TAM for these categories. Founders will set their sights on unseating these legacy incumbents backed by private equity. All they have to do is look at their portfolio pages for category ideas 3/ AI Rollups This is one of the most direct ways that VC is eating PE VC backed AI platform businesses are not just selling software but acquiring legacy business services companies to own the value chain end to end. As an example, our @speedrun company AgentAstra is acquiring freight forwarding services businesses with mostly debt and integrating AI deeply into their operations These companies aim to increase margins by at least 2x and make them “AI native” tl;dr - While the west coast, Patagonia-wearing VCs and the east coast, PE suits used to live in different universes, in 2026 with AI, I believe, those worlds converge

A simple thesis - AI is going to be writing dramatically more code in the future. - We'll need substantially better testing infrastructure to trust it. - Antithesis are the best in the business. Proud to be a backer!

Jane Street can speak to the technical innovation behind Antithesis much better than I could, but the reason I'm long (literally) is that Will Wilson (the CEO) is one of the most polymathic and interesting people I've ever met.