Start with the problem, because it’s the reason this company exists. PFAS, the “forever chemicals,” are a family of compounds engineered to resist heat, water, and grease. That’s what makes them useful in semiconductors, aerospace, and pharmaceuticals, and it’s also what makes them nearly impossible to break down. They sit in drinking water, industrial wastewater, groundwater plumes, and firefighting-foam sites, and they don’t leave on their own.

The standard playbook is to capture them with filters, concentrate them, and then ship the concentrate to a landfill or an incinerator. You haven’t destroyed anything. You’ve moved the problem and kept the liability.

Claros Technologies is built around a different verb. It destroys PFAS. Its ClarosTechUV system breaks the carbon-fluorine bonds that make these compounds so durable, and the company reports greater than 99.99% destruction across long, short, and ultra-short chain PFAS in field work. Ultra-short chains are the hardest to touch, so that claim matters if it holds up independently. Alongside the destruction hardware, Claros runs ClarosLabs, an accredited analytical division that detects and quantifies PFAS. You get the diagnosis and the cure from the same company.

On July 28, 2026, Claros announced a $55 million Series B, which the company says is inclusive of a previously announced convertible note. Read that carefully: the headline number isn’t all newly priced equity. Treehouse Family Capital led the round, with participation from Daikin America, Veralto, the Bush Foundation, Nord Asset Management, and the Saint Paul & Minnesota Foundation. No valuation was disclosed. For context on trajectory, the company raised $22 million in 2024 and a $10 million convertible note in 2025, and Twin Cities Business reported roughly $39.6 million raised as of December 2025. This is its largest financing to date.

The company was founded in 2018 as a University of Minnesota spinout. CEO and co-founder Michelle Bellanca spent 24 years at 3M, including running its corporate venture arm in Asia, with earlier stops at IBM and Japan’s Ministry of International Trade and Industry. Co-founder and CTO Abdennour Abbas is the university scientist behind the core chemistry, and John Brockgreitens rounds out the founding team on product. A researcher who knows the science, an operator who knows how large industrials actually buy, and a product lead. That’s a useful combination for selling into slow, regulated buyers.

The Market

The incumbents here are granular activated carbon, ion-exchange resins, and membrane filtration, sold by large water players. They work, and they’ve earned their place. But they share one structural feature: they capture and concentrate PFAS rather than eliminate it, which leaves a spent-media disposal problem behind. And that disposal step is turning into its own liability as bans on incomplete incineration spread. When your solution creates a second problem the regulator is about to outlaw, the door opens for someone selling permanence.

The numbers behind the door are real, though you should treat every forecast as a third-party estimate. Grand View Research valued the PFAS treatment market at $2.8 billion in 2025 and projects $4.8 billion by 2033, with the destruction segment growing fastest at a 10.1% CAGR. Future Market Insights sizes the PFAS concentration and destruction systems market at $2.7 billion in 2025, reaching roughly $8.0 billion by 2036. Persistence Market Research notes the U.S. Department of Defense has identified more than 700 contaminated military sites with cleanup costs expected to exceed $3 billion over the decade. And the 3M settlement, widely reported at $10.3 billion, reset how every chemical, military, and airport operator prices this risk.

Timing is the part that’s hard to argue with. The EPA has set a 4 parts-per-trillion limit for PFOA and PFOS in drinking water and put close to $1 billion behind state programs plus $945 million for drinking-water projects. The EU’s Drinking Water Directive limits took effect in January 2026, and broader restrictions are under discussion in Europe. Insurers, meanwhile, have started excluding PFAS claims outright, which forces companies to fund remediation from their own budgets. When the law tightens and the insurance disappears, “manage it later” stops being an option.

Read more

Corgi has reportedly raised money three separate times since early May. Nobody outside the cap table has seen a loss ratio.

Here’s the sequence, per Forbes and TechCrunch. In January, the Y Combinator alum raised a $108 million Series A at an undisclosed valuation (PitchBook estimates roughly $630 million post-money). In early May, a $160 million Series B led by TCV set the price at $1.3 billion. Three weeks later, an extension from those same investors added $106 million at $2.6 billion. And on July 22, Forbes reported, citing multiple people familiar with the matter, that a second extension has closed at a $4 billion valuation. That’s a reported 54 percent markup in about eight weeks.

Important caveat: the $4 billion figure and the round itself come from unnamed sources, not from a company announcement. Same for the revenue trajectory. When Corgi announced its Series A, the founders said the company had hit $40 million in annualized run rate. Sources told Forbes it’s now tracking toward $450 million by year end. That’s a forward-looking projection from people who aren’t named, not an audited result.

Corgi sells business liability coverage to startups (general, cyber, tech and AI liability), and it also sells data room software and runs coffee shops. Deel and Artisan are named customers.

The model: Narrative Fallacy.

Narrative fallacy is our habit of wrapping a clean story around a set of facts that would support several stories equally well. The story everyone’s telling about Corgi is momentum. Three raises in eight weeks, therefore red hot, therefore the market has lost its mind. Maybe. But insurance is a balance sheet business. When you write more premium, regulators and reinsurers require you to hold more capital against it. Grow the book fast enough and you have to keep feeding the balance sheet, whether or not anyone’s excited about you. Same facts, completely different story.

The contrarian close.

The valuation is the least informative number in this whole thing. In insurance, you can post spectacular growth and terrible underwriting at the same time, and you won’t find out which one you had for two or three years, because claims show up long after premiums do. Written premium tells you how fast you’re selling. Loss ratio tells you whether you should have. Ten straight up-rounds don’t change that, and neither does a $4 billion headline.

Watch what Corgi discloses next. If the next milestone they publicize is another valuation instead of a combined ratio, you’ve learned something.

——————————————————————————————————————

If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, you can grab my book, Mental Models: How to Think, Act, and Win, right now.

Get it Here

——————————————————————————————————————

Next week’s deep-dive:

——————————————————————————————————————

Prime Intellect, founded in 2024 and led by Johannes Hagemann, raised a $130 million Series A at a $1 billion valuation, led by Radical Ventures, with NVIDIA Ventures, Intel Capital, and Dell Technologies Capital all in, alongside a roster of angels that includes the founders of Perplexity, Box, Harvey, and Cognition. Total funding now tops $150 million. The company runs what it calls an “open superintelligence stack”: open-source, distributed infrastructure that lets a company train and own its own AI models instead of leaning entirely on a frontier lab. About 6,000 customers, including Ramp and Zapier, and roughly $100 million in annualized revenue, per the company.

The proof point is the best part. Ramp used Prime Intellect to train a small reinforcement-learning agent to find answers inside spreadsheets. Ramp says it beat the frontier models on accuracy, ran faster, and cost a fraction as much. That’s the whole thesis in one story: for your specific task, a small model you own can beat a giant one you rent.

The mental model here is Jevons Paradox. In 1865, the economist William Stanley Jevons noticed something odd. As steam engines got more efficient and coal got cheaper to use, total coal consumption went up, not down. Cheaper access expanded demand faster than efficiency shrank it. The same thing is about to happen to model training. Prime Intellect makes training your own model dramatically cheaper and more accessible. The instinct is that this threatens the compute giants. The opposite is true. When training gets cheap enough for 6,000 companies instead of five labs, total compute demand explodes. That’s exactly why NVIDIA and Dell wrote checks. They don’t care whether OpenAI or Ramp trains the model. They sell shovels to both.

Here’s my take. Everyone’s framing this as decentralization versus the frontier labs, David against Goliath. That’s the wrong lens. Prime Intellect isn’t shrinking the pie. It’s multiplying the number of people baking. I watched a version of this at Robinhood: when you take something gatekept and make it suddenly accessible, you don’t just redistribute the demand that exists, you manufacture a wave of new demand that didn’t exist before. So the contrarian read isn’t “the labs are in trouble.” It’s that the biggest winners of democratized AI training might be the exact infrastructure players people assume it threatens. Cheap doesn’t mean less. Cheap means more.

——————————————————————————————————————

If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, you can grab my book, Mental Models: How to Think, Act, and Win, right now.

Get it Here

——————————————————————————————————————

Next week’s deep-dive:

——————————————————————————————————————

Auger, the Bellevue supply-chain startup Clark founded in 2024, closed a $50 million Series B led by Eclipse, with existing backer Oak HC/FT in again. That brings total funding to $150 million in under two years, for a company of about 130 people. Clark isn’t a typical founder. He spent 23 years at Amazon, most recently running its worldwide consumer business, then a stint as CEO of Flexport. Customers already include Meta’s Reality Labs, Fanatics, and Kimberly-Clark. At Fanatics, Clark says roughly 85% of supply-chain decisions now run automatically through Auger.

Here’s what it actually does. Auger doesn’t try to be your system of record. It sits on top of the ERP, warehouse, transportation, and planning systems you already run, unifies the data, and turns AI agents loose to make the thousands of small operational calls on their own. Clark also raised early, before he needed the money, to avoid fundraising during a busy fall of onboarding, at a valuation he says is roughly double the Series A. For context, PitchBook and the NVCA pegged first-half 2026 venture activity at $412.7 billion, with 86% going to AI. Raising into that tape was a choice, not a necessity.

The mental model here is Path Dependence. Once a company builds its operations around SAP or Oracle, it’s stuck, and not because the software is beloved. It’s because everything else, every process, integration, and trained employee, is wired to it. Ripping it out means re-wiring the whole company. That’s path dependence, and it’s why “we’ll replace your ERP” has killed more startups than it’s built. Clark learned that lesson at Amazon scale, so Auger doesn’t fight the lock-in. It rides on top of it. The old system stays. Auger just makes the decisions the old system never could.

Here’s my take. The loud AI bet right now is replacement: rip out the legacy thing, drop in an agent. The quiet, better bet is the layer on top: assume the incumbent system is immovable and sell the thing that acts on it. Replacement founders burn years fighting switching costs. Clark is using those switching costs as a moat, because the same lock-in that keeps SAP in place will keep Auger in place once it’s the one making the calls. The unglamorous truth of enterprise AI is that the winners won’t be the ones promising to burn down the old stack. They’ll be the ones who make peace with it and quietly take the wheel.

Next week on Startup Spotlight

Every company above made noise this week. Next Wednesday’s paid deep-dive goes the other direction: the biggest AI infrastructure bet of the moment might be the one that never touches the cloud. I’m breaking down the on-prem AI shift, who’s quietly winning it, and why “private by default” is about to matter far more than it sounds. Paid subscribers get the full teardown.

Share

——————————————————————————————————————

If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, you can grab my book, Mental Models: How to Think, Act, and Win, right now.

Get it Here

——————————————————————————————————————

Next week’s deep-dive:

——————————————————————————————————————

SambaNova completed the first close of a $1 billion Series F at an $11 billion post-money valuation, according to the company’s announcement, with General Atlantic leading and Seligman Ventures, T. Rowe Price, and Capital Group joining. Reuters and CNBC reported the same figures. The nine-year-old San Jose company builds inference chips, the hardware that runs trained AI models rather than training them, and it’s pitching itself as an alternative to the usual GPU stack.

Big number. Familiar story. Money chasing anyone who isn’t Nvidia.

Here’s the line most people skimmed past. On the same day, SambaNova said JPMorgan Chase had selected it as an inference infrastructure partner, deploying its systems for secure, on-premises AI inside the bank’s own walls, per CNBC and Reuters. Not in someone else’s cloud. On-prem. Inside the firewall.

Run that through second-order thinking, one of the models from my book. First-order thinking stops at the raise: investors want a Nvidia alternative, so a billion dollars showed up. Fine. Second-order thinking asks what the raise is a symptom of, and what happens next. The answer is sitting in the JPMorgan line. One of the most compliance-bound institutions on earth just decided its most sensitive AI workloads shouldn’t leave the building.

Subscribe now

That’s the signal. Not the check size. The buyer’s choice.

When a bank pulls inference in-house, it’s telling you something about where regulated AI is heading. Control, auditability, and data residency stop being nice-to-haves and start being the spec. The company that wins that buyer isn’t the one with the flashiest benchmark. It’s the one that fits inside the constraints regulators already impose.

I’ve watched this from the inside. At /mkt we build in regulated markets, and the lesson repeats: compliance isn’t a layer you bolt on after the product works. It shapes the architecture from the first line. The teams that treat it as a design input, not a tax, are the ones institutional buyers can actually say yes to.

So here’s the contrarian read. Everyone’s counting the billions flowing into Nvidia challengers. The more durable question is who controls the compute once the model is trained, and where that compute physically sits. If the biggest, most regulated buyers keep choosing on-prem, the winners won’t be decided by raw speed. They’ll be decided by trust. And trust is a slower, stickier moat than any chip spec.

The billion dollars is the headline. Where JPMorgan wants its AI to live is the story.

If you want the mental models behind breakdowns like this, my book, Mental Models: How to Think, Act, and Win, is on Amazon now.

Get it Here

Startup Spotlight is for informational and educational purposes only. It is not investment advice, an offer, or a solicitation to buy or sell any security. Company metrics are self-reported, and funding details are drawn from public reporting and company statements. Figures may change.

What Novig Actually Is

There’s a number every sports bettor knows and most don’t think about: the vig. It’s the commission a sportsbook takes on every bet, baked into the odds before you ever place a wager. On a coin flip, you’d expect 50/50. A sportsbook offers you something closer to 47/53 that gap is the vig. On most platforms, it runs between 5% and 10%. Over time, it makes it nearly impossible for recreational bettors to be profitable. The house doesn’t need to be right. It just needs to be in the middle.

Novig was founded to remove the middle.

The New York-based company operates a commission-free, peer-to-peer sports prediction market using an order-book model. Rather than betting against a house that sets fixed odds, Novig users trade contracts against each other at market-driven prices. There’s no vig baked into the spread, no house edge, and no limit on winning players a practice traditional sportsbooks use to cut off anyone who gets too good. The name is a play on the term itself: no vig.

<cite index=”103-1”>On February 18, 2026, Novig announced the close of a $75 million Series B round led by Pantera Capital, with participation from Multicoin Capital, Makers Fund, Edge Equity, and existing investors Forerunner Ventures, Perceptive Ventures, and NFX. The round brings total capital raised to more than $105 million and followed a period of explosive growth: a 10x increase in trading volume during 2025.</cite>

Then, on June 17, 2026, Novig received CFTC approval as a designated contract market one of the fastest DCM designations in CFTC history, completed in less than seven months. That approval gives the company the same federal regulatory designation as Kalshi, the current market leader, and clears a path to operate sports prediction markets in all 50 states under a single national compliance framework.

The founding team: <cite index=”104-1”>Co-founders Jacob Fortinsky (CEO) and Kelechi Ukah (CTO) built Novig from the ground up as a sports-first exchange.</cite> Fortinsky came to prediction markets intellectually tracking the 2016 election and Brexit referendum in high school and studying how markets aggregate information better than pundits. Ukah brings the engineering depth to build the exchange infrastructure. They’re not gambling executives who stumbled into prediction markets. They built the product from first principles: what would a fair sports market look like if you designed it like a financial exchange?

<cite index=”103-1”>Novig’s annualized trading volume exceeds $4 billion, with the company reporting more than $5 billion in cumulative volume.</cite> Monthly volume recently crossed $300 million. The platform currently operates in 42 states.

Subscribe now

The Market : Why Sports Is the Right Prediction Market Wedge

The U.S. online sports betting industry is one of the fastest-growing consumer financial markets in the world. <cite index=”104-1”>DraftKings’ handle alone topped $53 billion during fiscal year 2025.</cite> The total U.S. sports betting market is estimated at $150-200 billion in annual handle and growing rapidly as more states legalize and digital adoption accelerates.

Here’s the structural fact that makes Novig’s timing compelling: sports already dominates prediction market trading, even on platforms that weren’t built for it. <cite index=”111-1”>By some estimates, sports accounts for between 80% and 90% of all trading across prediction market exchanges.</cite> Kalshi, which started with political and economic contracts, now derives the majority of its volume from sports. Polymarket, built for crypto and election markets, sees the same pattern. Sports is the demand. But the dominant prediction market infrastructure wasn’t built with sports traders in mind.

That’s the gap Novig is filling.

The macro environment is also moving in Novig’s direction. A 2024 court ruling gave Kalshi the green light to offer sports event contracts under CFTC oversight, which triggered a wave of new entrants applying for DCM status. <cite index=”104-1”>The stocks of FanDuel parent Flutter Entertainment and DraftKings have both fallen by more than 50% in the last six months</cite> as the market reprices the competitive threat from federally regulated prediction markets.

Traditional sportsbooks built their moats on state-by-state licensing, brand recognition, and distribution. A federally regulated prediction market exchange needs one license the CFTC DCM designation to operate nationally. That structural difference changes the competitive dynamics entirely.

There’s also a customer quality argument. Sportsbooks are optimized to extract money from recreational bettors and kick out anyone who gets good. Prediction markets are optimized for price discovery. <cite index=”103-1”>Novig reports that its users are 10 times more likely to be profitable than users on traditional sportsbooks, with 23% of Novig users achieving profitability compared to roughly 2% on conventional platforms.</cite> These are company-reported figures not independently verified but they reflect a fundamentally different market structure: one where serious traders are welcome, not banned.

Read more

Private markets are famously opaque. A pension fund with $50 billion in private equity commitments spread across 200 funds and thousands of underlying companies has to track all of it valuations, cash flows, capital calls, performance benchmarks, and regulatory reporting across investments that don’t trade on an exchange, don’t publish quarterly results, and don’t update in real time. For decades, the industry’s answer to that problem was spreadsheets, PDFs, and a lot of analysts.

Chronograph’s answer is a cloud-based portfolio monitoring, valuations, and analytics platform that serves both sides of the private capital relationship: the limited partners (LPs) who allocate capital into funds, and the general partners (GPs) who manage those funds and report back to their investors. The platform ingests data from both sides, structures it into a consistent system of record, and gives institutions the infrastructure to run reporting, performance analytics, valuations, and increasingly AI-powered queries across their entire private markets portfolio.

On June 16, 2026, Chronograph announced a $140 million+ minority growth equity investment from Sixth Street Growth, the dedicated growth investing platform of Sixth Street a firm with over $130 billion in assets under management and committed capital. All existing investors, Summit Partners, Carlyle AlpInvest, Nasdaq Ventures, and Sidekick Partners maintained their positions. Michael Bauer and Alex Goodman of Sixth Street Growth joined the board. Total funding now stands at approximately $160 million since the company’s 2016 founding. Valuation was not disclosed.

Read more

Most people start a sports league by copying the one above them. The Rabil brothers did the opposite, and just raised $100 million for it.

On June 30, the Premier Lacrosse League closed a $100M Series E led by Ares, a $644 billion asset manager, and Joe Tsai, the Brooklyn Nets owner who’s backed the league since 2019. The PLL and Forbes both call it the largest capital raise in the history of professional lacrosse. ESPN added a minority stake, and the cap table now includes Glen Powell, Rob McElhenney, and David Blitzer’s family office. The league has pulled in more than $200 million since a roughly $3 million seed in 2018. The post-round valuation wasn’t disclosed.

Here’s the part that actually matters. The PLL doesn’t run independent franchises. It’s a single wholly-owned entity that owns all eight teams. Players sign with the league, not a club. Games tour the country instead of anchoring to home cities. When the Rabils started in 2018, the incumbent pro league, Major League Lacrosse, ran the traditional franchise model, and it fought for two decades before the PLL absorbed it in 2020.

The mental model: First Principles.

Most founders reason by analogy. The NFL has franchises, so we’ll have franchises. First-principles reasoning throws out the template and asks what’s actually true on the ground. For a young sport with a thin talent pool and no established local markets, independent franchises spread the product too thin and give away control before there’s anything worth controlling. So the Rabils built the inverse: one entity, full command of schedule, media, and talent, and a tour model that puts the best players on the same field every single week. They didn’t inherit the answer. They rederived it.

It’s the same instinct behind building in regulated markets. At /mkt, the athlete-ownership structure isn’t lifted from someone else’s playbook. It’s built up from what the rules actually permit and what aligns athletes with the value they help create. When the ground is new, first principles beats “how it’s always been done” almost every time.

Now here’s the twist. With the model working, the PLL is openly weighing whether to sell individual teams, the exact franchise structure it rejected, using SailGP as the template. That’s not a contradiction. That’s the whole lesson.

First principles isn’t a religion. You build from the ground up when the map doesn’t exist, and you adopt the conventions once they start adding leverage instead of stealing it. The mistake was never the franchise model. The mistake would’ve been running it on day one because everyone else did. Get the structure right for where you actually are, not where the incumbents ended up.

That’s the quiet reason smart capital showed up. Not because lacrosse is having a moment, but because the people running it think from scratch.

Disclaimer: This post is for informational and educational purposes only. It is not investment, financial, legal, or tax advice, and it’s not a recommendation or solicitation to buy, sell, or trade any security, contract, or instrument, or to participate in any prediction market. Valuations, volumes, and funding figures reflect public reporting and third-party estimates that vary by source and change over time; past performance and private valuations are not indicative of future results. Regulatory status referenced here is described generally and is subject to ongoing litigation and change. /mkt is referenced only to illustrate how regulated-market infrastructure is built, and nothing here is an offer or solicitation with respect to any security.

If you want the mental models behind breakdowns like this, my book, Mental Models: How to Think, Act, and Win, is on Amazon now.

Get it Here

Startup Spotlight is for informational and educational purposes only. It is not investment advice, an offer, or a solicitation to buy or sell any security. Company metrics are self-reported, and funding details are drawn from public reporting and company statements. Figures may change.

On July 1, Together AI closed an $800 million Series C at an $8.3 billion post-money valuation, led by Aramco Ventures, the venture arm of Saudi Aramco, with Nvidia, Vista Equity Partners, General Catalyst, and Emergence Capital joining. Per TechCrunch, that’s up from a $3.3 billion valuation just 16 months ago. The company says annual bookings crossed $1.15 billion last quarter and that usage of open models on its platform tripled over the past year. Those last two are company-reported, so hold them a little loosely.

Here’s the business in plain English. Together AI is an “AI neocloud.” It rents out GPU clusters, and more importantly, it layers its own software on top that it says cuts the cost of running popular open-weight models like DeepSeek and Nemotron by up to 80%. Customers like Cursor and Cognition run those open models instead of pricey closed APIs to protect their margins. Together sells the cheaper road.

Now apply Commoditize Your Complement, one of the models in my book.

A business wins when the thing sitting right next to its product gets cheap and abundant. Microsoft wanted PC hardware dirt cheap because it sold the operating system on top. The cheaper the complement, the more demand pools into you. Together AI’s complement is the model itself. Every time an open-weight model gets good enough and nearly free, the valuable question shifts from “which model?” to “where do I run it cheaply?” Together isn’t betting OpenAI or Anthropic loses. It’s betting the model layer keeps commoditizing, and that the neutral place to run any of them is where the durable margin sits. The pick-and-shovel version is “in a gold rush, sell shovels.” The sharper version is “make sure the gold keeps getting cheaper, so everyone needs more shovels.”

My take? Read the cap table. Aramco leading this has almost nothing to do with oil and everything to do with a sovereign fund deciding that compute and energy, not any single model, are where value sits for the next decade. Nvidia’s in because every workload Together runs, whatever model’s on top, still runs on Nvidia chips underneath. When the strategics quietly agree the money is under the models, that’s worth noticing.

The loudest question in AI is “which model wins?” For builders, that’s the wrong question. Ask “what stays valuable no matter which model wins?” The answer is usually the infrastructure and the rails. It’s the same instinct behind what we build at /mkt: the durable spot in any market is the regulated infrastructure everyone has to route through, not the one asset riding on top.

One honest caveat: $8.3 billion prices in a lot going right. The hyperscalers and Nvidia itself can lean on independent neoclouds, and cheap capacity can tip into a brutal price war. Selling shovels only works while the digging still pays. That’s the risk to watch.

None of this is investment advice.

Disclaimer: This post is for informational and educational purposes only. It is not investment, financial, legal, or tax advice, and it’s not a recommendation or solicitation to buy, sell, or trade any security, contract, or instrument, or to participate in any prediction market. Valuations, volumes, and funding figures reflect public reporting and third-party estimates that vary by source and change over time; past performance and private valuations are not indicative of future results. Regulatory status referenced here is described generally and is subject to ongoing litigation and change. /mkt is referenced only to illustrate how regulated-market infrastructure is built, and nothing here is an offer or solicitation with respect to any security.

If you want the mental models behind breakdowns like this, my book, Mental Models: How to Think, Act, and Win, is on Amazon now.

Get it Here

Startup Spotlight is for informational and educational purposes only. It is not investment advice, an offer, or a solicitation to buy or sell any security. Company metrics are self-reported, and funding details are drawn from public reporting and company statements. Figures may change.

Meta has more than 3 billion users and just decided not to point a single one of them at real money.

Per NPR’s reporting this week, Zuckerberg tried to buy Kalshi before building his own thing. When the talks stalled, Meta started work on its own prediction market app. Here’s the tell: Meta’s version uses play money, not real cash. Its AI writes the questions and settles who’s right. The most powerful distribution machine ever built walked up to a booming market and left out the part that actually prints revenue.

And it is booming. Per The Block, monthly volume across Kalshi and Polymarket went from roughly $28 billion in June 2025 to nearly $220 billion a year later, most of it sports. Kalshi was valued at $22 billion in May, up from about $2 billion a year earlier, and the Financial Times reports it’s now in talks at around $40 billion, possibly closing in Q3. Polymarket sits near $10.7 billion per PitchBook and has been chasing a round closer to $15 billion. These are reported figures, and private valuations move fast, so treat them as snapshots, not gospel.

Now apply Counter-Positioning, one of the models I break down in my book.

Counter-positioning is when a challenger picks a business model the giant can’t copy without hurting itself. Everyone assumes it works in the upstart’s favor. Flip it here. Kalshi runs as a federally regulated exchange under the CFTC. That status is expensive and comes with lawsuits, surveillance duties, and a running fight with state regulators. But it’s also the exact posture Meta won’t take, because putting real-money, gambling-adjacent contracts next to its ad business invites a regulatory mess it has every reason to avoid. So Meta counter-positioned itself into play money. The regulated real-money model isn’t the risky part Meta’s dodging. It’s the moat Meta can’t step into.

Here’s my read, and it’s the opposite of the headlines. People see Meta entering and think Kalshi’s in trouble. Read the product spec instead. When the biggest platform on earth shows up and deliberately leaves the money out, it’s telling you the license is the hard part, not the interface. Anyone can build the buttons. Almost no one can carry the regulatory weight.

I built prediction markets at Robinhood, so I’ve watched up close how the regulatory perimeter, not the UI, decides who gets to offer what. It’s the same reason we build athlete offerings at /mkt under Reg A+ with tZERO as the trading infrastructure. The compliance architecture is the barrier a competitor can’t clone over a weekend.

One honest caveat: this doesn’t make Kalshi a lock. A new Illinois law treating sports-tied contracts as wagers takes effect today, Kalshi has sued, and the broader state-versus-federal fight is still live. Tighten the perimeter enough and it squeezes the regulated players too. The moat cuts both ways. That’s the risk to watch.

None of this is investment advice.

Disclaimer: This post is for informational and educational purposes only. It is not investment, financial, legal, or tax advice, and it’s not a recommendation or solicitation to buy, sell, or trade any security, contract, or instrument, or to participate in any prediction market. Valuations, volumes, and funding figures reflect public reporting and third-party estimates that vary by source and change over time; past performance and private valuations are not indicative of future results. Regulatory status referenced here is described generally and is subject to ongoing litigation and change. /mkt is referenced only to illustrate how regulated-market infrastructure is built, and nothing here is an offer or solicitation with respect to any security.

If you want the mental models behind breakdowns like this, my book, Mental Models: How to Think, Act, and Win, is on Amazon now.

Get it Here

Startup Spotlight is for informational and educational purposes only. It is not investment advice, an offer, or a solicitation to buy or sell any security. Company metrics are self-reported, and funding details are drawn from public reporting and company statements. Figures may change.

Your biggest client didn’t reply to your email. They texted you on WhatsApp at 9pm, and that’s where the deal actually moved.

That side door, the one every compliance officer pretends doesn’t exist, is now a nine-figure business. On June 30, LeapXpert announced a $180 million growth investment led by Riverwood Capital, with existing backer Portage Ventures joining. Per the company’s public disclosures, that brings total reported funding to roughly $236 million. Valuation wasn’t disclosed.

LeapXpert doesn’t sell a chat app. It sits between the consumer channels your clients already use (WhatsApp, iMessage, Signal, WeChat) and the regulated firms that legally can’t let those conversations vanish. Financial services were the first market. Per the company, government is now its fastest-growing segment. Public enforcement actions over off-channel messaging have run into the billions across the industry, so the demand here isn’t theoretical. It’s a line item.

Here’s where it gets interesting. Apply Second-Order Thinking, one of the models I break down in my book.

First-order move: capture the messages so you don’t get fined. That’s a cost. Everybody sees it.

Second-order move: once every governed conversation is captured and structured, you’re sitting on one of the largest untapped datasets your company owns. The archive you built to stay out of trouble quietly became your most AI-ready asset. That’s the real bet Riverwood is making. Not “messaging compliance is big.” Rather, “the compliance byproduct is the prize.”

Most people stop at the first order. The winners price in the second.

My take? Watch where the growth is coming from. Reporting around the round points to rising demand from prediction-market and sports-finance compliance. I built prediction markets at Robinhood, so this one lands for me. Every new market that touches money creates fresh surfaces a regulator wants governed. That’s not friction to engineer away. It’s the toll you pay to operate, and the toll is the moat.

I see the same structure every day at /mkt, where we build athlete offerings under Reg A+ with tZERO as the trading infrastructure. In regulated markets, the compliance layer isn’t the tax on the business. It’s the part a competitor can’t copy over a weekend. LeapXpert just raised $180 million to prove that at scale.

None of this is investment advice.

If you want the mental models behind breakdowns like this, my book, Mental Models: How to Think, Act, and Win, is on Amazon now.

Get it Here

Startup Spotlight is for informational and educational purposes only. It is not investment advice, an offer, or a solicitation to buy or sell any security. Company metrics are self-reported, and funding details are drawn from public reporting and company statements. Figures may change.

Let’s break down the company, the market, and what it means for you.

Company Overview

In plain English, Taktile builds the control room for AI-driven decisions inside banks and insurers. When a lender approves a loan, a compliance team clears a customer through KYC and AML checks, or an insurer pays out a claim, that’s a high-stakes decision with money and regulators attached. Taktile lets those firms build and run those decisions using a mix of AI agents, business rules, live data, and human oversight, all inside a system that a head of credit or a fraud officer can actually see and control. The founders call it an agentic decision platform. Think of it as the operating system that turns a frontier model into a dedicated, supervised worker for a bank’s most sensitive tasks.

Read more

That’s the headline that should make you stop scrolling. Stark Defence, founded in 2024, just closed a €500 million round (about $570 million) led by Sequoia Capital and Peter Thiel’s Founders Fund, with the NATO Innovation Fund, Project A, Air Street Capital, 201 Ventures and Döpfner Capital joining. Reported valuations differ by outlet, but every source puts it north of €3 billion, roughly triple where the company sat back in February. Total raised now runs around €640 million, up from roughly €140 million before this round.

Here’s the part the headlines skip. According to Financial Times reporting, Stark’s Virtus drones failed to hit a single target across four attempts in trials with British and German forces last November. Germany approved a €269 million Bundeswehr contract anyway, with options stretching toward €2.86 billion. More than 80% of the new money is earmarked for manufacturing and R&D, not go-to-market. And Stark is now on the clock to equip a full German brigade in Lithuania by the end of 2026.

So why would two of the sharpest funds on earth write that check?

Subscribe now

This is where one of the models from my book earns its keep: Second-Order Thinking.

First-order thinking reads the trial result and stops. The drone missed, so it’s a bad product, so it’s a bad bet. Clean, fast, wrong.

Second-order thinking asks what happens next, and then what happens after that. The thing being priced here isn’t this quarter’s hit rate. It’s European strike-drone manufacturing capacity during a rearmament cycle, sitting on top of a procurement position inside contracts that are already signed. In a war of attrition, a good-enough system you can build by the thousand beats a perfect system you can’t ship. Capacity is the scarce asset. Stark is being financed like a factory, because a factory is what the buyers actually need.

I’ve spent enough time around regulated, infrastructure-heavy markets to recognize the shape of this. At /mkt, we build inside systems where the demo is the easy part and compliant execution at scale is the hard part. Defense procurement rhymes with that: signed contracts, certifications, and hitting a delivery deadline carry more weight than a single flashy test.

Now the contrarian flag, and I’ll keep it conservative, because I’m not here to tell anyone what to buy. When capital races to “be in the room,” price can run out ahead of proof. A signed contract isn’t a battlefield result, and €3 billion-plus is a lot of expectation to load onto a two-year-old company that hasn’t yet shown it can manufacture at the volumes it’s promising. The second-order logic is sound. The valuation is where the risk actually lives.

Watch the factories, not the press releases. The next eighteen months are a single question: can Stark build what it already sold?

If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, you can grab my book, Mental Models: How to Think, Act, and Win, right now.

Get it Here

Startup Spotlight is for informational and educational purposes only. It is not investment advice, an offer, or a solicitation to buy or sell any security. Company metrics are self-reported, and funding details are drawn from public reporting and company statements. Figures may change.

Subscribe now

That single decision is why this company is worth your attention this week.

Company Overview

In plain English: Adaptive Innovations is a home health care provider that runs on its own AI operating system. When a hospital discharges an elderly patient who needs care at home, a traditional agency handles the intake, scheduling, clinical documentation, and compliance billing through a tangle of legacy software and human coordinators. Adaptive automates that back office with AI and sends its own clinicians into the home. It isn’t selling the tooling to someone else. It’s the licensed provider getting paid to deliver the care.

The company is at the Series A stage. It came out of stealth in early June with $60 million raised to date: a $50 million Series A led by Felicis, plus a previously undisclosed $10 million seed round that Bain Capital Ventures led. The Series A drew in Bain Capital Ventures, Optum Ventures, Sunflower Capital, Conviction, BoxGroup, SV Angel, Dorm Room Fund, and Constellation, along with angels from healthcare services and frontier AI labs.

The founding team runs deep on both sides of the problem. Co-CEOs Alex Wendland and Logan Stinson lead alongside CTO Ryan Tolsma and COO Hunter Stinson, working out of New York City and Dallas. The company says its bench includes engineers from Palantir and Jane Street and operators from McKinsey, Harvard, Stanford, and the U.S. Army Rangers. That’s an unusual mix: people who can ship hard software sitting next to people who’ve actually run operations under pressure.

What have they built so far? According to the company, 200 clinicians have completed more than 100,000 home visits in 18 months. Adaptive reports a rehospitalization rate below 5% against an industry average around 11%, and says its AI workflows have cut clinician documentation time by roughly 80%. Treat those as company-reported figures, because they are. They’re still the kind of early numbers that make sharp investors lean in.

Read more

On June 23, Cadence, the clinical AI company managing chronic care for older adults, announced a $100 million Series C led by Spark Capital. The syndicate is worth reading closely. Thrive Capital, General Catalyst, Coatue, and B Capital came in alongside three health-system investors: Corewell Health Ventures, Memorial Hermann, and Duke Health. Cadence also announced a new affiliation with Duke Health the same day.

The product fits the moment. Cadence supports more than 90,000 patients across 20-plus health systems, using AI to handle the high-volume routine work of chronic care, monitoring vitals and surfacing risk, so clinicians spend their time on the decisions that need a human.

Here’s the part the headline skips. Medicare’s innovation center rolled out an experimental program called ACCESS that pays providers a set amount per patient, roughly $180 to $420 in year one depending on the condition, and pays the full amount only if patients hit outcome targets. Compare that to a standard remote-monitoring setup, where a provider can bill around $100 per patient per month for doing the work regardless of results. Cadence said it would be among the first to sign up.

This is where I’d reach for Skin in the Game, one of the models in my book. The idea is simple: you carry real downside, not just upside, and your incentives are wired to the result instead of the activity. ACCESS flips healthcare’s default. The old model pays for motion. ACCESS pays for outcomes. Cadence is volunteering for a structure where it eats the risk if patients don’t get better.

That also explains the cap table. When three health systems write checks into the company they’re buying from, they’ve got skin in the game in its success too. That’s not a logo grab. That’s alignment.

The contrarian read: the $100 million is the least interesting number in this story. The interesting numbers are $180 to $420 per patient paid on results, against $100 a month paid no matter what. Cadence is betting it can use AI to make the outcomes math work where labor-heavy competitors can’t. If patients don’t improve, the model doesn’t pay. Most companies run from a constraint like that. The good ones build toward it.

I think about this at /mkt, where we operate inside a Reg A+ structure with tZERO as our trading infrastructure. In regulated markets you don’t write the rules, you build to fit them. The structures that force you to win only when your users win can look like constraints. They usually turn into moats.

If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, my book is coming soon.

This newsletter is for informational and educational purposes only. It is not investment, legal, tax, or financial advice, and it is not an offer or solicitation to buy or sell any security. Figures are drawn from publicly available reports as of June 23, 2026, and are subject to change. Do your own research and consult a licensed professional before making any decision.

Read more

What Corgi Actually Does

Corgi is an AI-native, full-stack commercial insurance carrier built for startups and high-growth companies. It was founded in 2024 by Nico Laqua (CEO) and Emily Yuan (COO), came out of Y Combinator’s Spring 2024 batch, and has spent less than two years going from zero to a $2.6 billion valuation.

On May 28, 2026, …

Read more

What Contraline Actually Does

Contraline, Inc. is a clinical-stage biopharmaceutical company headquartered in Charlottesville, Virginia. It’s building what could become the first approved pharmacological contraceptive for men. The company has two investigational products in active clinical development: NES/T Gel and ADAM.

On June 2, 2026, Contraline closed a $92.5 million Series B, co-led by BVF Partners and RA Capital Management, with participation from GV (Google Ventures), Lumira Ventures, Invus, and other investors. The round funds Phase 3 readiness for NES/T and continued advancement of ADAM.

The company was co-founded by Kevin Eisenfrats, who serves as CEO. Eisenfrats has been building in men’s reproductive health since his undergraduate days at the University of Virginia, where the core concept behind ADAM was first developed. He’s not a career biotech executive who wandered into a hot category — he identified the unmet need, raised early capital from GV, ran first-in-human trials in Australia, and is now steering a late-stage clinical program backed by some of the sharpest life sciences investors in the business. Iris van Alderwerelt van Rosenburgh from BVF Partners joins the board alongside the Series B close.

The two investigational male contraceptive products:

• NES/T Gel — a daily hormonal topical gel combining Nestorone (a progestin) with testosterone. Suppresses sperm production while maintaining physiological testosterone levels, avoiding the side effects that tanked prior hormonal male contraceptive programs. Licensed from the Population Council and the NIH’s NICHD. Phase 2b completed, enrolling 462 couples globally. Phase 3 start projected for 2027.

• ADAM — a non-hormonal injectable hydrogel implanted into the vas deferens via a minimally invasive procedure. Physically blocks sperm transport and is designed to eventually biodegrade, making it reversible. First-in-human trial completed across three Australian sites. 25 participants enrolled, 100% successful implantation rate. 24-month efficacy data shows azoospermia in participants who’ve reached that milestone. Phase 2 now in planning.

Read more

Every generative AI product you’ve used so far works the same way. You give it instructions. It runs computations. Eventually, it hands you a result. That lag — even if it’s milliseconds — is a ceiling. It means AI can’t react. It can only respond.

A San Francisco startup called Reactor just raised $59 million to blow past that ceiling entirely.

Reactor emerged from stealth on May 28th with a Series A led by Lightspeed Venture Partners. WndrCo, Amplify Partners, Sky9 Capital, and FPV Ventures also participated. The company was co-founded by Alberto Taiuti and Bryce Schmidtchen, both former technical leads on the Apple Vision Pro. Taiuti previously co-founded Luma AI, where he built the infrastructure behind one of the most widely used 3D and video generation platforms in the market. They know what it takes to ship this stuff at scale.

The product: a unified SDK and API that lets developers build real-time interactive AI applications in just a few lines of code. No batch processing. No prompt-and-wait. The video generates as you interact with it — dynamically shaped by user input, in the moment it’s happening. Reactor calls this the shift from AI as a tool you prompt to AI as an experience you step inside.

Jeffrey Katzenberg, the Hollywood mogul turned investor, joined through his WndrCo holding company as a board observer. His read on it: “Literally every facet of the production pipeline can be enhanced using these tools. Live action, TV, film, commercial work, animation — any visual medium.” That’s not a media pitch. That’s an infrastructure pitch wearing a Hollywood jacket.

Read more

Most fintechs spend a decade trying to capture banking-like economics without becoming a bank. Mercury just did the opposite.

The headline last Wednesday was the money. Mercury closed a $200M Series D at a $5.2 billion valuation, led by TCV, with Andreessen Horowitz, Coatue, Sequoia, Sapphire, Spark, and CRV all participating. That’s up 49% from a $3.5B valuation just 14 months ago. CEO Immad Akhund told CNBC the company is at $650M annualized revenue with four years of profitability. Total funding to date sits at roughly $700M across primary and secondary.

The number that actually matters is buried lower in the announcement. Mercury submitted an application to the Office of the Comptroller of the Currency for a national bank charter, and applied for federal deposit insurance with the FDIC. The company has stated plans to file with the Federal Reserve to restructure as a bank holding company.

For context, this is the hardest path in fintech. Estimated timeline is 12 to 24 months. Capital requirements are steep. Operating constraints are real. Very few fintechs in the modern era have made it through.

So why do it?

Read more