Crosscurrents | The Power of Price, The Great Capital Shift, Collecting the Toll
Sanjeev Krishnan | October 13, 2026
My Rent is Too Damn High — The Pricing of Pricing in the Era of Market Concentration and AI
Think in Circles, Not in Lines — Part Two
My Rent Is Too Damn High — The Pricing of Pricing in the Era of Market Concentration and AI
The price of living has become the price for power.
In 2024, incumbent political parties lost vote share in every developed Western country that held a national election, the first clean sweep in roughly 120 years of records, and in most of those campaigns, the cost of living was a central issue. A dozen eggs in the United States peaked at $6.23 in March 2025. The price of things became a cultural headline.
We have spent four years arguing about why: snarled supply chains, stimulus checks, pandemic labor shortages, tariffs, volatile commodities. Each explanation treats price as the sum of its inputs plus a stable margin. If that assumption holds, the input shocks of 2021 through 2025 should have squeezed margins. But the opposite happened.
I’ll give you three data points to illustrate. First, the U.S. Bureau of Economic Analysis’ (BEA) measure of after-tax profit per dollar of real nonfinancial corporate output hit 19.5 cents in the second quarter of 2026, a record in a series that began in 1947. Second, the S&P 500’s net profit margin was 14.8% in the first quarter of 2026 and 17.0% in the second, against a five-year average of 12.6%. The second-quarter figure is inflated by one-off investment gains at Alphabet and Amazon, but FactSet’s estimate excluding both is 15%, still a record. Third, and most importantly, because it predates the pandemic, De Loecker, Eeckhout and Unger estimate in the Quarterly Journal of Economics that the aggregate markup of US public companies rose from 21% above marginal cost in 1980 to 61% in 2016, with most of the increase coming from the top of the distribution. The median firm’s markup did not move. The large firm’s did.
The Kansas City Fed found that markup growth alone could account for more than half of US inflation in 2021. The European Central Bank found that unit profits explained about two-thirds of Euro-area domestic price pressure in 2022, double their long-run share. You can argue about whether that is pricing power or firms pricing ahead of expected costs. But price was behaving less like a sum of inputs than a judgment from the market.
Pricing power has always been a privilege. Warren Buffett has said that pricing power is the single most important thing to judge in a business. For most of economic history, the ability to set price belonged to the scarcest layer of the economy: the monopoly or oligopoly, the physical chokepoint, the brand, or the owner of the channel.
Look at who has held it for over a century and the pattern is repetitive. Pricing power often goes to whoever faces a fragmented counterparty. One platform facing a million merchants, one railroad facing ten thousand suppliers. And over time, it migrates to whoever intermediates the channel. Since the late 1990s, more than three-quarters of US industries have become more concentrated, and firms in industries with the largest increases in concentration earned higher margins without measurable efficiency gains. The investor’s shorthand followed the data: back the firm with high market share, high switching costs and the best knowledge of its customers’ demand. For forty years that shorthand worked, because those three variables were the inputs to pricing power.
The era of AI may reset that. The three variables could be coming apart, and the pricing of pricing could move from who has the share to who has the information and the compute. Three shifts are already visible.
Pricing becomes continuous.
Delta Air Lines began deploying an AI pricing engine from Fetcherr in 2024; by mid-2025 it covered roughly 3% of domestic fares with a stated target of 20% by year end, and the rollout subsequently drew congressional scrutiny, while Delta has not publicly updated its adoption numbers. Instacart ran live price experiments through its platform until December 2025, when Consumer Reports found nearly three-quarters of tested items showing different prices to different customers in the same store at the same time, with gaps up to 23%. Amazon was estimated to be changing prices 2.5 million times a day as far back as 2013. These firms are emblematic of a trend in which the unit of pricing has shrunk from a season to a second.
Pricing intelligence democratizes.
The scale threshold is falling. Repricing software that only an airline or a supermarket chain could afford a decade ago now ships as a feature in a point-of-sale system. A 2026 small-business survey found 35% of owners already using AI-supported pricing tools. The survey comes from an advocacy group, so treat it as directional, but its implications about the broader shift are clear. Demand-side knowledge, the third leg of the old shorthand, is becoming a utility rather than a moat.
Pricing coordinates without a meeting.
This third shift is the most troubling to me. Calvano and co-authors showed in the American Economic Review that simple reinforcement-learning agents, set loose to price against one another in a simulated environment with no communication, can learn to sustain supracompetitive prices, capturing roughly 70% to 90% of the gains from moving toward monopoly profits. Assad and co-authors confirmed this pattern in Germany, where gasoline stations that adopted pricing algorithms raised margins by 9%, and in local duopolies where both stations adopted the algorithm, by 28%. RealPage, which the Justice Department alleged controlled at least 80% of the commercial revenue-management software market for multifamily housing, reached a proposed settlement with DOJ in November 2025 that would restrict the use of competitors’ nonpublic data to generate rental pricing recommendations. California enacted new restrictions on shared pricing algorithms used for anticompetitive coordination; New York’s restriction on algorithmic rent-setting was partially enjoined in September 2026 on free-speech grounds; the FTC sought public comment on a proposed enforcement policy statement regarding personalized pricing in August 2026. Regulators are discovering that the cartel no longer needs the cartel.
For investors, the old test for pricing power asked three questions about a company. The new test asks four questions about a system.
Who owns the demand data, and is it proprietary or purchasable? Who owns the repricing capacity, and does it reside within the firm or with a vendor that also serves the firm’s competitors? Where is the constraint that pricing cannot route around: the slot, the acre, the interconnection, the lease? And can the price be set without triggering the Balancing mechanism, whether that is a regulator, a legislature, or a customer who notices?
Market share answered the first three questions for forty years. It may not answer any of them for the next ten. The scarce layer of the economy is shifting towards whoever holds information about a position and whoever owns the algorithm that prices through it. That is a different company, a different moat, and quite possibly a different antitrust régime.
We should also note, a reminder for anyone who invests in physical chokepoints, that the one thing an algorithm cannot reprice its way around is a scarce constraint.
Retirees to the Real Economy — Resetting the Clocks of Money and the Implications for Global Capitalism
Boomers started turning 65 in 2011.
That year, Americans purchased $240 billion of annuities. In 2025, they bought $464 billion, a record for the fourth year running, and the first half of 2026 set a new high. Add it up, and over fifteen years, roughly $4.1 trillion in cumulative annuity sales has provided insurers with a substantial and recurring source of investment capital. Over the same period, global private debt assets under management grew from approximately $358 billion in 2011 to $1.7 trillion by 2025, according to Preqin. The two numbers are not a coincidence in my opinion.
In the scale of capital markets, annuities are still small. That $4.1 trillion is about 6% of the US stock market and barely more than 1% of global stocks and bonds. However, they are increasingly the source of new liquidity for these markets (more new flow and increasing stock). Life insurers now hold approximately $3.9 trillion in corporate and foreign bonds, 23% of everything outstanding, up from 22% a year earlier. They have been the largest institutional holder of corporate bonds since the 1930s.
For a century, banks turned deposits into loans. After the 2008 global financial crisis, capital rules made long, illiquid lending expensive for them, and banks’ share of corporate lending fell from 48% in 2015 to 29% in 2025, by the Fed’s own count. Those borrowers still need capital. A growing share of corporate, asset-backed, and real-asset credit now comes from insurer balance sheets, and the platforms that manage those balance sheets are increasingly owned by large alternative asset managers. Private-equity-affiliated insurers held about 10% of life-industry general-account assets in 2014 and about 14% by 2024; on a broader definition of private ownership, nearly 20% of life-industry invested assets, roughly $1.2 trillion, up from 2.5% in 2011. They write 18% of the annuity market and a third of indexed annuities.
These platforms make money the way banks do: borrow at one rate, lend at another, keep the spread, often with leverage. Apollo’s Athene manages $414 billion in gross invested assets; KKR’s Global Atlantic, $220 billion; Brookfield’s wealth solutions business. Separately, Blackstone manages approximately $290 billion in assets on behalf of insurance clients. In 2025, Apollo earned more from spread than from fees. The spread machine has two tailwinds that are not going away: societies aging into a preference for guarantees and banks that do not want duration.
We should also note that roughly one in ten private-sector workers now participates in a defined-benefit pension, down from 38% in 1979. About half participate in a defined-contribution plan. The employer that once carried longevity and market risk has handed it to the individual. The individual is now handing it back, one annuity at a time, to an insurer. Meanwhile, the pensions that remain are paying insurers to take over their obligations. U.S. pension risk transfer transactions totaled approximately $52 billion in 2024 and $49 billion in 2025, with outstanding single-premium pension risk transfer assets reaching approximately $342 billion. Either way, the money ends up in the same place.
One underappreciated consequence of this shift is how it changes the investment horizon and risk framework governing retirement capital. When capital moves from pensions to annuities, it changes the clock of money.
A defined-benefit pension is measured against a market benchmark, marked to market, and governed by a liability with a duration of 14 years or so. An annuity is measured against a guaranteed crediting rate and a regulatory capital ratio, carried at book, and governed by rating-agency definitions of risk. The traditional life book that preceded both was built on mortality tables rather than spread and ran longer still. Each step shortens the clock and narrows the tolerance for loss.
That is the mechanism by which retirees reshape the real economy. The money arrives with a target return by a target date, and the transmission mechanism is rated, contracted and collateralized. It cannot own equity in a transition; it must lend against one. So the assets it reaches for are the ones that can be made to look like bonds: private placements, which grew from $386 billion to $849 billion between 2014 and 2024 and now make up 46% of life insurers’ bonds; asset-backed securities, 14.5% of bonds and the highest share since before the financial crisis; and the BBB bucket, which has gone from under 20% of insurer bonds before 2008 to roughly 37% today, with another 5% below investment grade. Private-equity-owned insurers lean harder on all three: 31% of their bonds are in structured products against 13% for the industry, and more than a fifth carry private letter ratings rather than public ones.
If these trends continue, and the demographics say they will, global capitalism becomes more debt-financed, shorter-dated, and more dependent on the assumption that a rating agency’s definition of safety will survive a cycle. A transition from an owner’s mindset to a lender’s mindset, applied at the scale of a national retirement system.
Over fifteen years, America has rebuilt how it funds retirement, and in doing so, who funds the real economy. Regulators have taken notice. The NAIC has new disclosure rules for private assets and affiliated origination landing this year, Bermuda tightened its reinsurance rules in January, and the Senate has been writing letters since May. Like every retooling of capital markets, this one will be judged by how it survives an actual credit cycle. This modern spread machine has not yet seen one. The retirees whose money runs it are counting on it not needing to.
Think in Circles, Not in Lines — Part Two
Last week, I ended with the claim that a system that leaks value is a system in which someone will eventually stand where the output comes back around and collect it. The best mainstream investors already stand on those spots. Finding them on purpose takes a different kind of underwriting.
What It Means to Invest in a Loop
A business is a function. Capital and inputs in, product and return out. A system is a set of loops, where the result of one action comes back around as the cause of the next. Feedback can amplify itself (Reinforcing), correct itself (Balancing), run through belief and change what the facts become (Reflexive), arrive late (Delayed), or hit a wall (Constrained).
To invest in a line is to buy a function and wait for its output. The thesis lives inside the company; the exit is someone else’s opinion of that output at a later date. To invest in a loop is to buy a position in the feedback itself. This thesis lives between companies, in the grid, the watershed, the supply chain, and the insurance pool. Each turn can generate a return.
A loop investor asks three things: where is value being created that the system cannot keep, which mechanism is letting it escape, and what would have to exist for the system to keep it? The answer to the third question is the toll.
The Loop Thinking That Is Already Proven
Berkshire Hathaway is a loop. Insurance float funds equity purchases; those businesses strengthen the balance sheet; a stronger balance sheet enables more insurance to be written at a lower cost of float. Each turn financed the next while the market valued the companies.
FICO’s scores make repayment histories legible to lenders; better credit information can lower borrowing costs for creditworthy borrowers; broader access to credit generates more repayment histories that can be scored. The leak was illegibility. FICO owns no loans, holds no risk, and takes a fee every time the loop turns.
Visa’s merchants and consumers reinforce one another. More merchants accept the card because more consumers carry it, and the reverse. The leak was trust between strangers; the toll is network fees tied to the payment activity its infrastructure enables.
Apollo is another useful example. Retained spread grows Athene’s equity, equity supports more annuities, annuities fund more loans, and the whole structure supported nearly $400 billion in gross invested assets by the end of 2025. In 2025, Apollo earned more from spread-related earnings than from fees. Owning the return path produced the alpha.
Why Lines Cannot Change Systems
The era of moderation rewarded lines because the macro tide lifted every function. In the era of volatility, the binding constraints are in the loops: grid interconnection, soil, water, insurance pools, and shipping compliance. A company that performs well inside a broken loop gets absorbed, offset, or arbitraged by it. Its output leaks.
Three kinds of leaks recur:
- Unpriced externalities: Heat, waste, methane, and mortality, with losses booked elsewhere.
- Unowned constraints: A permitted project nobody can energize, a charger nobody keeps up.
- Illegibility: Value that exists but cannot be underwritten, such as a farmer’s organic transition or a smallholder’s yield history that lives in no database.
Most durable leaks are all three at once. Together, they defeat any investor who sees only one mechanism. Fixing a leak means changing the feedback. You build what lets the system keep what it was losing, and the system pays you to keep it there. That is the toll, and it is also the theory of change.
The spread is yours for as long as the fix holds. Extract at the chokepoint, and the Balancing mechanism eventually responds: a regulator, a competitor, or a bypass. Widen it, and the system tolerates you because you have lowered its cost. A toll that fixes the leak earns the loop. A toll that deepens it earns a correction.
The five mechanisms enter roughly in the order a deal unfolds. Reflexivity ignites with a proof point. Delay lets you buy the toll cheaply because most capital cannot wait. Constraint puts it at the node everyone must pass through. Reinforcement pays as each turn produces the data, track record, and cheaper capital that funds the next. Balance decides whether the toll lasts.
Putting Loop Thinking into Systems Investing Practice
Several investments are early, and systems can humble anyone who thinks they have mapped one. Still, when I lay our portfolio against the three leaks, I see the same pattern.
Food & Agriculture
- Clear Frontier: The 36-month leak. Farmers carry organic costs and lower yields at conventional prices for three years before certification. Traditional lenders often struggle to underwrite the transition, and conventional land leases may not accommodate the timeline. We co-founded Clear Frontier in 2019 with 770 acres to buy land and lease it on crop-share terms through the transition. It now manages more than 30,000 acres across four states. The toll is land tenure matched to the transition’s clock; each transitioned acre makes the next cheaper to underwrite.
- Growers Edge: The adoption leak. The risk of trying a new input or practice sits entirely with the grower, so retailers under-sell and farmers under-adopt. Growers Edge writes yield warranties on crop plans through retailers and manufacturers, covering more than a million acres. The toll is risk transfer for the farmer betting a season on it.
Energy
- Kanin Energy: The heat leak. Industrial waste heat goes up the stack because plant managers cannot justify the capital or power-market expertise to recover it. Kanin develops, owns, and operates recovery projects at the host site. The toll is a behind-the-meter offtake contract against a resource with zero incremental fuel cost.
- Resurety: The price leak. Nodal prices, basis, and capture rates are difficult to forecast and manage, exposing renewable energy projects and PPAs to significant pricing and revenue risk. Resurety makes these risks more transparent through its CleanSight analytics platform and enables transactions and hedging through CleanTrade, its CFTC-registered marketplace. Its toll is price discovery.
Oceans
- Oxzo: The oxygen leak. Warming water and algal blooms starve salmon pens of dissolved oxygen; trucked liquid oxygen is expensive and logistically challenging. Oxzo generates oxygen at sea and leases its oxygenation systems under multi-year contracts, with reported mortality down roughly a quarter and biomass up 5% to 10%. The toll is paid out of fish that did not die.
- Echandia: The dry-dock leak. An early battery failure can require dry-docking and recertification, making ferry operators hesitate to electrify. Echandia guarantees 20,000 to 30,000 cycles, so battery life is designed to equal vessel life. The warranty is the toll: it lets a port or ferry authority underwrite the conversion.
Why We Believe Loops Produce Systems Alpha
Each turn makes the next cheaper. Clear Frontier’s first transitioned acre was a bet; its 10,000th is a recursive underwriting model. The return depends on the loop turning, which the investor can see and potentially control.
A loop investor who owns the toll is paid from avoided losses, a steadier source of return than hoped-for gains. Oxzo is paid from fish that did not die; Kanin recovers heat already being thrown away. The leak was costing the system something before you arrived.
Most capital cannot hold through three years of organic transition, four years of loan seasoning, or the gap between a permit and an energized battery — the investor who can earn a liquidity premium. And a toll at a node the system must pass through is designed to be defensible: competitors can build a better widget more easily than a second node.
Reflexivity can also turn against you. Apollo trades at about 12 times forward earnings against 18 or more for Blackstone because the market has not yet decided whether the Balancing mechanism in private credit is as strong as advertised. Belief has to hold long enough for the base case to prove itself.
Systems investing starts with the reflexive move and ends with the recursive one.
A loop lets output return as input, changing the system with each turn. The investor who closes the leak is paid every time that happens. Find the leak. Originate the fix. Collect the toll.
—
Companies shown are S2G portfolio companies selected to illustrate discussed themes, not based on performance, and are not representative of all S2G investments. There is no guarantee that S2G’s strategy will succeed or that any company referenced will generate positive returns or succeed. Metrics are company-reported and have not been independently verified by S2G. A complete list of S2G portfolio companies is available upon request. Content is for informational purposes only and reflects the author’s current views, which may change without notice. It is not an offer to sell, or a solicitation of an offer to buy, any security or interest in any S2G-managed fund or vehicle. Portfolio companies are referenced for illustrative purposes only, are not recommendations, and are not representative of all S2G investments or results any investor will achieve. Certain information comes from third-party or portfolio company sources believed to be reliable but not independently verified by S2G. Statements about investment theses, impact, or expected outcomes are forward-looking, involve significant risks and uncertainties, and may not be realized. Private company investing involves substantial risk, including loss of all invested capital. Past performance is not indicative of future results.
The views and opinions expressed are those of the S2G team members identified and do not necessarily reflect the views of S2G. This content is for informational purposes only, should not be taken as legal, business, tax or investment advice, or be used to evaluate any investment or security, and is not directed at any investor or potential investor in any investment vehicle sponsored by S2G. Investing involves risk, including the risk of loss. Specific companies mentioned in this podcast are for educational purposes and should not be construed as an endorsement of any kind. Please note that S2G may maintain investments in some of the companies discussed on this podcast. Any past performance discussed is not indicative of future results. For more important information, please see s2ginvestments.com/disclosures.