Living Forever is an Asset Allocation Problem

In this issue:

• Living Forever is an Asset Allocation Problem—If there are advances in longevity, and they're unevenly distributed, they illustrate some of the economic and political problems that aging societies already have.
• Books—Genre fiction is rapidly getting automated, but it's an easier kind to systematize.
• SPVs—When venture funds are deliberately small, they have to contort their model a bit to do follow-on investments.
• Blockchain—Sometimes you do turn out to rely on trusted third parties.
• National Balance Sheets—Japan's government has borrowed heavily, but partly to fund some levered bets that have done well.
• A Voluntary Y2K—US automakers are reducing their reliance on Chinese software. At least on paper.

The Diff February 9th 2026

0:00

/723.226122

1×

Ask questions about this post on read.haus.

Living Forever is an Asset Allocation Problem

As Bloomberg is occasionally very insistent on informing people, Bryan Johnson, founder of Braintree, has dedicated all the money and free time that comes with a successful exit to the task of living forever. He's approaching this with the usual startup scene suite of behaviors: A/B test everything, be so studiously indifferent to looking weird that the result is basically camp, find a monetization model eventually, and, in the meantime, throw money at every problem that can be trivially solved with dollars, on the assumption that if it's clear that you know what you're doing, more dollars will materialize on demand.

There are some health interventions that are pretty easy, and once you've knocked out the big ones—avoid a few obvious bad habits, get enough sleep, work out, etc.—it increases the salience of other improvements. And, if you're bullish on the concept of aging as a long list of loosely-related factors that lead to cumulative increases in mortality rates, you also get some convexity: early cancer detection, screening for rare genetic conditions, or getting your plasma replaced by that of someone younger will all have a magnified impact on mortality if you aren't regularly riding your motorcycle while drunk.

But one of the questions this raises is: how do you save for retirement if you think you might live for five hundred years? Let's suppose, for the sake of argument, that significant life extension ends up being possible, but expensive, and that it takes a while for the cost to come down. One way to dodge this is to argue, as Johnson has, that you really just need to stay alive long enough for AI to take off. At that point you (optimistically) will have access to a superabundance of incredibly high-quality medical care, plus a superintelligent entity that can persuade you to adhere to whatever the healthiest lifestyle turns out to be. Pessimistically, the AI will briefly appreciate that your organs are younger than your chronological age while it's converting them into paperclips or whatever.

But let's leave that aside as a cop-out: hypothetical technological advantages can bail you out of whatever problem you care to ignore, but it's hard to tell when things will asymptote. Someone who decided circa 1970 that the transistor was as small as it was likely to get, but that all you had to do was plot a log chart of how long it took to cross the Atlantic over the last few centuries to know that the early 21st century would be an era of hypersonic travel. You can't just extrapolate trends; you have to underwrite asymptotes.

And that means you have a few risks to worry about:

1. Someone who has predictable cash outflows over an extremely long period is short duration. Fortunately, the hypothetical of an exceptionally long-lived investor with high fixed costs actually nicely illustrates duration risk. Right now, if you need $1m/year pretax for the next thirty years, you need $20.5m. But suppose you buy $20.5m worth of new-issue 30-year treasury bonds with that $20.5m, and live off the interest, and then they mature in 2056 and you need to reinvest—but if by then the 30-year yields 1% (perhaps because there are so many near-immortal retirees hedging their duration risk), the pretax income from rolling that position over drops to $205k.
2. They also have to worry about inflation. At 2.5% inflation, even the original $1m in income now has only $477k of purchasing power. If you want to hedge duration risk and inflation risk in the same instrument, you can—just buy gold, the perpetuity that perfectly hedges your inflation and duration risk at the low, low cost of having a 0% return.
3. Okay, so they can invest in other assets, but now they're taking on other risks. A handful of countries have, since the inception of their stock markets, avoided a drawdown to zero roughly. But America and Australia are probably just lucky in that respect, and it's likely that the distribution of returns for any country's equity markets includes a scenario where equity holders get roughly zeroed.
4. Which is fine! The world is bigger than stocks, bonds, and one particularly yellow metal. A duration-matched portfolio for a relatively immortal individual investor could include very long-lived infrastructure, which tends to combine current income with inflation protection. This hypothetically long-lived investor would probably want to allocate money to ports, airports, railroads, toll roads, etc.
5. Which is fine until populist politicians start railing against multicentenarian multicentimillionaires extracting wealth from the rest of us.

A long time ago, The Diff explored the question of how to allocate capital if you run the Long Now Foundation's endowment, with the ultimate conclusion that it should be a barbell portfolio with mostly low-risk investments and moonshots that basically buy options on anything that disrupts those low-risk investments—but with the added kink that the real persistence is that ideas stick around for a long time, and if a near-apocalypse happens in 2300, we might be able to reconstruct the Long Now Foundation by 2400. But you can't reconstruct yourself in the same way. If you need a steady income each year, in order to keep yourself going, the entire sociopolitical system that makes that possible is part of your extended phenotype.^[1]

But the risk of hitting zero is just impossible to avoid. At some point, the efficient frontier is defined by the fact that an indefinitely long-lived person can afford to take some risks that nobody else can, but also that if they and others like them end up claiming a large enough share of society's output, the real risk is that they'll get their wealth expropriated. In extreme cases, rights are always defined by what members of the winning coalition can get away with, and they're eliminated to the extent that they benefit the losers. Prospective losers can't really hedge this except by holding on to power at all costs, but that's a miserable way to live—Kim Jong Il is one of the unluckiest people ever to live, in that he has to spend all of his time pretending to be a superhero lest he get deposed.

This model is somewhat fanciful, in that it presupposes that a small number of wealthy people live to extreme ages. But what it illustrates is the political dynamic of societies where there's extreme income inequality by age—which is every society that doesn't treat the net present value of future salaries as part of people's economic balance sheet. In a scenario where young people don't have a lot of money, but know that they'll make a lot in the future and can borrow against it today, you can have a healthy equilibrium where young and old people both get benefits and the financial system profitably intermediates between them: older people will spend more than they earn, and pay for it by having earned more than they spent when they were younger. But if young people either don't experience that or don't think they will, the whole system breaks down. Aging countries will have to manage this, and younger countries will have to think ahead to how they'll do so. If we're lucky, unevenly distributed advances in radical lifespan-enhancing technology will give us a nice early-warning system first.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• Ex-Citadel/D.E. Shaw team building AI-native infrastructure that turns lots of insurance data—structured and unstructured—into decision-grade plumbing that helps casualty risk and insurance liabilities move is looking for forward deployed data scientists to help clients optimize/underwrite/price their portfolios. Experience in consulting, banking, PE, etc. with a technical academic background (CS, Applied Math, Statistics) a plus. Traditional data scientists with a commercial mindset also encouraged. (NYC)
• Series A startup that powers 2 of the 3 frontier labs’ coding agents with the highest quality SFT and RLVR data pipelines is looking for growth/ops folks to help customers improve the underlying intelligence and usefulness of their models by scaling data quality and quantity. If you read axRiv, but also love playing strategy games, this one is for you. (SF)
• YC-backed startup automating procurement and sales processes for the chemicals industry, which currently relies on a manual blend of email, spreadsheets, legacy ERPs, etc. to find, price, buy, and sell over 20M+ discrete chemicals, is hiring full-stack engineers (React, TypeScript, etc.). Folks with exposure to both startups and bigtech, but also an interest in helping real-world America with AI preferred. (SF)
• Ex-Bridgewater, Worldcoin founders using LLMs to generate investment signals, systematize fundamental analysis, and power the superintelligence for investing are looking for machine learning and full-stack software engineers (Typescript/React + Python) who want to build highly-scalable infrastructure that enables previously impossible machine learning results. Experience with large scale data pipelines, applied machine learning, etc. preferred. If you’re a sharp generalist with strong technical skills, please reach out.
• A hyper-growth startup that’s turning the fastest growing unicorns’ sales and marketing data into revenue (driven $XXXM incremental customer revenue the last year alone) is looking for a senior/staff-level software engineer with a track record of building large, performant distributed systems and owning customer delivery at high velocity. Experience with AI agents, orchestration frameworks, and contributing to open source AI a plus. (NYC)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.

Elsewhere

Books

Last week's Diff touched on the decline of the trade paperback. Conveniently, the NYT has a look at the new world of genre fiction: hyper-niche romance authors using AI to crank out hundreds of titles a year. Romance and porn are both prone to extremely narrow micro-genres, because they're both categories where the viewer is bringing plenty of their own preferences/experiences/emotional baggage and is looking for something that hits the right specific themes, rather than for excellent production values for an off-target performance. A zero marginal-cost world like the digital distribution of text is a good laboratory for understanding what market share will look like in places with nonzero, but declining, marginal costs: everything's a bit worse, but the exact thing you're looking for is definitely there. But it's also a misleading one, in that a genre that's always been hyper-targeted will have more training data around which tropes readers expect. Writers of literary fiction don't think of themselves as trying to produce out-of-distribution tokens, but if every model thinks that enemies-to-lovers is a standard expectation, enemies-to-still-enemies-but-with-more-nuanced-reasons-to-hate-each-other or whatever is something AI will have a harder time replacing.

SPVs

Benchmark Capital has notoriously kept their fund size small and focused on finding the best companies early. This presents a problem when a company they've invested in starts to get more capital-intensive, and they've chosen the natural solution: when one of their portfolio companies, Cerebras, raised a later-stage round at a high valuation, they apparently raised two new vehicles just to buy into that round. This is very reasonable for a VC who has access to more good deals than ready capital (I, too, raise SPVs). And it's a way to take advantage of the "gentleman's pro-rata": pro-rata rights mean that an investor who owns 5% of a company can invest enough to maintain 5% ownership in subsequent rounds. A gentleman's pro-rata is when someone doesn't have that legal right, but the startup is happy to let them raise more to participate. For LPs in such a fund, it means that some of them are buying direct exposure to the fund's investments, and the rest are buying that plus the privilege of investing in follow-on vehicles. And this is part of what subsidizes small funds: their smaller investors are looking for a return, but their biggest investors are happy to subsidize access.

Blockchain

The core initial pitch for decentralized finance was that it didn't rely on trusted third-parties. If you wanted to transfer money to someone else, the two of you could do so, and the payment network was technically incapable of stopping you. This was appealing to three kinds of principled libertarians: the ones who valued privacy and independence for its own sake, the ones who didn't care for following the rules of the IRS, and the ones who felt the same way about the DEA. Things have evolved since then, and the blockchain has shifted from being decentralized-by-default to being decentralized in theory but optionally controlled by a single entity. So Tether has decided to freeze some criminals' funds, at the behest of the government of Turkey. Right now, the marginal Tether user is not someone with the ideological motivations mentioned above; it's someone who wants a dollar-pegged asset and, for whatever reason, would prefer not to go through a bank that uses dollars, knows what KYC and AML stand for, etc. This user isn't ideological, but is pragmatic, and if their asset isn't as safe as they thought it was, they'll pull money out of the Tether ecosystem and into whatever they think is the safest adjacent bet.

National Balance Sheets

The FT has an overview of some fun research into Japan's balance sheet, with the upshot being that Japan has profited from intervening in both currency and equity markets and is much less levered than it looks ($, FT). This is separate from the usual line that the Japanese government's deficits are largely financed by local savings, but adds a bit to that argument: when governments spend money, they're getting something for it, and if what they're getting is an investment then that deficit spending can be sustainable indefinitely.

A Voluntary Y2K

US auto companies have until mid-March to comply with a new directive that they not use code written by Chinese companies, which involves either rewriting lots of code that runs things their suppliers' suppliers sold them, or just encouraging those suppliers to sell to a US-controlled entity ($, WSJ). Jurisdictions do actually matter in cases like this; Chinese and American regulators have both pushed for companies to store their data locally, for example, so it's easier for the local legal system to demand access later on. But it's hard to work backwards from this policy to a coherent threat model.