For funds raised and deployed during the zero-rate era, waiting is no longer a passive choice. It is an active bet that the old cost of capital will return.
On August 18, the 30-year Treasury yield touched 5.337% intraday, its highest level since April 2007. Although it has since retraced, the signal was never the print. It was confirmation that the long end of the curve no longer permits investors to treat a return to zero-rate assumptions as the base case.
Startups are primarily financed with equity, which carries no coupon, no maturity, and no refinancing calendar. Historically, no clear line ran directly from the long bond to a startup’s cap table or valuation. Today, that insulation is gone.
The fact that the curve has anchored near 4.7% at ten years and 5.3% at thirty matters to venture capital, not because every startup carries floating-rate debt, but because every venture valuation contains a risk-free rate, whether explicitly in a model or implicitly in the price an investor will pay. A startup that has never issued a bond still borrows its valuation from the bond market. During ZIRP, abundant capital and rising comparables allowed the industry to ignore that fact. That option has disappeared. The transmission runs through three channels:
The discount rate. Venture equity is a long-duration claim: much of its value depends on cash flows or an exit many years away. A higher risk-free rate reduces the present value of those distant outcomes and raises the return required for uncertainty, illiquidity, and dilution.
The exit market. Strategic acquirers, buyout funds, and public-market investors price assets using today’s financing costs and public comparables, not the environment in which the venture round was completed.
The funder. An LP able to earn roughly 5% in long Treasuries requires more than paper appreciation to accept venture’s illiquidity, dispersion, fees, and ten-plus-year duration. That channel reprices more than the asset. It reprices the next fund.
That is the valuation argument. The distribution data makes it a governance problem.
PitchBook puts the average 2021-vintage fund at 0.05x DPI after five years, which is the lowest five-year DPI for any vintage this century. The same report shows a 17.1% one-year horizon IRR for U.S. venture, while cumulative net cash flow to LPs has totaled negative $202 billion since 2022. These measure different populations, which is the point: even the flattering number remains materially dependent on marks rather than distributions. The concentration is equally stark, as four companies accounted for 93.5% of YTD 2026 U.S. venture exit value.
A 2021 carrying value is not wrong merely because Treasury yields rose. But it embeds assumptions about exit timing, multiples, capital needs, dilution, and buyer financing. If those assumptions still require 2021 conditions, the mark is a macro forecast disguised as accounting.
The repricing is not a forecast. It is already in the documents. Across the insider bridge financings we have reviewed at CRAGSI since June (a sample, not a survey), terms have moved from simple 1x non-participating structures toward participation, seniority, and pay-to-play. Price discovery is happening at the round level, one financing at a time, regardless of whether it has reached the quarterly mark.
Continuation vehicles, tender offers, and cross-fund solutions can create liquidity and preserve upside. They cannot eliminate price discovery. A properly run GP-led process requires an independent market-clearing price, conflict management, and a real choice for existing LPs; the transaction tests the legacy mark rather than shielding it from scrutiny.
A managing partner recently described to us a portfolio company held for nearly a decade. It grows 3% to 5% annually, has contracted revenue, durable customer relationships, and predictable cash flow. It is neither a venture winner nor an obvious restructuring. Another venture fund is unlikely to finance it, and a buyer will not pay the carrying value. This is the category many mature portfolios avoid naming: a viable company housed in the wrong security, at the wrong mark, for the wrong duration.
The cost of waiting is calculable
Hold that company three more years and ask what it must do simply to justify the hold.
At the most charitable hurdle, the long bond at roughly 5.3%, crediting venture with no illiquidity or dispersion premium, the position must appreciate about 17% over three years to match the opportunity cost of what the LP could have earned in Treasuries. I use the long end rather than the three-year point deliberately: three years is a decision interval, not the asset’s duration. The position has no maturity, and this one is already a decade old. At an illustrative required return of 20% to 25%, the same hold requires 73% to 95%.
Even assuming value compounds one-for-one with 3% to 5% annual growth and the multiple holds, the position appreciates only 9% to 16% over three years. It neither clears the Treasury floor nor is within reach of a venture hurdle. That math credits the position with growth alone and no distributions. If the company is genuinely throwing off cash the fund can take, that changes the arithmetic, and it also changes the category: the position is a harvest, not a hold, and it belongs on a distribution schedule rather than in a reserve model.
The 9% to 16% also assumes a constant exit multiple, which is the generous case. PitchBook’s Q2 2026 US VC Fundraising and Returns Report shows the median step-up at exit fell from 62.8% in 2021 to 15.5%, and the median company exiting above $500 million had raised $323.7 million to get there, which is more than twice the $157 million median a decade earlier. In 2021, patience was cheap. Today it has a hurdle rate, and many ZIRP-era holds do not clear it.
The obvious rebuttal is that the fund cannot sell at the mark, so the comparison is unfair. Correct, and that is the finding.
If an orderly market-clearing process values the position at, say, sixty cents on the carried dollar, then the economic question is no longer how to preserve the carried dollar. It is whether the sixty cents of current value can earn an adequate return from here. The hold decision is not carry-versus-Treasuries. It is whether the sixty, compounding at 3% to 5%, beats what the same sixty earns anywhere else. It does not. Even at the top of that band, five percent against a 5.3% long bond is a shortfall before anyone prices the equity risk, the illiquidity, fees, or fund-life constraints.
A hold that requires the multiple to expand and rates to fall is an unpriced macro bet, not a plan.
The objection worth answering
None of this proves that today’s yield level is permanent. It proves something more useful for underwriting: a return to ZIRP cannot responsibly remain the base case. In the Federal Reserve’s daily constant-maturity series, the 10-year Treasury averaged 5.80% from 1990 through 2007 and 2.71% from 2008 through 2025. It has traded near 4.7% this year, which is between the two, and closer to the former.
The most compelling counterargument is that the earlier period was one of the strongest stretches in the history of the asset class. Rates averaged well above 5% yet venture generated exceptional returns.
That is true, and it is the strongest argument against my thesis, so let me answer it directly: those funds were underwritten at those rates. Their asset prices, acquisition financing, public-market multiples, LP opportunity costs, and capital availability all cleared in a world where those rates existed.
The 2020 through 2023 vintages were not. The problem was never the level of rates. It is the gap between the rate a position was underwritten at and the rate at which it will be exited. A 2021 mark is not suspect because 5% is high. It becomes suspect because it was struck in a zero-rate regime and has not moved.
The structural case is not that the Fed can never cut. It is that policy-rate cuts need not restore the old long-rate equilibrium. Federal Reserve research describes the recent increase in far-forward nominal rates (which reflect longer-run rate expectations and term premium beyond the current policy cycle) as the largest in decades.
This does not dictate a single yield path, and I am not forecasting one. It widens the plausible range of long-term rates and raises the price of underwriting to the bottom of that range. I do not know where rates go, but I know what your mark assumes.
The portfolio response
Managing partners should conduct a time-boxed re-underwriting of every material ZIRP-era position while the fund still controls the process. The alternative is not that the question goes away. It is that someone else asks it first: an auditor, a valuation committee, a limited partner advisory committee, or the diligence process on the next fund. None of those four will time it to suit the fund. The work answers four questions.
Re-underwrite. What is the company worth using current public comparables, financing costs, dilution assumptions, and a realistic exit calendar, and not using the last preferred round?
Segment. Is the asset a genuine compounder, a fixable operating situation, a cash-yielding harvest, or an impaired position? Different categories require different owners, reserves, and governance intensity.
Choose. Fund to a defined inflection point; change the operating plan or leadership; recapitalize or merge; pursue a strategic sale, secondary, tender, or continuation vehicle; operate for cash; or wind down. Assign an owner, a deadline, and explicit stop-loss conditions.
Communicate. Give LPs the base, upside, and downside cases; the incremental capital required; the expected timing and amount of distributions; and the decision that will follow if milestones are missed.
For a company with genuine liquidity risk, add a 13-week cash forecast and a liquidity-versus-solvency assessment. Do not impose a distressed-company tool on every slow grower. The purpose is to match the intervention to the diagnosis, not to relabel an aging portfolio as a restructuring mandate.
The costliest assumption is not that rates will stay high forever. It is a portfolio plan that requires ZIRP to return.
The long bond has already withdrawn that assumption from the market. Withdraw it from your underwriting while the choice is still yours.
Originally published on LinkedIn on August 25, 2026.