Financing the Build-Out Era
What history teaches us about the future of financing capital-intensive innovation
We’ve been here before
Things have a way of ending up back where they started. For the last fifteen or so years during the ZIRP era, venture capital thrived by sprinkling modest (and not-so-modest) checks into deterministic software companies that could spin up AWS instances on a credit card and scale before the next board meeting. Cheap money, huge addressable markets, and fast exits made the ten‑year closed‑end fund feel like an immutable law of physics.1
But capital has a cost again, software markets appear shakier, and a lot of the real‑world problems we still need to solve — from food security to biomedicine to climate resilience — require things like heavy iron, long lab cycles, and physical infrastructure. In other words, these ingredients look nothing like most of the AI app layer currently taking the lion’s share of venture funding.2
When an old playbook starts to look worn, perhaps it is time to dust off an older one.
The Allocator’s Dilemma: why the modern VC model breaks down outside pure-play software
As venture became a more mature asset class and new funds emerged with regularity over the last decade, more and more fund managers began to explore investing in software startups for complex, legacy industries. Typically, some precipitating factor creates a (real or perceived) inflection point for an industry. This could be something like:
“The passage of the Affordable Care Act will create regulatory tailwinds for healthcare services startups”
“Monsanto acquiring Climate Corporation means we can have high exit multiples for farm management startups”
The Sharing Economy Meme (“let’s Uber-ize everything”) became a weekly LinkedIn slop post from a fund desperate to back asset-lite businesses.
Something something something “we wanted flying cars…”
While all of these factors on their face may have been true at some point in time, the memes became reality and nearly an entire generation of capital allocators came of age having never seen a real downturn, never investing in CapEx-heavy companies, and never encountering the sting of interest rates above 2%.3
As Michael Dempsey at Compound VC so crisply described elements of this shift:
Venture capital is, in my opinion, the highest consensus, lowest conviction financial asset class in human history. And so what that means is that conviction pushes people into consensus quite quickly.
And part of this might be that more mature financial markets kind of implicitly, if you have consensus, you just can’t make money…You must by definition be somewhat high conviction and somewhat anti-consensus.
At some point in the cycle, the Allocator’s Dilemma materializes, driven by some combination of factors.
The Duration Mismatch
Field trials for biologicals can take multiple growing seasons to move the needle, even for early-adopter farmers. It takes three to seven years to commercialize a Class III medical device from concept through FDA approval. A standard software-focused VC fund has already deployed reserves and is gearing up for its next raise long before those technologies hit commercial scale. When one has too much software in the diet for too long, VC Brain simply cannot comprehend what the commercial path looks like for capital-hungry opportunities.
The Ownership Illusion
Look at how a $25 million seed fund’s math really works: after fees, expenses, and follow‑ons, the initial check is about $500K for 5 % ownership.4 Miss a pro‑rata and the stake can plausibly melt to 2%. In capital‑intensive sectors, later rounds are bigger, dilution is nastier, and the path to DPI (distributions to paid‑in) stretches well past fund life.5
The Megaround Mirage
Median figures for biotech financing rounds in all quarters of 2024 were at or above $100 million. Far from de‑risking science, they concentrate talent and capital in a few fashionable bets (often consensus therapeutic areas), crowding out early‑stage diversity and turning a single failed Phase II trial into a systemic shock for all but the largest asset managers. The milestone‑based financing discipline that made 1970s VC so powerful is fraying.
The Exit Compression
While some industries offer the benefit of both rapid feedback loops and high margins, other sectors (e.g. healthcare services or certain areas of industrial goods, like AgTech) routinely face good-but-not-great margins due to labor costs or the nature of commodity markets.6 Where this becomes a problem is when it comes time for ventures in these spaces to graduate (e.g. IPO or acquisition).
In AgTech, exits rarely hit nine figures, let alone deliver the 10× outcome a seed investor dreams of. Looking further back, over the last 20 years, the history of all IPOs or M&A above $1B — not mark-ups, but liquidity events — has remained roughly constant (around 20 a year).7 AgTech accounts for at most two of those.8 As a longtime biotech investor once shared with me, “Compared to biomedical sciences, AgTech takes twice as long with half the end-stage valuations.” When the ceiling is lower and the runway is longer, the power‑law math that justifies spray‑and‑pray (or, really, almost any early-stage investment) falls apart.
Paradoxically, investing in asset-lite software-as-a-service platforms is now riskier because of the increased competition and lack of innovation. Unlike biotech, in which a company in any particular therapeutic area might have half a dozen to a dozen potential acquirers, AgTech startups typically have fewer than half-a-dozen acquisition options, primarily the major input and machinery companies. These startups find themselves between a rock and a hard place, since the prevailing wisdom is that they also likely need these larger incumbents for distribution (or, at a minimum, to not gatekeep).
Healthcare is a similar story. There have been a handful of truly-generational companies (Omada, Oscar, etc.) that have navigated multiple market cycles; however, the telehealth M&A of the 2018-2022 era resulted in significant goodwill impairments in the public markets. While certainly better than the dearth of exits in AgTech, the performance of both tech-enabled services and pure-play software ventures has been underwhelming.
The Dearth of Risk Capital
AgTech along with many other “hard” sectors underwent a massive swing in capital allocation over the last few years. While there are glimmers of light in healthcare and biotech again, and startups in aerospace and defense (“American Dynamism”) are reviving certain elements of the industrials sector, the funding landscape is still fairly barren. The tourist generalist VCs are gone (having been burned by bad investments, or simply closing up shop themselves). Growth-stage funds won’t touch much of the sector due to slow growth or improbability of exits on their timelines (and the general decline of growth-stage VC outside of mega-funds).
Many of the corporate venture funds in these spaces are also strained, due to both macroeconomic pressures along with the end of multiple commodity supercycles and the push towards deglobalization. What is left is a rather small and largely unappealing set of financing options for startups with some growth or commercial success, but that don’t have traditional SaaS-style annual growth.
The result is that industry players still encourage startups to grow at AI-like breakneck pace, as if they can simply “play the game on the field.” It’s the financial and operational equivalent of being a NASCAR driver and slamming into the wall. Instead, they need to be focused on the road ahead, which likely requires a return to long-term-thinking and evaluating the commercial and financing paths over the first 40 months of a venture.
The conversation has already shifted
The idea of making money solely through trading small sums of equity for cash isn’t what venture originally was. It will also not survive the decade. The future is fewer funds, more products, across the whole capitalization and corporate lifecycle. Josh Wolfe at Lux Capital expects a contraction of 30-50% of funds over the coming years; it’s unlikely that the next generation of fund managers would fill their place with the same fund models.9
“If one asks how exactly VCs do that they do, it is not clear that the answer today is much different from half a century ago. The dominant form of organization is still the limited partnership with an ephemeral fund life, even though this places constraints on the time scale of investment returns. Although there have been some organizational structure and strategy innovation, these have been paradoxically rare in an industry that finances radical change.”
― Tom Nicholas, VC: An American History
Venture is also bifurcating into seed stage funds and mega-fund beta-harvesters and asset managers (General Catalyst now owns a hospital, Andreessen Horowitz makes more in management fees than most seed funds have in AUM, etc.). It’s unclear how this plays out for the asset class, but it is clear this is local maximum with the risk of self-destruction.
The emerging narrative is littered with investors admitting that conventional term sheets feel brain‑dead for ventures in the applied science and advanced technology realms. Some argue we need a new asset class focused on underwriting $300 million “doubles and triples,” not chasing unicorns.10 Others call venture a blunt instrument for non‑software businesses and think we need more PE‑style creativity at seed.
There’s reason to think most of the AI-enabled roll-up strategies will fail. However, a larger problem is that most generalist venture funds today operate more like small-cap PE funds who don’t know how to actually operate a business, and most PE funds don’t have the in-house expertise or risk appetite for high-risk, capital-intensive technology investing.
Packy McCormick put it bluntly: capital intensity isn’t the enemy; where the capital lives is. SpaceX, BridgeBio, and others are proof that hardware, biology, and other science-based ventures can be wildly capital‑efficient when financed the right way. OpenAI’s capital structure should be commended as much as its underlying technology — this is a case study in matching capital form with technical ambition.
Many of the outsider funds entering these sectors (and those that have over the last few years) will likely make the same mistakes as the last cycle by not being original enough; they will wind up investing in incremental technology while having less ownership at exit.11 For most, the returns simply will not be there.12
Funds willing to back more unconventional models can likely make the next generation of science-based startups work. This will not look like the last few cycles of AgTech, for example, where most of the founders and most of the investors made bets on incrementalism, largely due to the risk-averse culture of the sector.13
Enter: Merchant Banking 2.0
The history of financing new ventures begins long before the days of the limited partnership venture fund model of the 1950s. For several hundred years prior to that, the groups that funded new innovation were often merchant banks, often closely-held private firms and what would now be called family offices or investment banks. If traditional venture is a sprint relay, merchant banking is something akin to the Badwater ultramarathon with a pit crew.
Historically, merchant banks combined advisory work, board‑level stewardship, and their own balance sheet to partner with family‑owned or capital‑starved companies over decades. Firms like Barings Bank, Rothschild & Co., and Hambros became household names in elite circles. In modern dress, this is a firm like Allen & Co., which over several decades built a reputation as the premier TMT boutique bank.14
What makes Allen & Co. special is that despite its size (at less than two hundred employees) it seemingly finds a way to provide counsel on nearly every major media or telecom transaction, even if that means sharing the tombstone with peers like Goldman or others. By seeing every possible deal, providing advisory services to some, and selectively investing its own capital in a select few, the firm can simultaneously operate with a long-term orientation while also being intimately involved with its clients’ management teams.
The common thread across merchant banks: no forced exit clock, skin in the game, and services that go beyond wiring a check.15
The multi-product fund structure
For those who aspire to fund capital-intensive endeavors, one path forward is for a structure that is part principal investor, part strategic advisor, part project‑finance arranger. This might include:
Evergreen or Interval Capital: Matches 12‑15+ year technology deployment arcs without hostage‑taking LPs.
Concentrated Bets: Sourcing wide (potentially through advisory work) and then maniacally doubling down on the promising ventures. High involvement, less dilution creep.
Integrated Stack: Equity + asset‑backed loans, revenue‑share notes, JV project vehicles — all mechanisms to fund the factory as well as the IP.16
Operator Bench: Those involved, whether in capital allocation or operator roles, will likely not look like they came out of Bay Area VC central casting. Instead, this is likely some combination of in‑house regulatory, go‑to‑market, and M&A talent that can move directly into fractional roles with clients.
The pitch to founders goes something like:
“We can fund your whole lifecycle” — Raise once, not five times. The merchant bank can sequence equity, credit, and revenue‑share as milestones unfold.
“We will walk through the gates of hell with you” — Need someone who has navigated clinical trial design or successfully negotiated offtake arrangements with an OEM? They’re on payroll, and ready to join your management team. It’s a return to Richard Rainwater’s the 3 C’s: Capital, Connections, and Culture.
“We know how to run a process for this” — When it’s time to spin up a project finance vehicle for a pilot plant, the founders can feel confident the merchant bank can put a term sheet together and syndicate it accordingly.
The pitch to LPs goes something like:
“We’re going to build an institution that is a long-term steward of your capital” — where GPs agree to a lucrative, but longer-term-oriented compensation package in exchange for relaxed duration constraints.
“We will go anywhere, do anything, see everything in our sectors, and invest in a small subset” — creating a large funnel not only keeps the lights on via advisory work, but minimizes the likelihood of missing a chance to invest in a generational investment.
“We can stitch opportunities together where traditional venture or private equity often struggles” — funds like Cibus show how to marry investing in both real assets and growth-stage technology that is accretive to the latter. The objective here is to move closer and closer to the metaphorical cash register, while finding ways to integrate and create leverage across previously-disjoint elements of a value chain.17
Some common objections
“It’s too complicated.”
So were SAFEs once. Templates now exist for revenue‑share, capped‑profit units, and asset‑backed venture loans, assuming a fund doesn’t want to do bespoke products. Complexity is a one‑time tax; illiquid, mis‑timed exits are a recurring fee. Infrastructure PE investing wasn’t a thing until a few decades ago, and now it is itself a mature asset class.
“Founders will balk at financiers and active investors on the cap table.”
Not if those bankers get their upside the same way founders do - by building, as opposed to passively investing and/or flipping. It is instructive to look at Thrive and Sutter Hill Ventures as possible alternatives models to the modal venture investor today: hands-on, actually involved, and with an ethos of company-building and co-creation.
“Founders don’t want to give up that much equity.”
New England biotech founders would like a word! Also, if anything, identifying non-dilutive sources of capital (asset-based financing, biobucks, etc.) removes some of the need to do the standard swap of more equity dilution in exchange for cash.
“LPs need ten‑year liquidity.”
Interval funds already manage it in real estate and private credit. Secondary windows plus asset‑backed coupons can provide early cash flow without fire sales.
“I’d be violating my LP Agreement with the structures you proposed”
Without passing too much judgment, it’s striking the high regard LPs hold their allocators in this situation: limiting them to effectively playing a game of Mad Libs with approximately a dozen or so variables on a standard NVCA term sheet template, but I digress.18
The history of the asset class shows that the greats like Don Valentine, Tom Perkins, and David Marquardt were instrumental in being actively involved in building, managing, and financing their winners. Today’s venture fund would be a totally foreign model to this cohort, who often functioned as investor-operators and interim Chief Business Officers in the early days of companies like Microsoft, Cisco, and Genentech.
Concentration is too risky; VC observes the power law”
Concentration also means higher ownership positions, removing some of the dilutive effects from the spray-and-pray model. It allows for aggressively doubling-down on winners. Given the relatively stagnant number of billion-dollar-plus IPOs over the last several decades, if you, as a fund or LP didn’t invest in those particular companies (largely generalist software firms), it’s difficult to see how you’d return a fund given recent fund sizes.
A benefit of the merchant bank model means you’re more likely to truly “see everything” in your circle of competence, investing somewhat widely via advisory work, while plowing your balance sheet into the few companies that will be generationally important.
Moreover, the idea of a permanent capital vehicle models means you’re less reliant on later-stage venture mark-ups and can take the time to build compelling businesses. The way to de-risk will be on staging capital deployment for any one investment, tied to specific milestones, and being creative with the capital stack.
A better future, if we want it
This isn’t about replacing venture capital; it’s about a return to funding a variety of business models, each of which requires a distinct blend from the capital stack. Software‑native VCs will still back the next Plaid. Traditionalist seed investors in “hard” sectors can still target lower-risk, lower-return startups that might each yield 3-5x returns as opposed to putting all their eggs in one high-risk basket. But the world also needs the next Corteva‑size biologicals platform and the next Boston Scientific med‑robotics giant and the next Standard Oil in clean energy. Those companies won’t be built on convertible notes, weekend hackathons, and “lean startup” methodologies.
The question of how to most effectively and efficiently route capital to innovators and founders is one of the most important things our society must solve in order to enable human flourishing. Prior financial innovations made our world materially better, funding new products and technologies responsible for lifting billions of people into modernity. Fixing the current stasis in funding models will lift billions more.
The above approach offers but one way to take bigger (more calculated) swings, and stay in the fight long enough to win — whether in health, food, energy, or industrials. In an era no longer buoyed by money-printer memes and stable economic growth, that might be the surest path to outsized returns, and, more importantly, outsized impact.
So when the next founder walks in with a plan to decarbonize fertilizer or treat a neurodegenerative disease, maybe the first question shouldn’t be, “What’s their pre‑money?”
Maybe it should be, “How might we partner for the long haul?”
Fast exits routinely meant acquisition by an incumbent Big Tech firm, rather than IPO, but liquidity is liquidity, I suppose.
Or the B2B SaaS playbook that a whole bunch of venture funds now update quarterly for AI app layer startups.
I’m excluding from this era both biotech funds and the important, but small, group of funds that truly made investments in applied science startups between. They funded “deep tech” before that became The Current Thing.
This assumes the valuation isn’t asinine, so, yes, the joke is on me in 2025.
A good historical example of this dynamic is in biotech, where the company risk profile and commercial path means management team upfront dilution is much higher (with exit values much lower) than SaaS, for example.
I’ve long-since accepted that my career across healthcare and AgTech ventures mirrors my endurance running hobby: long, fuzzy feedback loops that reward the borderline-masochistic willingness to stay the course.
Sequoia independently ran these numbers, too, and I’m going to trust Roelof Botha on this more than my own analysis.
This depends on how you classify Trimble / AGCO and the acquisition price net of fees for Climate Corporation / Monsanto.
Of course, Josh and the Lux team, with a multi-billion-dollar AUM, are being self-serving with very-convenient statements that benefit their fund size and structure; it doesn’t mean they’re inherently wrong, however.
Again, there are parallels that can be drawn here to the traditional and successful “financier” approach to biotech, but seeing 2021 repeat again in 2023-2025 suggests the modal venture investor has learned approximately nothing.
It’s possible the only two archetypical structures that can do this successfully will be either mega-funds with ginormous teams, or low-overhead boutiques catering to an extremely narrow slice of generational companies. Underwriting itself is not the hard part; access to quality investments is.
At least until the bursting of the AI bubble also leads to a broader end to the 15-year venture mega-cycle.
Risk aversion applies across the whole sector, from farmer to investor (CVC with natural bureaucratic risk aversion, generalist VC with an aversion to being burned again, etc.) to acquirers.
It should be the goal of every capital allocator to emulate Herb Allen’s firm, which, to this day, still lacks a public-facing website.
I’m not referring to the services that every venture fund claims to have like a “platform support” person or staff, or some nominal “jobs” page on their website. If anything, these platform roles are such an antipattern than LPs should ask why their management fees are being used on those line items. I’m referring to much more complex legal, financial, and operational guidance, which traditional PE or banking excels at.
It should go without saying, but equity financing will still be necessary to fund the R&D work, whereas the deployment phase can plausibly be funded by debt or hybrid structures.
A particularly appealing factor for LPs that want to see both financial returns and some sort of “impact” factor.
LPs flocking to the safety of the mega-funds is equally at fault here; one cannot simply blame GPs alone for LP herd mentality.