Neither is automatically better. Venture capital fits companies chasing fast, uncertain growth where early profits make no sense. Revenue-based financing fits companies with steady, predictable sales that want cash without giving up shares. Dictionary.com defines a venture as "a risky undertaking," and that risk cuts both ways — founders risk control, investors risk the whole check. This piece breaks down how each option works, what it really costs, and which businesses each one fits.

How does venture capital actually work?

A venture fund raises money from institutions and wealthy individuals, then invests it in young companies with high growth potential. In exchange, the fund receives preferred stock — shares with special rights spelled out in a term sheet. Those rights usually include a liquidation preference, meaning the investor gets paid before common shareholders if the company is sold. Our explainer on liquidation preferences covers why that single term decides who gets paid first. We covered a connected angle in Liquidation Preferences: The Term That Decides Who Gets Paid First.

The fund does not expect most of its investments to succeed. The model works if a few companies return many times the investment, which is why venture investors push for aggressive growth over early profitability. It also explains the strings attached: board seats, protective provisions, and pro-rata rights in future rounds. Founders who take venture money are signing up for a follow-on process too — each round sets expectations for the next, as our guide to running a Series A process shows. Readers following this should also see Bridge Rounds: The Runway Math Before You Take One.

What this means in practice: venture capital is cheap money if the company becomes enormous, and expensive money if it doesn't. The dilution is permanent. There is no repayment schedule, but there is an implicit one — the pressure to grow fast enough to raise again.

How does revenue-based financing work?

Revenue-based financing, sometimes called revenue-based loans or royalty financing, gives a company cash up front. The company repays a fixed multiple of that amount — the multiple is set in the contract — by remitting an agreed percentage of monthly revenue until the total is paid. When revenue is strong, the balance clears faster. When revenue dips, payments shrink, because the payment is a percentage, not a fixed bill.

Three features define the structure. First, no equity changes hands, so founders keep their ownership and their cap table stays clean. Second, repayment is tied to revenue, which cushions slow months — a fixed-term loan does not do that. Third, the total cost is known in advance: you know exactly what multiple you will repay, unlike a venture round where the true cost depends on what the company is worth later.

The trade-off is capacity. Because repayment comes out of revenue, the structure only works for businesses that already generate consistent sales. A pre-revenue startup cannot service revenue-based payments at all, which is why the instrument lives almost entirely outside the classic venture pipeline.

What does each option really cost?

This is where the comparison gets honest, because the two structures price risk in completely different currencies.

Our analysis: founders often compare the headline numbers and miss the asymmetry. Venture capital's cost is invisible until an exit, then it is enormous. Revenue-based financing's cost is visible and capped from day one, but it drains cash exactly when the business is trying to reinvest. A company growing 10% a month may find that servicing a revenue-based obligation starves the very growth that made it attractive. A company with flat revenue may find venture money buys a runway it cannot justify.

Who should take venture capital — and who shouldn't?

Venture capital fits businesses where the prize is winner-take-most and speed decides the winner: marketplaces, network-effect software, categories where being second means being irrelevant. It also fits companies building something that needs years of spending before revenue appears — deep tech, biotech, frontier AI. The investor is underwriting the outcome, not the cash flow. Our coverage of megarounds, like the terms behind OpenAI's $40 billion round, shows how far that model stretches when investors believe the outcome is historic.

Venture capital does not fit businesses with modest ceilings. A profitable services firm, a niche e-commerce brand, or a company whose founders want to keep control should not sell preferred stock with liquidation preferences to chase growth it does not need. The obligation runs forward: once you raise a priced round, the expectations compound, and a miss can force a down round — see what a down round does to a cap table for how badly that reprices everyone.

Who should choose revenue-based financing?

Revenue-based financing fits companies with three traits: predictable revenue, gross margins healthy enough to absorb the remittance, and a use of cash with a clear, near-term return — inventory, customer acquisition, expansion into a proven channel. SaaS companies with stable subscriptions and e-commerce brands with repeatable ad economics are the classic candidates.

It does not fit companies that need to spend heavily for years before earning anything, and it does not fit companies whose revenue is lumpy or seasonal enough to make remittances unpredictable. It also does not replace a round for companies that need an investor's network, hiring help, or credibility. Money is the only thing a revenue-based provider sells.

Practical steps before signing either structure:

1. Model the full cost. For a round, model dilution at several exit values. For revenue-based financing, model the multiple against your actual margin.
2. Stress-test repayment. If revenue drops for two quarters, can you still remit without cutting growth spending to zero?
3. Read the control terms. Board composition, protective provisions, and any personal guarantees matter more than headline amounts.
4. Check the follow-on path. Venture money commits you to a fundraising cadence; revenue-based money commits you to a repayment schedule. Know which commitment you can keep.

What this means for your decision

The evidence supports a simple split. Venture capital is a bet on a huge, uncertain outcome, paid for in ownership. Revenue-based financing is a bet on a steady, known outcome, paid for in cash flow. If your business needs speed to win a big market and can survive years without profits, venture capital is the tool. If your business already earns money and you want to accelerate it without selling shares, revenue-based financing is the tool. Many companies use both at different stages — venture capital to build the engine, revenue-based financing to fuel it once it runs. What remains unknown for any specific founder is the exit value and the revenue curve, and no structure protects you from guessing those wrong. Price your own uncertainty before you price the money.

Sources: merriam-webster.com · dictionary.cambridge.org · capitalone.com · dictionary.com

The mechanics matter more here than at any later stage. A bad term on a friends-and-family check can resurface years later and complicate a Series A. This explainer walks through the round in order: who invests, what documents get signed, how the money converts into shares, and what a founder should have ready before asking.

Who actually writes pre-seed checks?

The earliest money comes from people who are betting on the founder, not the spreadsheet. The typical ladder looks like this:

• Friends and family. Personal contacts who invest because they know you. These checks are often the smallest and the least documented, which is exactly the problem.
• Angel investors. Individuals, often current or former operators, who invest small amounts across many companies and can also open doors.
• Pre-seed funds and accelerators. Small institutional checks, sometimes bundled with a program, mentorship, and a demo day.
• Early seed funds. Some larger funds will take a first look at a company here, though they more often wait for traction.

Each group wants something slightly different. Friends and family want to help you. Angels want asymmetric upside. Funds want a repeatable pattern. A founder who understands which motivation is in the room negotiates better.

What documents does a pre-seed round use?

Most pre-seed rounds avoid setting a valuation at all. The standard instruments are simple agreements for future equity, known as SAFEs, and convertible notes. Both defer the hard question — what is this company worth? — until a later priced round.

A SAFE is a short agreement that converts into shares when a priced round happens, usually at a discount or with a valuation cap that rewards early investors. A convertible note does the same job but is structured as debt with an interest rate and a maturity date. The trade-offs between the two are covered in detail in our comparison of Convertible Notes vs SAFE Notes: Which Instrument Founders Actually Choose.

Occasionally a pre-seed round is priced: investors and founders agree a valuation and sell actual shares. This is cleaner on the cap table but slower to negotiate, and at this stage the company often has too little data to defend any number.

What this means in practice: the instrument you pick at pre-seed determines how much of the company the earliest backers own once the dust settles. Caps set too high barely reward early risk. Caps set too low can hand away more equity than founders realise until the priced round forces the conversion.

How does the round actually come together, step by step?

The process is less formal than later rounds, but it has a shape:

1. Define the milestone. Work out the one proof point the money must buy — a working product, a first set of paying customers, a waitlist that converts. Everything else follows from this.
2. Size the raise to the runway. Estimate monthly costs, add a buffer, and raise enough to reach the milestone with months to spare. The runway logic is the same as in larger bridge financing, which we break down in Bridge Rounds: The Runway Math Before You Take One.
3. Line up the first commitments. Early checks attract later ones. A first committed angel, even a small one, gives other investors social proof.
4. Standardise the paperwork. Use a consistent instrument and consistent caps across investors. A cap table with five different terms from five different uncles is a due-diligence problem waiting to happen.
5. Close in waves. Most pre-seed rounds close on a rolling basis rather than on a single date. Money wires in as commitments firm up.
6. File and track. Record every agreement, every cap, every conversion trigger. Future investors will ask for all of it.

What do pre-seed investors look for?

With no revenue history and often no product, investors at this stage weigh a handful of things:

• Founder-market fit. Does this team have an unfair advantage in this specific problem — domain knowledge, distribution, or a technical edge?
• A sharp problem. Not a feature, a problem painful enough that someone would pay before the product is polished.
• A credible path to a seed round. Pre-seed money is a bridge, and investors want to see the next round is reachable with what this raise buys.
• Reasonable terms. A clean cap table and standard documents signal a founder who understands the game.

Notice what is absent: detailed financial projections. Nobody at this stage believes a five-year model. What they assess is whether the team can learn faster than the money runs out.

What can go wrong in a pre-seed round?

The failure modes are well known and mostly self-inflicted:

• Over-raising on friendly money. A large friends-and-family round at a generous cap feels free at the time and expensive later, when the conversion dilutes the founders more than expected.
• Sloppy documentation. Handshake deals and inconsistent terms surface in diligence for the seed round and slow it down, or kill it.
• Raising against the wrong milestone. Money spent on offices and headcount before the core hypothesis is tested leaves the company raising again from a weaker position.
• Too many small investors. Fifty angels on a SAFE means fifty signatures and fifty questions at the next round. Consolidation helps.

Our analysis across the funding lifecycle is consistent on one point: the cheapest mistakes to fix are the ones fixed before the wire hits. Once terms are signed, they only get renegotiated under duress — the dynamic that drives down rounds, which we examine in What a Down Round Actually Does to Your Cap Table, Ratchet by Ratchet.

Where does pre-seed fit in the longer funding arc?

Pre-seed is the first rung on a ladder that runs through seed, Series A and beyond, each stage buying a different kind of proof. The stages later in the sequence — priced rounds, liquidation preferences, board seats — are covered in our guide to Liquidation Preferences: The Term That Decides Who Gets Paid First, and the full landscape of rounds and instruments lives in our funding section. Readers following this should also see Liquidation Preferences: The Term That Decides Who Gets Paid First.

The through-line is simple: every later round inherits the terms set here. Pre-seed is where the cap table's DNA is written. Founders who treat it casually spend their Series A paying for it; founders who keep it clean spend their Series A building.

What the evidence of how companies actually progress supports is this: the pre-seed round's job is narrow. Buy one proof point, on standard terms, with documented agreements, from investors who understand what stage risk means. Everything else a startup needs is earned later, round by round.

Sources: innerbody.com · en.wikipedia.org · anatomy.app

The funding of science works differently from venture funding in almost every respect. There is no term sheet, no cap table and no exit. The currency is peer approval, the timeline is set by budget cycles, and the payoff is knowledge that may take decades to become a product. Understanding how the money moves explains a lot about why basic research looks sluggish from the outside — and why it produces things venture capital rarely will. We covered a connected angle in What a Down Round Actually Does to Your Cap Table, Ratchet by Ratchet.

Where does the money start?

Most public research money begins as a line in a government budget. Elected officials allocate it to science agencies, and those agencies decide how to split it across fields, programs and competitions. The agencies do not run experiments themselves. They fund them, mostly at universities and research institutes, through competitive grants.

The logic is straightforward. Taxpayers pay for research that markets underfund. A drug mechanism discovered in a university lab has no revenue for a decade, so no private investor will carry it alone. Public money covers that early stretch, on the theory that society captures the value later — in medicines, materials and trained scientists — even when no single company can.

This is also where the pipeline gets political. Budget priorities shift with administrations, and an agency's priorities shift with them. A lab's funding can rise or fall on decisions made two buildings away from any bench.

How does a grant actually reach a lab bench?

Usually through a university, and rarely in full. A professor writes a proposal, the agency's reviewers score it, and if it wins, the award is made to the institution — not to the individual. The university then takes a share off the top for what are called indirect costs: building maintenance, utilities, compliance staff, grant administrators. What remains is the professor's operating budget.

From there the money turns into salaries for graduate students and postdocs, equipment, and supplies. In most academic labs, people are the largest line item. A grant that looks generous on paper can be mostly spoken for before the first purchase order goes out.

The professor's job, in funding terms, is closer to a founder's than most people realize. They raise money continuously, in competition, and their lab's payroll depends on winning the next award before the current one runs out. The difference is that the pitch is a research plan and the investors are volunteer peer reviewers.

Why does everything take so long?

Because the review process is deliberately slow and deliberately skeptical. According to Encyclopaedia Britannica, science is "any system of knowledge that is concerned with the physical world and its phenomena and that entails unbiased observations and systematic experimentation." That definition cuts both ways for funding: the same systematic caution that makes science reliable makes its financing slow. Every claim in a proposal gets scrutinized by other scientists, revisions get requested, and panels meet on fixed calendars — often only once or twice a year for a given program.

Then the clock keeps running after the award. Equipment has procurement rules. Hiring a postdoc can take months. Animal and human-subject studies need ethics approvals before work starts. A three-year grant can lose most of its first year to setup.

Renewals add another layer. Grants are typically short relative to the research they fund, so labs spend a meaningful share of their time writing the next application instead of doing the current work. The pipeline is not just slow at the start; it never fully stops being an application process.

How is this different from venture funding?

Almost everything. A startup raises money against a growth story and spends it fast, because speed is the product. A lab spends slowly, because rigor is the product. The comparison is worth spelling out:

• Who decides: venture partners decide in weeks; grant panels decide on fixed cycles, often months after submission.
• What's bought: venture capital buys equity and acceleration; a grant buys a bounded research plan with no ownership stake.
• Failure: a failed startup shuts down; a failed hypothesis still gets published, which is the point.
• Follow-on: startups raise bigger rounds on traction; labs win renewals on results and reputation.

The two systems do connect, though. Basic research funded publicly often becomes the science a startup later commercializes, and the people trained on grants become a startup's early technical staff. Readers who follow private rounds — how a bridge round stretches a company's runway, or what a down round does to a valuation — will recognize the underlying pattern: capital is always priced against uncertainty. Public science funding simply prices that uncertainty near zero for the investor, because the investor is everyone. For related coverage, see Bridge Rounds: The Runway Math Before You Take One.

What this means for people watching the startup economy

Three practical takeaways from the way the pipeline is built.

First, timeline claims deserve a discount. When a company says its technology came out of "years of research," the research phase was funded on grant cycles that reward patience, and the commercial phase runs on venture cycles that punish it. The handoff between those two clocks is where a lot of hard-tech timelines slip.

Second, the people matter more than the grant list. A lab's output depends on graduate students and postdocs whose salaries the grants pay. Funding continuity is a hiring story, and hiring is a capability story — the same reason a startup's balance sheet matters more than its press release.

Third, public money is upstream of a surprising amount of private value. Founders and investors who understand where the pipeline starts can read the research landscape earlier than the market does. That is an informational edge, not an endorsement of any company — and it is the same discipline this desk applies to funding coverage generally: follow the money to its source, then check what the source actually bought.

What the pipeline shows

The funding of science is a system designed to be slow on purpose. Money moves through budgets, agencies, peer review and university overhead before it becomes an experiment, and the results are expected to outlast any single grant. That structure trades speed for reliability, and it explains both the frustration and the payoff.

What the record supports is this: the pipeline produces knowledge that markets will not fund on their own, and it trains the people who later build companies on that knowledge. What it does not tell us, in any given case, is how long the gap between discovery and product will run. That gap is set by science, by markets and by luck — and no budget line controls all three.

Which one founders pick depends less on the paperwork and more on who is across the table. SAFEs dominate early-stage and angel deals because they are short, standard and cheap to negotiate. Convertible notes show up more often with bridge investors, venture debt-style lenders and anyone who wants a fixed end date and interest for the wait. The differences are small on paper and large at conversion time.

This piece walks through the four terms that separate them — interest, maturity, valuation caps and conversion mechanics — and where each instrument fits. For the broader contrast with priced equity rounds, see SAFE Notes vs Priced Rounds: Mechanics, Cost, and Control.

What exactly is a convertible note?

A convertible note is debt, full stop. The investor lends money to the company. The note carries an interest rate and a maturity date — the day the loan comes due if no priced round has happened. When a priced round does happen, the note converts into shares, usually at a discount to the round price or at a valuation cap, whichever gives the investor the better deal.

Because it is debt, the note sits on the balance sheet as a liability. If the company runs out of money before the maturity date, note holders are creditors. In a shutdown, they stand ahead of common shareholders in line for whatever is left. That creditor position is the quiet reason some investors prefer notes.

What exactly is a SAFE?

A SAFE — a Simple Agreement for Future Equity — is not debt. There is no interest rate and no maturity date. The investor puts money in now, and the agreement converts into shares at the next priced round, again usually with a discount or a valuation cap. If no round ever happens, the SAFE can simply sit there; it does not come due.

The standard SAFE documents are short and widely used, which is their main selling point. A founder and an angel can sign one in an afternoon without term-sheet negotiation over interest, covenants or repayment. The trade-off is that the investor takes on more risk with fewer protections — no maturity forces a reckoning, and no interest compensates the wait.

How do interest and maturity change the deal?

These are the two terms SAFEs simply do not have, and they matter most when a round takes longer than anyone planned.

• Interest. A note's interest typically accrues and adds to the amount that converts. The investor ends up with slightly more shares than the face amount of the check. A SAFE converts at face value only.
• Maturity. If a note hits its maturity date with no priced round, the note holder has leverage. Depending on the terms, they can demand repayment, negotiate an extension — often with sweeter terms — or in some structures force a conversion. A SAFE holder has no such trigger.

For founders, that maturity date is the risk. A bridge that was supposed to last six months can become a repayment demand during the worst fundraising stretch of the company's life. That scenario is one reason bridge math deserves its own analysis — see Bridge Rounds: The Runway Math Before You Take One.

What do valuation caps actually do?

Both instruments usually carry a valuation cap: the maximum company valuation at which the early investor's money converts. If the next round prices above the cap, the note or SAFE holder converts as if the company were worth only the cap — earning more shares for the same check. If it prices below the cap, the discount usually governs instead.

The cap is where the real money changes hands. A cap set too low means early investors take a larger slice of the next round, and the founders' and employees' ownership shrinks more than the headline round size suggests. Because SAFEs have no interest to offset, founders and investors often bargain harder over the cap itself. Multiple SAFEs with different caps, signed at different times, can stack into a messy conversion waterfall — worth reading alongside What a Down Round Actually Does to Your Cap Table, Ratchet by Ratchet.

Which instrument fits which situation?

The honest answer is that the market has sorted itself into rough lanes.

• Pre-seed and angel rounds: SAFEs, overwhelmingly. The amounts are small, the investors want simplicity, and nobody wants to negotiate a maturity schedule over a $25,000 check.
• Bridge rounds from existing institutional investors: convertible notes more often. The investor is extending credit to a company it already knows, and wants interest and a maturity date as compensation for the risk that the next round never prices.
• Venture-debt-style lenders: notes or note-like structures, always. These are lenders in fact, and debt documents come with covenants SAFEs cannot offer.
• Larger pre-priced rounds with sophisticated parties: either, negotiated term by term.

When a founder wants the loan structure without equity conversion at all, that is a different product entirely — Venture Debt: When a Loan Beats a Round covers where straight debt beats both instruments.

Our analysis: the maturity date is the real decision

Strip away the legal vocabulary and the choice comes down to one question: who carries the risk if the next round is late? With a SAFE, the investor carries it — no interest, no repayment right, no deadline. With a note, the company carries it — the clock runs, interest accrues, and maturity gives the investor a lever.

Founders choosing a note should price that lever in. Ask what happens at maturity, who decides, and what an extension costs. Founders choosing a SAFE should assume the investor knows they are carrying more risk, and expect that to show up in a lower cap or a larger discount. Neither instrument is free; they just bill in different currencies — one in interest and repayment risk, the other in conversion economics.

What the record does not settle is how often notes actually convert at maturity versus get extended, and how that varies by stage. That figure exists somewhere in aggregate fund data, but this desk has not seen a source for it, so we leave it as an open question rather than guess. What is clear is the trend line: standard, short documents have won the early stage, and the note survives where debt-like protections matter — which is exactly where the money is largest and the riskiest.

Practical steps before you sign either one

1. Model the conversion at your realistic next-round price, at the cap, and at the cap with accrued interest. The spread between those numbers is what you are giving away.
2. If you sign a note, write down what happens at maturity and who has to agree to an extension. Ambiguity here favors the creditor.
3. If you stack multiple SAFEs, keep a running list of caps and discounts. Conversion math gets ugly fast when the caps differ.
4. Remember both instruments are temporary. They defer the valuation conversation; they do not eliminate it. The priced round — and its liquidation preferences — is where control is actually settled.

Neither document is a verdict on your company. They are postponements with different interest rates on the waiting. Pick the one whose risks you can actually carry, and know the date you are promising to make the deferral end.

Sources: therighthairstyles.com · latest-hairstyles.com · timeless-hairstyles.com · flawlesshair.com

Down rounds stopped being rare after 2021. The median pre-money valuation for US venture rounds fell for consecutive quarters through 2022-2023 from the 2021 peak, per PitchBook-NVCA Venture Monitor reporting, and repricings became common enough that the mechanics are worth knowing cold.

What triggers anti-dilution adjustment?

The trigger is issuing new shares — usually preferred — at a price below the conversion price of an existing preferred series. When that happens, the older series' conversion price adjusts downward, meaning each old preferred share converts into more common shares than before. The investor's stake grows arithmetically; nobody buys anything. The adjustment is automatic under the charter, not a negotiation reopened out of goodwill.

Two formulas dominate. Broad-based weighted average, the market standard per NVCA documents, scales the adjustment by how many new shares are issued at the low price relative to the company's size. Full ratchet, rarer and harsher, resets the old conversion price to the new round price outright — as if the earlier investor had bought at the down-round price from the start.

How different is broad-based from full ratchet in practice?

The table's last rows are why full ratchet appears mostly where the investor has leverage: bridge rounds into companies that cannot raise otherwise. Founders signing term sheets should read the anti-dilution clause before the valuation clause, because the second number is what the first one does to you.

What else does a down round reset besides price?

Option pools and liquidation preferences, mostly. A new round typically reprices the employee option pool at the low price, which helps new grants but marks existing underwater options — and companies often run a repricing program afterward, exchanging old strikes for the new fair market value, a move that requires board approval and an accounting expense. Liquidation preference stacks can also grow: a down round negotiated with participating preferred or senior preference gives the new money first claim on exit proceeds, compounding the common stockholders' position.

Employee retention is the quiet casualty. A 2021-vintage option struck at the peak is worthless-looking at a quarter of the price, and the fix — the repricing — announces the situation to the whole team at once.

How often do down rounds actually happen?

Cyclically. After the 2021 peak, the share of US venture rounds that were down rounds rose steadily through 2023, with later-stage companies hit hardest — several analyses through 2023 put down rounds at their highest share since the 2008-2009 period, per law firm deal analyses and PitchBook-NVCA Venture Monitor data. The 2008 cohort showed the lag pattern too: repricings cluster 12 to 24 months after valuations peak, because companies raise on the old mark until the money runs out.

Is a down round worse than the alternatives?

Not automatically. The honest alternatives list is short: a bridge round that defers the repricing while spending runway, an aggressive cut to extend runway without new money, a sale, or shutdown. A down round reprices the company truthfully and funds it; a flat round at a stale mark, or debt that dodges the question, can cost more in the end. What the evidence supports: the mechanism is arithmetic, the ownership shift lands on common, and the stigma is pricing a real number instead of defending a stale one. What it does not support is the folk claim that down rounds predict failure — outcome data on repriced companies is mixed and sparse, and the honest line is that the question stays open.

The mechanics behind that kind of reset apply well beyond one fintech company. Anti-dilution clauses, pay-to-play provisions and the choice between a down round and a dressed-up "flat round" show up whenever a startup raises at a lower price than its last one.

What Actually Happens to Existing Investors in a Down Round?

Anti-dilution provisions reset the price at which earlier investors' preferred shares convert into common stock, and they come in two common forms, according to TechCrunch's breakdown of anti-dilution mechanics. A "full ratchet" clause drops the earlier conversion price all the way down to match the new, lower round price — the most aggressive version, and the one that dilutes founders and common shareholders the most. A "weighted average," or broad-based, clause instead blends the old and new prices with a formula tied to how much capital came in at each price, producing a smaller adjustment. Weighted-average terms are the more common structure in U.S. venture deals; full ratchet tends to show up when a company is negotiating from a weaker position.

A separate term, "pay-to-play," punishes investors who sit out the new round rather than protecting the ones who join it. Existing investors who don't put in new money can lose their anti-dilution protection and other preferred rights, effectively forcing a choice between reinvesting or being treated closer to a common shareholder.

Why Do Startups Take Down Rounds Instead of Alternatives?

Founders often resist down rounds because a lower price is public and permanent, showing up in every future pitch deck. Crypto lender BlockFi illustrated the alternative path in 2022, when it sought $100 million at a $1 billion valuation, down from the $3 billion figure it carried just 15 months earlier, according to TechCrunch's reporting at the time.

Some startups instead negotiate a flat round that preserves the old headline valuation but loads the new money with punitive terms: a 3x liquidation preference that pays new investors back three times their money before anyone else sees a dollar, participating preferred stock that stacks a fixed return on top of ordinary upside, or tighter anti-dilution rights. Venture investor Brad Feld has argued the flat-round path is worse for a company, telling TechCrunch that "just doing a clean resetting — at whatever the valuation so that everybody is aligned and dealing with reality — is much, much better for a company" than layering hidden terms onto an inflated number. The tradeoff in that reporting: a down round is harder on morale once employees see their options reprice, but it keeps the terms simple, and it doesn't hand later investors a stack of preferences from a round that never reflected the business.

How Did Klarna's Valuation Actually Move, Round by Round?

The repricing wasn't specific to Klarna. Buy now, pay later stocks fell broadly in 2022 as inflation and rising rates hit consumer-lending valuations across the sector — Affirm's shares were down 77% for the year and Block's down 61% by the time Klarna's round closed, while Apple had just entered the installment-loan market that June, according to CNBC's coverage of the round. Pulled together, Klarna's own numbers and contemporaneous reporting show a round-trip: a peak in 2021, an 85% cut in 2022, and a partial recovery at its 2025 listing that still landed well short of the earlier high.

Who Backed the Down Round, and What Did They Get?

Klarna's $800 million round came from a mix of holdovers and new entrants: existing backers Sequoia Capital, Klarna's founders, Danish retailer Bestseller, Silver Lake and Commonwealth Bank of Australia put in more money, while Abu Dhabi's Mubadala Investment Company and Canada Pension Plan Investment Board joined as new investors, with Goldman Sachs advising, according to Klarna's announcement of the round.

Klarna CEO Sebastian Siemiatkowski framed the round as validation despite the price cut: "It's a testament to the strength of Klarna's business that, during the steepest drop in global stock markets in over fifty years, investors recognized our strong position," he said in the announcement. Sequoia partner Michael Moritz made a similar case, saying Klarna's "business, its position in various markets and its popularity with consumers and merchants are all stronger than at any time." Both are company- and investor-supplied characterizations, not independent findings — the announcement does not disclose revenue or profitability figures alongside them.

What Happened to Klarna's Valuation After the Down Round?

Klarna spent the three years after the down round pushing toward profitability while expanding in the U.S. market with the new capital, per the company's statements at the time of the raise. It went public on the New York Stock Exchange on Sept. 10, 2025, under the ticker KLAR, according to Klarna's own announcement of the listing. Shares priced at $40, implying a valuation of roughly $15.2 billion on a fully diluted basis and raising about $1.37 billion, according to Axios's reporting on the IPO pricing. Shares then closed their first trading day up 14.6%, at $45.82, according to Fortune's coverage of the debut.

That arc — $45.6 billion to $6.7 billion to roughly $15 billion at listing — is what a down round's aftermath can actually look like: a real cut, a real but partial recovery, and a return to public markets at a price still less than half the 2021 peak. Nothing in the public record shows Klarna's value returning to that earlier high.

What is the honest runway math?

The calculation founders should run before signing any bridge. Inputs: months of runway the bridge adds (net of its own raise costs), the milestone it must produce, and what the next round requires that milestone to be. The arithmetic's core test: does the bridge's time plausibly produce the delta between your current metrics and the next round's bar? A company at $600,000 ARR growing 2x annually needs roughly a doubling of both to clear a Series A; a 9-month bridge that ends at $750,000 ARR has bought time and purchased nothing — the next raise happens in a weaker market position with a burnt bridge behind it. The documented failure signature: serial bridges — each one smaller, flatter, and from a narrower set of insiders, each deferring a valuation event the metrics cannot support, until the insiders stop.

When do bridges work?

The documented success cases share a structure: a named milestone with a deadline, an investor writing a real check at real terms, and a next-round thesis that the milestone completes. The canonical good bridges: the revenue quarter — six months to close the enterprise pipeline that converts a $1.2 million ARR story into a $2 million one; the product release — the version that unlocks a buyer category the current product cannot reach; the correction — burn cut deep enough that the bridge reaches default-alive, converting the raise question from existential to optional. In each, the bridge is a project with a deliverable, priced and scheduled like one.

What are the structural costs?

Even good bridges carry terms and signals. Terms: bridge SAFEs stack with the existing SAFE history and convert at the next priced round — a stack of escalating caps from serial bridges can consume a round's dilution before the new money prices it. Insider dynamics: bridge participation is the existing investors' honest vote — the documented pattern of some insiders declining the bridge is read by outside investors at the next round as diligence data, the signaling problem in its purest form. And the market's arithmetic: a company that has bridged once raises the next round's questions — 'why did the A not happen?' — which is survivable with a good answer and fatal without one.

What should founders negotiate in a bridge?

Six items from documented practice. The milestone, in writing, agreed as the bridge's purpose — not a slide's aspiration but a metric with a date. A cap (or discount) that respects reality — serial bridges with escalating caps against flat metrics simply pre-load the next round's collapse. Insider pro-rata discipline — who participates signals the syndicate's view to the market. A length calibrated to the milestone: bridges that are too short convert to serial bridges; too long and the burn they fund produces nothing fundable. Warrants or bonus coverage for bridging investors — market-standard compensation, negotiable in size. And the pre-agreed plan for both outcomes — what happens if the milestone lands (who leads, at what bar) and if it does not (the cut plan, decided before the bridge is signed, not after it fails).

What are the alternatives worth pricing?

Before any bridge, the honest comparison set: a deep cut — reducing burn to extend runway without new capital, which weakens the company but preserves the cap table and the optionality; venture debt — where a real milestone and an asset base make a loan cheaper than a stacked SAFE; a small flat priced round — crystallizing a valuation but clearing the deck for new investors; and the acquisition conversation — the option founders resist exploring until it is too late to negotiate. The bridge is one instrument among these, appropriate exactly when the milestone is real, close, and fundable — and a slow leak otherwise.

The bridge test is a sentence: name the milestone this money buys, the round it unlocks, and the date both arrive. If the sentence fills in cleanly, take the bridge. If it needs adjectives, the bridge is buying adjectives — and those convert at zero.

How does the instrument actually work?

The standard structure: a three-to-four-year term loan, interest-only or light-amortizing early, with repayment from the next equity round's proceeds in the base case; warrants for roughly 5 to 10 percent of the loan amount's worth of equity; and — the load-bearing terms — covenants and triggers. The lender's real underwriting is the cap table: they lend against the company's ability to raise the next round, which is why venture debt follows equity rounds and why lenders want the same investors still supporting the company. Cost, all-in, typically lands well above traditional bank debt once warrants are counted — the honest comparison is not to a bank loan but to the dilution of the equity the debt defers.

When does venture debt clearly make sense?

Four documented use cases. The bridge to a stronger round: six to nine months from a raise that needs one more quarter of metrics — debt buys the quarter, the round prices the improvement, everyone is paid back. The asset-financing case: capital expenditure with recoverable value — equipment, and in the AI era GPU fleets, where the asset secures the loan and the debt matches the asset's life. The non-dilutive top-up: a company near breakeven that wants to delay a flat round without adding a priced down-round event to the record. And acquisition or working-capital needs with known repayment sources. In each, the common denominator is a high-confidence repayment event inside the loan's life — debt against a plan, not against hope.

When is it the wrong tool?

The documented graveyard pattern: companies that took venture debt when equity was actually unavailable — the lender underwrote the last round's investors' continued support, the support evaporated, and the covenants turned. The mechanics of the failure: a missed milestone or a down-round trigger can make the debt due, hand the lender warrant coverage at reset prices, or deliver default economics — and the documented 2022-2024 wave of venture-debt workouts showed lenders converting to control positions in companies whose only remaining asset was the IP. The rule of thumb practitioners print: venture debt extends a working equity story; it does not repair a broken one. If the next round was already doubtful, the debt makes the next chapter a negotiation with a creditor rather than an investor.

What terms deserve the most negotiation?

The triggers, which matter more than the rate. Milestone covenants: what technical or financial event must occur, by when — negotiate triggers the company controls. The MAC clause — material adverse change — which can be sprung in a downturn; narrow its definitions. Minimum cash covenants: the level below which default is automatic; set it at real operating floors, not round-figure comfort. Warrant coverage and strike: post-money mechanics, and whether coverage resets on a down round. And the intercreditor reality: which lender is secured, on what collateral, ahead of whom — the question that decided the 2022-2024 workout outcomes.

How did the market behave in the last cycle?

The documented cycle: venture debt expanded through 2021 as a top-up on cheap equity; in 2022-2023, as equity froze, the debt stack became the stress point — SVB's collapse in March 2023 briefly froze the largest lender out of the market entirely, an event every borrower learned from (concentration risk in lenders is real); through 2024-2025, the market normalized with specialty lenders (Hercules, Horizon, Trinity, Western Technology Investment) and bank platforms repricing risk higher. The AI era's variant: GPU-backed lending — debt against accelerator fleets and inference contracts, the compute era's version of equipment finance, and the financing layer underneath several of the GPU cloud companies.

Venture debt is a good instrument used for its purpose: matching capital to a known repayment event, at the cost of a covenant rather than a board seat. It is a terrible substitute for equity the market will not provide — the lender's patience is contractual, and the contract's triggers are written for the downside.

Who is Ilya Sutskever and why does the bet price his name?

The documented record: co-founder and chief scientist of OpenAI, co-inventor of the sequence-modeling lineage that led to GPT, author of some of the field's most cited papers (the 2014 sequence-to-sequence work with Sutskever, Vinyals and Le being a foundation of modern deep learning). In 2023 he co-led the board's removal of Sam Altman, then backed his return, then left OpenAI in May 2024 and joined the Superalignment safety team's dissolution by departing. That record — technical founder of the field's central company, present at its deepest governance crisis, exiting with concerns he declined to fully specify — is the asset the $32 billion prices. Investors buying SSI are buying the market's most credentialed researcher's claim that he knows something about the path the frontier labs are mismanaging.

What is SSI's actual thesis?

Printed on its own materials: one product, one goal — safe superintelligence — with no interim commercialization, no product releases, no revenue timeline. The business structure documented at founding: a for-profit company headquartered in Palo Alto with a Zurich research arm, designed to avoid the distraction of commercial pressure — the explicit contrast with OpenAI's and Anthropic's product businesses. The company has publicly stated it will not ship anything until the goal is achieved, a stance maintained through 2025 with no product announcements, no model releases, and minimal research publication — a near-total information blackout that the market interpreted as discipline, and skeptics as absence of anything to show.

How does the round compare with the market?

The April 2025 window was the mega-seed's peak: Thinking Machines' $2 billion at $12 billion the same month; SSI's $2 billion at $32 billion setting the valuation ceiling for pre-revenue research companies. The syndicate's composition tells the structural story — Greenoaks leading, with Alphabet and Nvidia reported among earlier backers and the round linking SSI to the same compute-supplier web as the frontier labs. At $32 billion, SSI was valued at more than half of Anthropic's early-2025 mark on the strength of a research agenda; the market's pricing said the distribution of possible outcomes for a Sutskever-led lab is wide enough to justify frontier-lab economics on personnel alone.

What are the documented risks?

An honest list. Time: superintelligence is not a milestone with a calendar, and a company with no revenue and multi-billion compute needs will raise again and again — each round a referendum on progress the company has committed not to demonstrate publicly. Talent concentration: the documented departures of senior researchers to other ventures are part of the 2025 record, and a pure research lab's only assets leave the building nightly. The safety thesis itself: if capabilities plateau or regulation pins the frontier, the premise of a dedicated superintelligence lab deflates with it. And verification: with no products, no benchmarks published, and no disclosures, outside investors are pricing faith in a founder and a small set of insiders' diligence — the least checkable large bet in the industry's history.

What would move the needle?

Observable milestones only: published research establishing a distinct technical direction, compute deals surfacing through supplier disclosures, hiring or departures at the senior level, and the next round's price — the market's periodic vote. Until then SSI remains the purest experiment the venture market has run: can the single most credentialed researcher in the field, given billions and no commercial pressure, produce something the frontier labs with their products and revenue cannot?

The $32 billion is the price of that question. The answer is years away by design, and the bet's elegance — no products, no metrics, no near-term test — is also its exposure.

What are the actual mechanisms?

Three channels. The tender offer: the company runs a structured buyback or facilitates a buyer's purchase of employee shares at a set price, with eligibility rules — tenure, level, per-employee caps — and company control over who sells how much. The company-led tender became the dominant form precisely because it preserves control. The direct secondary: a shareholder sells to an outside buyer — a secondary fund, a high-net-worth vehicle — which almost always requires company consent under the shareholders' agreement's rights of first refusal and transfer restrictions. And the structured programs of the big funds: crossover investors and dedicated secondary funds buying blocks of later-stage companies, often at discounts to the last primary mark. All three routes price against the last primary round, usually at a discount of 10-40 percent, wider in weak markets.

Why do companies control the process so tightly?

Because secondary flow creates problems boards care about. Cap-table hygiene: hundreds of small outside holders complicate future financings and any eventual IPO. Information leakage: buyers of secondaries conduct diligence, and the company does not want its metrics circulating. Signal management: a large discount secondary repricing the company's paper downward is a financing event the company did not choose — the documented 2022-2023 experience, when forced secondaries cleared 40-60 percent below prior primary marks and functioned as de facto down rounds. And retention philosophy: liquidity reduces the golden handcuffs, so companies ration it — eligibility caps of a fraction of vested holdings are the norm, and some companies' documented practice ties tender participation to staying employed.

What are the rules employees actually face?

The standard constraints in U.S. startup equity documents: vested options must usually be exercised before sale — meaning cash for the strike and the tax — though some structured deals exercise-and-sell simultaneously; the company's ROFR lets it (or its chosen buyer) match any outside offer; transfer restrictions prohibit selling to anyone without consent, with violation grounds for forfeiture in aggressive documents; and tender offers are company-scheduled events, not standing rights. Taxes run by instrument: ISOs exercised and sold quickly lose favorable treatment and become disqualifying dispositions; the alternative minimum tax interplay makes exercise-before-sale decisions genuinely nontrivial, which is why the honest advice — get a tax advisor who has seen your specific documents before acting — is boilerplate because it is true.

What does the 2024-2025 market record show?

The AI wave's distinctive pattern: secondaries at eye-watering marks. OpenAI's employee tender — reported at $300 billion in early 2025, one of the largest ever — and the follow-on structured programs around the late-2025 $500 billion restructuring round; Anthropic's tenders reported at $183 billion; xAI, Databricks, Stripe, and Canva all running structured liquidity at multi-billion marks, per Reuters reporting. The concentration matters: the secondary market's volume migrated overwhelmingly to the top of the AI stack, while the median venture-backed company ran no tender at all — liquidity, like everything else in this cycle, bifurcated. Employees at the winners monetized pre-IPO; employees at the middle of the market held illiquid paper through a window that mostly did not open for them.

What should employees and founders take from the mechanics?

Employees: model the after-tax outcome of any tender — the interplay of strike, 409A, and holding periods decides whether selling is even rational; treat eligibility caps as the company's retention tax, priced in advance; and remember that a secondary at a discount tells you the market's honest price, information worth more than the proceeds. Founders: tender policy is retention policy — design it deliberately (frequency, caps, eligibility) rather than reactively; disclose marks honestly, because the employees who sold at discounts while the company's messaging held the old mark remember it; and resist the founder-secondary reflex at early stages, the misalignment documented in the 2021 vintage's aftermath.

Secondaries completed the private market: capital in, liquidity out, no listing required. The machinery works — for shareholders of the companies the market wants, rationed by the companies themselves, at prices the last round flatters.

How is a 409A actually produced?

An independent valuation firm — a specialist shop or an accounting firm — models the company from its own data: financials, projections, comparable public companies, and the terms of the latest preferred round. The standard methods: the market approach (revenue or earnings multiples from comparable public companies, discounted heavily for being private and small), the income approach (a discounted cash flow on the company's own projections, haircut by the probability the projections are fiction), and — the decisive piece — the option pricing model and backsolve: working backward from what investors just paid for preferred shares, subtracting the preferences those investors hold, to derive what common is worth after the preferred stack eats first. The output lands predictably: 409A values commonly sit at roughly 20 to 40 percent of the last round's post-money per-share price, with earlier-stage companies showing wider gaps.

Why is it always lower than the round price?

Because preferred and common are different instruments. Investors paid the round price for preferred stock carrying liquidation preferences, anti-dilution rights, and control terms; common stock sits last in the exit waterfall and votes last in governance. The discount is the measured value of those differences plus illiquidity — common shares in a private company cannot be sold. The discount is not pessimism about the company; it is the arithmetic of the stack. Employees should internalize the corollary: the strike is set low relative to the preferred price precisely so options have value at the bottom of the exit range, which is also where the preference stack does its damage.

When does a company need a new 409A?

The safe-harbor rules let a valuation stand for up to 12 months, but material events reset the clock: a new priced round, an acquisition offer, a material change in projections or business condition. The documented practical rhythm: a fresh 409A after every round, occasionally intra-round when something material happens — a big contract, a product failure, a market repricing. In down markets the resets cut both ways: post-2022, hundreds of companies took lower 409As, which was good news for new option grants — lower strikes — and irrelevant-to-painful for existing grants struck at 2021 prices, now underwater. The 'cheap' strike of a down-market 409A is one of the few documented silver linings of a repriced company.

What should founders know about running the process?

Four practical notes from practice. Timing: commission the 409A immediately after a round closes — the firm prices from the round documents while they are fresh, and option grants wait on it. Choice of firm: the report's defensibility is the product; a discounter's report invites IRS challenge, and the fee difference is trivial against the penalty exposure. Projections discipline: the income approach runs on management's own forecasts, and heroic projections produce higher 409As that look flattering right up until they make every option grant riskier. And board hygiene: the 409A is evidence the board exercised fiduciary care in setting strikes — the grant-date discipline matters as much as the number.

What should employees understand about it?

The 409A is the strike, not the dream: the value of an option is what the exit price does above the strike, and both the strike (set by 409A) and the exit waterfall (governed by preferences) are printed in documents employees are entitled to ask about. The two questions worth asking in any offer: what is the current 409A strike, and what is the total preference stack over common. A company that answers both cleanly is doing equity right; a company that cannot produce the first number is not issuing options lawfully, and one that will not discuss the second is hiding the arithmetic that decides whether the options pay.

The 409A is the startup's most honest number — conservative by statute, backward-looking by method, and deliberately unglamorous. Founders should treat it as plumbing; employees should treat it as the floor of their equity math; and everyone should remember that the headline valuation and the 409A measure different things on purpose.

What actually drove a 3.3x re-rating in seven months?

The documented ingredients. First, the AI product cycle landed in revenue: Databricks' positioning — data platform plus model training and deployment on the customer's own data — is the architecture enterprises chose for AI workloads, and the company reported its AI products as the fastest-growing part of the mix, with model-serving and vector-search consumption scaling through 2025. Second, the metric conversion: the company said it passed a $4 billion revenue run-rate at roughly 60 percent growth — at that combination, the $204 billion price is roughly 50x current revenue, a growth-adjusted multiple the round's investors judged against public comparables trading near 20x with half the growth. Third, the December round's floor: the $62 billion raise included a reported floor on secondary pricing that stabilized the private mark, making the July re-rating a market move from an anchored base rather than a rescue.

What is Databricks, in one paragraph?

A data-and-AI platform built on lakehouse architecture — the company's term for unifying data warehouse and data lake paradigms — selling consumption-based cloud infrastructure plus subscription products for analytics, ETL, governance, and machine learning. Founded 2013 from the Berkeley team behind Apache Spark, commercializing open-source infrastructure and defending the franchise with owned products (Delta Lake, Unity Catalog, the Mosaic AI line acquired in 2023 for a reported ~$1.3 billion) that ride on top. Its long rivalry with Snowflake — the warehouse to Databricks' lakehouse — is the defining competitive structure of the data platform market, and both companies' numbers make it the best-documented private-vs-public comparison in software.

How does the round compare with Snowflake?

The comparison investors priced: Snowflake, public, disclosed fiscal 2025 product revenue of roughly $3.6 billion growing around 29 percent, with its stock trading at roughly 15-20x forward revenue through 2025. Databricks reports a larger, faster-growing base at similar-or-better net retention, but consumes capital differently — it sells consumption infrastructure with cloud-cost pass-throughs, making gross margins lower than Snowflake's ~68-70 percent. The round prices Databricks at a premium multiple to Snowflake on the growth differential and the AI-native positioning; the bear case, documented in analyst commentary, is exactly the margin structure and the consumption model's cyclicality. Both readings are on the record; the private market chose the first.

Why does this round matter beyond Databricks?

Three signals. That AI infrastructure value is concentrating in the data layer: the labs monetize models, but the companies monetizing enterprise data gravity — Databricks, Snowflake, Palantir's resurgence — are the documented winners of enterprise AI spend so far. That the mega-round is no longer only an AI-lab phenomenon: $2 billion for a 12-year-old, revenue-heavy software company redefines what growth equity now funds — effectively late-stage public-private arbitrage, with IPO-ready companies choosing private capital's speed and nondisclosure. And that Thrive Capital's 2025 — this round, OpenAI's $40 billion syndicate, and the reported Anthropic discussions — made it the most consequential single investor of the AI cycle's financing leg.

What happens next on the documented timeline?

The company has said an IPO is a matter of timing, with reporting through 2025 pointing to a listing attempt when market conditions suit — the round's structure, providing liquidity and capital without disclosure obligations, removes any urgency. The watch-items: the Snowflake comparison each quarter (public data versus company-reported), AI product revenue disclosure granularity at the eventual S-1, and whether the 60 percent growth holds as the base passes $5 billion. The mathematics of patience: twelve years of compounding, then a year in which the AI wave tripled the price of the compounding.

Databricks is the counter-narrative of the AI cycle: not a lab, not a wrapper — the boring data plumbing under everything, monetized at the exact moment the models made data the constraint.

How does a SAFE actually convert?

A SAFE converts at the next priced round in one of two flavors. With a valuation cap: the SAFE buys shares at the capped price however high the round prices — a $5 million cap on a company that later prices at $20 million means SAFE holders buy at a 75 percent discount. With a discount instead: typically 10 to 20 percent off the round price, whatever it is. Uncapped, no-discount SAFEs also exist — pure deferred risk, investor-favorable in the other direction. MFN ('most favored nation') clauses let early uncapped SAFEs adopt the terms of any better SAFE signed before conversion. The conversion math compounds: every SAFE's discount dilutes the round's investors, who respond by adjusting the price, which lands on the founders — the mechanism nobody watches until the stack is three layers deep.

What does the deferral actually cost?

The documented failure mode of the 2021-2022 vintage: companies stacked SAFEs at escalating caps — $10 million, $20 million, $40 million — and then the market repriced, the priced round landed at $25 million, and the SAFEs' conversion arithmetic consumed the round. Founders discovered that the effective valuation of their company was set not by the headline of the new round but by the stack beneath it; employees' option pool repriced downward; some rounds became impossible and became bridge extensions instead. The SAFE is cheap at signing and settles its bill at conversion, with interest — the interest being the uncertainty every later investor prices in when they see an unaudited stack.

What does a priced round buy for its cost?

For the legal spend — typically $30,000 to $100,000 across counsel at current rates — a priced round buys certainty: a valuation, a share price that sets the 409A and employee option strikes, a lead investor with a board seat and skin in the game, pro-rata rights that organize future financing, and a clean cap table that the next investor can read in an afternoon. It also buys friction: six to ten weeks instead of days, negotiation over protective provisions, and dilution crystallized at a moment the founders may consider the worst possible timing. The control terms — board composition, veto rights over future raises and sales — are the real transfer in a priced round, and the ones founders should negotiate hardest.

When is each instrument the right one?

The market's working answer: SAFEs for pre-seed and seed rounds below roughly $3 to $5 million, or between priced rounds when speed dominates — the canonical uses are a pre-seed from angels, a seed top-up, or an extension. Priced equity when a lead institution is taking a board seat, when the round is large enough that deferral risk compounds, or when the company needs the certainty — for an employee liquidity program, debt financing collateral, or an acquisition currency. The 2024-2025 market added a practical note: as round sizes inflated, the 'seed' label moved onto $5 million-plus raises, and more of those became priced rounds with SAFEs only for small early money — the instrument following the check size.

What should founders watch in SAFE terms?

Four terms decide everything: the cap (or its absence), the discount, the maturity — modern SAFEs have none, meaning no repayment obligation ever forces the issue — and the post-money versus pre-money framing. The post-money SAFE, Y Combinator's current standard, fixes the investor's ownership percentage at signing, which makes dilution from later SAFEs land entirely on the founders rather than being shared across the stack — a one-paragraph distinction worth real percentage points. Founders should model the full conversion waterfall before signing any SAFE: at three different next-round valuations, who owns what. The model takes an hour; skipping it has cost founders companies.

The choice is not speed versus rigor; it is which questions you answer now versus at conversion — and the interest rate on deferred questions is higher than founders expect.

What does the timeline actually look like?

Weeks minus four to zero: preparation — the metrics deck, the model, the data room, and a warm intro path into 15 to 25 target firms. Weeks one to three: first meetings, run in parallel, all sourced within days of each other so that interest compresses into the same window. Weeks four to six: partner meetings and data-room diligence for the firms that advance. Weeks six to eight: term sheet negotiation and signature. The compression is the strategy: offers that arrive in the same fortnight compete; offers that arrive months apart do not, and the founder negotiates against a stale process instead of a live market.

What materials does the process need?

Four documents, in order of importance. The deck: 12 to 18 slides telling the growth-machine story — problem, wedge, traction curve, why now, team. The metrics: an honest cohort and retention view, pulled from billing data, presented in whatever level of detail diligence will request. The model: 18 to 24 months of use of funds, showing what the money buys and what milestones it reaches — Series A investors are buying the Series B story. And the data room: cap table, contracts, employment agreements, key customer terms, clean before the first meeting, because slow document production is read as operational sloppiness.

Which metrics actually decide the round?

The investor screen runs on five numbers. Growth rate on a real revenue base. Net revenue retention — expansion inside existing customers, where above 110 percent reads as enterprise durability. Cohort payback — months for a customer's gross profit to cover acquisition cost, with under 18 months as the standard mark. Gross margin — software buyers expect 70 percent plus, and AI companies face the documented margin question on inference costs. And concentration — if the top three customers exceed half of ARR, the round needs a story for why that de-risks rather than signals design-partner dependence. Founders should know all five before the first meeting, because every partner meeting will reach them within twenty minutes.

How do you build the target list?

Twenty firms beats fifty. The documented pattern for a good list: half sector-specialists with portfolio adjacencies whose partners have written or spoken about the space; a quarter generalist firms at the right stage; a quarter wildcards — firms that have done the most recent comparable rounds, listed in press coverage. Warm intro requirements mean the founder works backwards from who can introduce them; the best intro comes from a portfolio founder, the worst from a cold email. Track everything in a pipeline — firm, partner, stage, next step, date — and treat a two-week silence after a partner meeting as the answer it is.

What kills processes, on the documented record?

The recurring failure modes: leaking the raise to a current investor whose terms arrive before the market competes; meeting firms serially so no window ever compresses; a metric that fails diligence late — the surprise churn cohort discovered in week five; founder indecision between two term sheets while both go cold; and accepting a term sheet without checking the investor's references, then discovering the partner's board style in the first board meeting. Each is avoidable with process discipline, and each ends processes that had live interest.

How do you evaluate the term sheet?

Price is one term of roughly ten that matter: board composition, liquidation preferences, the option pool shuffle (whether the pool is created before or after the new money — a pre-money pool costs the founders percentage points), pro-rata rights, information rights, and the partner's actual availability. Check references with three founders the partner has backed through a hard year, not a good one. The right lead investor at a fair price beats the highest price with a distracted or disengaged board member, and the founders who learn that distinction early are the ones whose Series B raises start from a functioning board.

The Series A is the first fundraise where the company is being bought, not backed. Run it like the sale it is: prepared, parallel, compressed, and decided on evidence.

How does a 1x preference work in a sale?

Start with the clean case. A company raises $50 million at a $200 million post-money valuation, investors taking 25 percent with a standard 1x non-participating preference. The company sells for $500 million. The investors choose the larger of their preference ($50 million) or their ownership share (25 percent of $500 million = $125 million) — they take $125 million, and everyone's paperwork matches the headline math. Now the same company sells for $60 million. The investors take their $50 million off the top; the remaining $10 million is split among everyone else. Founders who owned half the company on paper collect $5 million, not $30 million. The preference did that.

What is the difference between participating and non-participating?

Non-participating preferred, the modern default, pays the investor either the preference or the converted-to-common share, whichever is greater. Participating preferred pays both: the preference first, then a pro-rata share of what remains — the investor 'double dips.' On a $100 million sale where an investor put in $50 million for 25 percent participating, they take $50 million plus 25 percent of the remaining $50 million: $62.5 million total. Participating terms became rare after the 2021-2022 repricing, but they return in down markets and in hard-money sectors like deep tech, and founders signing quickly should model them before the signing, not after.

Why do multiples above 1x exist?

1x is the market standard; 1.5x to 3x multiples appear when investors perceive elevated risk — bridge rounds into troubled companies, venture debt with equity kickers, or sectors where exits are typically below the raised capital. A 2x preference on $100 million means the first $200 million of any sale belongs to that investor. In 2023-2024's repriced market, ratchets — agreements that retroactively reset an investor's purchase price if a later round values the company lower — also reappeared, functioning as a preference that moves. Every one of these terms is legal, common, and invisible in the valuation announcement.

What is the stacked-preference problem?

Preferences stack by seniority. The default is that the newest round is paid first (last-in-first-out in some stacks, first-in-first-out in others — it is negotiated), and a company with $300 million raised across rounds carrying preferences can carry $300 million-plus of stacked claims on the first dollars of any exit. The arithmetic consequence: below roughly the total preference stack, common stock is worth close to nothing, and employee options priced at a high 409A strike are worthless paper. This is the mechanism behind the quietest wealth destruction in startups — employees at companies that sold for 'hundreds of millions' who received nothing, because the sale price sat under the stack.

How should a founder or employee actually model this?

Founders: before signing any round, model the exit grid — at $50 million, $100 million, $500 million, $1 billion, who gets what under the actual terms. Ask for the preference stack in writing at every round, and negotiate caps on participation if participating terms are unavoidable. Employees: the right question in offer diligence is not 'what is the strike price' but 'what is the total preference stack above common, and does the company's realistic exit range clear it.' Companies that answer transparently are telling you something; companies that do not are telling you something too.

What is the current market state of the term?

As of the 2024-2025 market, 1x non-participating remains the overwhelming standard for priced rounds — the founder-favorable equilibrium that settled after the 2001 and 2008 corrections taught everyone the same lesson. Deviations cluster where risk does: bridges, debt-adjacent instruments, and the AI sector's mega-rounds, where structural protections for billion-dollar checks have included liquidation multiples and IPO-conversion terms worth reading twice. The pattern to remember is that preference terms loosen in exactly the markets where founders feel they have the least leverage to refuse — which is when the terms matter most.

The preference stack is the cap table's floor plan. Valuations describe the penthouse; preferences describe who can reach the exit first when the building is on fire.

Who led the round and who followed?

SoftBank led, with its Vision Fund 2 deploying alongside the parent balance sheet, in a syndicate reported to include Microsoft, Coatue, Altimeter, and Thrive Capital. Microsoft's participation mattered beyond the money: the round coincided with a renegotiated Microsoft partnership under which OpenAI gained the right to source compute from third-party clouds — a change that made the $300 billion enterprise architecture possible at all, because a company of that scale cannot run on one supplier's capacity.

What did the tranches actually condition on?

Per Reuters reporting at the time, the initial roughly $10 billion arrived at close, while the later $30 billion depended on OpenAI completing its corporate restructuring — moving commercial operations into a public benefit corporation controlled, through a complicated equity chain, by the original nonprofit. That condition was met when the restructuring closed in October 2025 at a reported $500 billion valuation, with SoftBank entitled to invest an additional $10 billion at that price if it chose. Tranching is common in megadeals precisely because it converts a bet on a company into a bet on named milestones; founders negotiating smaller rounds should recognize the same device scaled down.

Where did the money go?

The stated purpose was compute. The round was announced alongside the acceleration of Stargate, the datacenter venture with Oracle and SoftBank announced in January 2025 with reported commitments of $500 billion over four years, and a reported cloud commitment to Oracle of roughly $300 billion over five years for OpenAI's capacity. In other words: the largest equity round ever raised was substantially a pass-through into infrastructure contracts, some of them with the lead investor's own ecosystem. That circularity — equity in, compute contracts out — is the structural risk nobody at the closing dinner mentions.

How does the valuation compare with the record?

At $300 billion, OpenAI became the most valuable private company in the world, ahead of SpaceX, and the round roughly doubled its $157 billion October 2024 valuation in five months. For context on the round-size league table: OpenAI's own $6.6 billion October 2024 round and xAI's $6 billion raises were the prior benchmarks — the SoftBank-led deal was six times larger than anything before it. Whether $300 billion prices in perfection is unanswerable from outside; what the record shows is that secondary markets and later reporting marked the company up further, to roughly $500 billion by late 2025.

Why does this deal matter for everyone else?

Three spillovers. First, the round reset the ceiling of what a private company can raise, making $10 billion raises look routine — Anthropic's reported $13 billion January 2026 discussions follow directly in this lane. Second, it anchored an AI fund-raising complex in which the same institutional names appear across labs, clouds, and datacenter ventures, concentrating both capital and dependency. Third, it demonstrated that governance uncertainty — the nonprofit-to-PBC question, contested in court by Elon Musk — no longer blocks nine-figure checks; it just gets tranched around.

The lesson for founders is not the size. It is that every term in a megadeal — tranches, conditions, warrants, compute tie-ins — allocates risk between investor and company, and the headline number is the least informative line on the term sheet.