Venture Debt for Startups: Terms, Warrants, MAC Clauses, and the SVB Lesson

Venture debt is a loan to a company that investors have already funded, sized against the last equity round and repaid from the extra runway it buys. It costs less than equity in dollars if the company does well, but it must be repaid on a schedule, it comes with legal terms that can bite in a bad quarter, and it depends on a small group of lenders. This guide covers the terms in a typical term sheet, the Silicon Valley Bank failure and what it showed about relying on one lender, current market data, and a worked example of the all-in cost including warrants.

It is a companion to two other guides. Revenue-based financing is the alternative for companies without institutional investors, and convertible debt is the alternative when you are still deciding on a valuation. The equity side of this comparison is in our venture capital guide, and lending to larger profitable companies is in private debt.

What lenders look at and how much they lend

Lenders underwrite a venture loan mainly on the strength of the equity behind it. Kruze Consulting's sample term sheet makes funding subject to "satisfactory conversations with Borrower's primary investors," and Mercury's term sheet guide says loan size is usually 20% to 50% of the previous venture round. That is why debt is normally raised soon after an equity round closes, when cash is high and the investors are engaged, and why it is hard to raise when a round has stalled.

The rest of the term sheet is a handful of numbers:

• Interest: mostly floating, often WSJ Prime plus a margin, with a floor. Kruze's example is Prime plus 1.00% with a 5.75% floor. Prime was 7.00% as of September 2026, per FRED.
• Interest-only period: Mercury says 12 to 18 months is typical, and it notes that the length of the period tends to influence the interest rate and the warrant pricing. After it, principal amortizes; our illustration below uses 36 months.
• Fees: an upfront fee, and often a final payment due at maturity as a percentage of the loan. Kruze's example has a 6.00% final payment, and some lenders add a prepayment fee.
• Warrants: options to buy company stock at a fixed price, described below.
• Covenants and conditions: financial tests, restrictions on other debt or asset sales, and the material adverse change clause.

The pricing sits on top of the market's cost of lending to technology companies. Hercules Capital, a large public venture lender, reported in its second-quarter 2026 results a core yield of 12.0%, a non-GAAP measure that leaves out one-time items such as early-repayment income, within its expected range of 12.0% to 12.5%. That figure comes from its own portfolio, and any single company's quote will differ.

Warrants, covenants, and the MAC clause

Warrants

A warrant is the lender's option to buy stock at a fixed price, normally the price of the last funding round or the latest valuation. Mercury says they most commonly last 10 to 12 years, longer than the loan. Kruze's example grants a warrant for 0.25% of the fully diluted company, half vesting at closing and half at the first draw, exercisable for 10 years, with cashless exercise allowed, and Kruze says 20 to 35 basis points is standard, with fund lenders usually asking for more.

Sources describe coverage in two ways. Some state it as a percentage of ownership, as above. Others state it as the strike value of the warrant divided by the loan. Kruze's suggestion is to work out the actual stock you are giving up and divide it by the loan amount. Ask the lender to state both so you can compare offers.

Covenants

Mercury separates light and heavy covenant structures. Non-bank lenders and venture debt funds usually require little more than a basic liquidity test, while banks tend to test debt-to-EBITDA, interest coverage, or minimum cash. Lighter covenants may come with a higher interest rate. For an early-stage company whose results are hard to predict, a heavy structure is dangerous: a single bad quarter can breach a ratio while the business is otherwise healthy. Read the negative covenants, which restrict additional debt, asset sales, and a change of control, as closely as the financial tests.

Many bank term sheets also require the borrower to run its cash management through the lender. Kruze notes that this protects the bank because it can offset deposits against the debt after a default. The clause matters more after 2023, and the next section explains why.

MAC clauses

A material adverse change clause gives the lender the right to declare a default, and in some documents to refuse further draws, if something adversely affects the borrower's financial position. Kruze's sample term sheet says the lender does not have to advance more money if a MAC occurs, and it warns that the bank can call a MAC if the burn rate is out of control. Because a MAC is subjective, it can be triggered by events like a failed fundraise or a large customer loss, which is when a company most needs the money. In negotiation, aim to delete it, or limit it to objective tests such as a minimum cash balance or a missed payment, and make sure any funding condition is not tied to a subjective view of your prospects. Also ask what happens to undrawn commitments if your lead investor changes its plans, since lenders treat investor support as part of the credit.

What the Silicon Valley Bank failure showed

Silicon Valley Bank (SVB) was a major bank and lender to venture-backed technology companies. The Federal Reserve's April 2023 review says the bank tripled in size between 2019 and 2021 on deposit inflows from the VC and technology boom. Those deposits were largely uninsured, and they came from a concentrated network of venture investors and technology companies. On March 9, 2023 the bank lost over $40 billion of deposits in a day, and management expected to lose over $100 billion more the next day; the outflow represented roughly 85% of the deposit base. California regulators closed the bank on March 10. A later outside review announced in September 2026 found, per CNBC's report, that the deposits were 94% uninsured and concentrated in venture-backed technology companies.

For borrowers, the immediate result was contained. On March 26, 2023 the FDIC announced that First-Citizens Bank & Trust would assume all deposits and loans of Silicon Valley Bridge Bank, and its branches opened under the new name the next day. The longer-term result was on the supply side. PitchBook's Q2 2026 venture debt analysis says the post-2022 retrenchment of traditional venture lending was accelerated by the collapse of SVB, a major lender in the sector.

Three practical lessons follow for a borrower:

• Cash concentration is a risk in itself. FDIC insurance covers $250,000 per depositor, per insured bank in each ownership category, so a company holding $18 million at one bank has about 98.6% of it uninsured. If your loan requires a primary operating account at the lender, negotiate how much cash it has to hold and for how long, and keep the rest at a second institution.
• A lender's failure changes who you owe. In SVB's case the loans moved to an acquirer along with the rest of the bank, and borrowers' obligations continued. Read the assignment and change-of-control provisions in your own documents.
• A single-lender relationship is a dependency. Ask a second lender for a term sheet even if you do not plan to use it.

The market now

The PitchBook-NVCA Venture Monitor for Q2 2026 gives the numbers. Venture debt reached $64.7 billion in the first half of 2026, boosted by a few outsized loans, including a $20 billion debt refinancing for SpaceX, but only 280 loans were recorded through the second quarter.

The averages are our arithmetic on PitchBook's totals. Between 2021 and 2025 the loan count fell about 32% while dollars rose about 69%. The Venture Monitor describes a market split between a small set of enormous infrastructure-driven facilities and a long tail that has been largely cut off, and it says private credit's expansion has sustained dollar volume while the availability of venture debt to the typical growth-stage company remains constrained and focused on AI. Excluding the SpaceX refinancing, the first-half average is still about $160 million, so the headline totals say little about what a $5 million loan costs or whether a lender will write one.

All-in cost, including warrants

Illustration: a company raised a $20 million Series A at an $80 million post-money valuation and has 100 million fully diluted shares, so the last round price and the warrant strike are $0.80. It borrows $6 million, which is 30% of the round. The terms are ours and sit within the ranges above: 48 months, with 12 months interest-only and then 36 months of equal principal payments; Prime plus 1.50%, which is 8.50% with Prime held at 7.00%; a 0.5% upfront fee ($30,000); and a 5% final payment ($300,000).

• Interest during the interest-only year is $42,500 a month. The first amortizing payment is $209,167 and the last is $167,847. Total interest over the loan is $1,296,250, so non-principal costs are $1,626,250.
• The all-in cost before warrants is 10.7% a year (10.2% nominal), from the internal rate of return on the net proceeds and all payments.
• With cash of $18 million and a burn of $1.5 million a month, the company has 12.0 months of runway. With the loan it has 15.2 months after paying debt service, or 3.2 months added, not the 4.0 months that $6 million would suggest. Debt service in months 13 through 24 is $2.43 million.

Warrants add to this only when the company's value rises above the strike, so the added cost depends on the eventual sale price:

These are annual costs assuming the warrant is exercised at the end of month 48 and the company is worth the stated multiple of the Series A price. A 0.25% warrant is 3.3% of the loan by strike value, and a 1.00% warrant is 13.3%, which is one reason to compare coverage on both bases.

Compare the same $6 million from equity. At an $80 million pre-money valuation it costs 6.98% of the company. If the company later sells for $120 million, $240 million, or $400 million, that stake is worth $8.4 million, $16.7 million, or $27.9 million. The loan with a 0.25% warrant costs $1.73 million, $2.03 million, and $2.43 million in the same three cases, and with a 1.00% warrant $2.03 million, $3.23 million, and $4.83 million. Equity costs less than the loan only if the company sells for under about $23 million, where the roughly $1.63 million of interest and fees exceeds 6.98% of the price.

That comparison flatters debt in one way. Equity does not have to be repaid when things go badly, and a $209,167 monthly payment starts in month 13 whether or not the next round closed. If the company cannot raise in the meantime, debt service, covenants, and a MAC clause combine. Size the loan so that you could meet it in your downside plan.

When venture debt fits and when it does not

It fits a company with institutional investors who intend to keep backing it, predictable enough revenue or milestones to bridge to the next round, and a specific use such as extending runway to hit a target that improves the next valuation. It fits poorly when the next round is uncertain, when burn is high relative to cash, or when the lender's covenants would be tripped by a normal bad quarter. Companies without institutional backers should look at revenue-based financing or bank options first.

This article is educational and does not replace advice from a lawyer, accountant, or your board. Loan documents differ on every point discussed here, and the illustration uses assumed terms, not a quote.