Licence anatomy, clause by clause
The core idea#
The licence is the document that puts the institution's IP into your company. For a deeptech startup it is often the single most important contract it signs, because investors are buying into the rights it grants. A good licence gives the company exclusive rights broad enough for its business, economics that leave room for future funding rounds, and the freedom to be sold. A bad one can make the company impossible to finance or acquire.
This playbook goes through a university licence clause by clause, explains what each clause does and what founders and investors should aim for, gives published benchmarks, and shows how to model what a licence will cost over time.
Stage: minus-one to pre-seed · Lens: Deeptech · Reading time: about 35 minutes · Not legal advice: have a lawyer experienced in university licences review the final draft.
Who this is for#
- Founders who have received a term sheet, option or draft licence from a TTO.
- Founders preparing to negotiate (playbook 2.10).
- Business co-founders and early investors reviewing a spinout's licence.
What you leave with#
- An understanding of every major clause and why it matters.
- Benchmarks to compare your terms against.
- A simple model of what the licence costs at different stages.
- A red-amber-green checklist for reviewing a draft.
The three questions every clause answers#
- What rights do we get? Grant, exclusivity, field, territory, licensed IP, improvements, sublicensing.
- What do we pay? Equity, upfront fee, patent costs, royalties, milestones, minimums, sublicence share.
- Who controls what happens next? Diligence, assignment, change of control, termination, patent prosecution, enforcement, government rights.
How the economics fit together#
A university licence pays the institution in several ways at once. No single term tells you whether a deal is fair; you have to look at the whole economic package (Gunderson Dettmer).
| Payment | When it's paid | Who it hurts most | Founder priority |
|---|---|---|---|
| Equity | At signing, as shares |
Founders and later investors, through dilution |
Keep it near published norms; limit anti-dilution |
| Upfront fee | At signing | A cash-poor startup | Keep it small, or replace it with equity |
| Past patent costs | At signing or spread over time |
A cash-poor startup | Spread over 1–3 years or defer to a funding round |
| Future patent costs | Ongoing | The company, from month 30 especially |
Control which countries are filed |
| Annual maintenance or minimum fee |
Every year | Pre-revenue years | Small, rising slowly; creditable against royalties |
| Milestone payments | On defined events |
Depends on the event |
Tie to events that bring money |
| Royalties on sales | Once products sell |
Gross margin, especially in hardware |
Low single digits; clean net sales definition; stacking relief |
| Share of sublicence income |
When you sublicense |
Business models built on partnering |
Lower share for later sublicences |
Trade-offs to understand#
Equity versus royalty. For a startup with no revenue, universities often take equity in place of, or alongside, a lower upfront fee, so their return follows the company's success (CASRAI). A higher equity stake can justify a lower royalty, and the reverse. A
practitioner survey of about 100 licences found the two rates moved against each other (Orrick).
- Startup versus large company deals. Universities ask large companies for big upfront and milestone payments and accept lower royalties; startups tend to get lower cash terms, a required annual minimum and equity (Mondaq).
- Business model matters. A company that will sell products cares most about royalty rate and net sales definition. A company that will partner or license out cares most about the sublicence income share. A platform company cares about field of use and improvements.
Rights clauses#
These clauses decide what the company actually owns the right to do. Get them right first; money terms are easier to fix later.
Grant#
What it says: the rights granted, usually "to make, have made, use, import, offer to sell and sell" licensed products. A real example: Stanford's licence to Bolt Biotherapeutics grants exactly these rights within the licensed field and territory (Justia).
Aim for: all the verbs your business model needs, including "have made" (so contract manufacturers can produce for you) and the right to practise licensed processes.
Exclusivity#
What it says: whether the university will license the same rights to anyone else. "Exclusive" in the Stanford–Bolt licence means Stanford won't grant further licences under the licensed patents in the field and territory; it ran until the last licensed patent expires (Justia).
Aim for: exclusive rights in your core fields for the life of the patents. Universities nearly always keep the right to use the technology for research and teaching, and often to let other non-profits do so. That is normal.
Field of use#
What it says: which applications the licence covers. Biotech licences are often split by therapeutic area, while standard technology licences more often use broad fields (CASRAI). The Stanford–Bolt licence's field was "all fields" (Justia).
Aim for: a field wide enough for the beachhead and the next two markets (playbook 1.07). If the university wants to keep other fields for other licensees, ask for a right of first negotiation on them.
Watch for: fields defined by your current prototype ("membranes for solvent recovery in pharmaceutical manufacturing") rather than the market ("separation of organic solvents from aqueous streams").
Territory#
What it says: where you can operate.
Aim for: worldwide. A territory limited to one country rarely makes sense for a startup and worries investors.
Licensed IP#
What it says: exactly which patents, applications, software, know-how and materials are licensed. Usually listed in a schedule.
Aim for: every filing in the family, including continuations and foreign counterparts; knowhow and data needed to use it; and materials such as cell lines or samples, under a material transfer clause.
Improvements#
What it says: whether later inventions from the same lab are included.
Aim for: an automatic inclusion of improvements made by the named inventors for a set period (for example, 2–3 years), or at least a right of first negotiation. Without it, a better version of your technology could be licensed to a competitor.
Sublicensing#
What it says: whether you can grant rights to partners, and on what terms.
Aim for: the right to sublicense without the university's approval of each deal, with a share of sublicence income paid to the university (see money clauses). Note that agreements for a contractor to make products for you are usually not sublicences; the Stanford–Bolt licence says so explicitly (Justia).
Money clauses#
Royalties#
What it says: a percentage of net sales of licensed products, often tiered so the rate rises as annual sales cross thresholds (Gunderson Dettmer).
Benchmarks: industry and TTO sources describe a broad range of roughly 1–10% of net sales, commonly in the low-to-mid single digits (CASRAI). A Gunderson Dettmer analysis of university licences found a median royalty of 2.38% where anti-stacking applied and 1.75% where it didn't (Gunderson Dettmer). Where the licensed patent is only peripheral to the product, the royalty may be around 0.5% (Guide to Biotech Startups). The UK USIT guide gives a range of 0–5% depending on the technology's maturity (Cambridge Enterprise).
Aim for: low single digits, with a clear net sales definition.
Net sales definition#
The royalty is only as fair as the base it's charged on. Make sure net sales:
- Deduct returns, discounts, taxes, shipping and insurance.
- Cover only the licensed product, not a bigger system it sits inside. If your patented membrane is 10% of a module's value, argue for a royalty on the membrane's value, or a lower rate on the whole.
- Exclude services, software subscriptions or consumables not covered by the patents, unless they are.
- Exclude sales between the company and its affiliates or sublicensees, so the same sale isn't counted twice.
Royalty stacking#
What it says: if you must also license other patents to make the product, you can reduce the university's royalty by part of what you pay others, down to a floor (CASRAI).
Benchmarks: anti-stacking appeared in 65% of university licences in the Gunderson Dettmer analysis; typically the licensee may reduce up to 50% of its royalty, by 50% of the amounts paid to third parties for other necessary licences (Gunderson Dettmer). These provisions are common in biotech and rarer in standard technology deals (CASRAI). Aim for: include it even if you don't expect to need it. Patent landscapes change (Orrick).
Sublicence income share#
What it says: the share of payments you receive from sublicensees that goes to the university. 94% of university licences in the Gunderson Dettmer analysis required it (Gunderson Dettmer).
Benchmarks: published express licences range from 15% (Nebraska) (UNL) to 33% (Missouri) (University of Missouri) of non-royalty sublicence income. Stanford's licences to Bolt Biotherapeutics set the share between the high single digits and low teens, depending on when the sublicence is signed (SEC filing).
Aim for: a share that falls over time or as you add value (for example, lower after a Phase I trial or a working prototype).
Upfront fee, maintenance fees and minimums#
Benchmarks: upfront fees appeared in 75% of licences in the Gunderson Dettmer analysis, with a median of $17,500; the median annual maintenance fee was $25,000 (Gunderson Dettmer). UNC's early Carolina Express model set annual minimums of $5,000–15,000 for three to five years, rising to $10,000–30,000 after year six (WIPO/AUTM).
Aim for: low early fees, rising slowly; and make annual fees creditable against royalties, as in the Stanford–Bolt licence (SEC filing).
Milestone payments#
What it says: payments when defined events happen: first sale, regulatory filing or approval, a funding round. Common in biotech, rare in standard technology licences (CASRAI).
Aim for: milestones tied to events that bring the company money or value, so you can afford them when they fall due.
Equity and anti-dilution#
Benchmarks: equity appeared in 57% of the analysed university licences, with a median of 4.2%; 83% of those with equity included anti-dilution protection, with a median threshold of $10 million raised (Gunderson Dettmer). University anti-dilution isn't the price-based protection investors get; it means the university receives more shares to keep its percentage until the company has raised a set amount (Orrick).
Aim for: a stake at or below published norms (playbook 2.11), and anti-dilution that ends at a modest threshold.
Control clauses#
These decide whether the company stays in control of its own future. Investors read them closely.
Diligence obligations#
What it says: milestones the company must hit (raise money, build a prototype, file for approval, make a first sale) by set dates, or the university can narrow the licence or end it. Diligence is one of the terms universities care most about (Orrick).
Aim for: milestones you control and can meet with realistic funding; dates measured from the funding you will raise, not from signing; a process to agree extensions; and a cure period before any termination. Tie the first milestone to "commercially reasonable efforts" rather than fixed dates where possible.
Assignment and change of control#
What it says: whether the licence can be transferred, including when the company is acquired.
Aim for: the right to assign the licence to an acquirer of the whole business, or of the business line using it, without the university's consent (or with consent not unreasonably withheld). A licence that needs consent for an acquisition hands the university a veto over your exit.
Patent prosecution and enforcement#
What it says: who controls filings and who can sue infringers. Large licensees often fight over who prosecutes patents and who calls the shots in litigation (Mondaq).
Aim for: a say in prosecution strategy (which countries, which claims), since you pay for it; and the first right to enforce the patents in your field.
Termination and what happens after#
What it says: when either party can end the licence, and what happens to sublicences and improvements afterwards (CASRAI).
Aim for: termination by the university only for material breach after a cure period; the company's right to end the licence on notice; and sublicences that survive termination so your partners aren't left exposed.
Government rights and other reserved rights#
What it says: rights the university must keep because of how the research was funded. Under Bayh-Dole, the US government keeps a licence, and exclusive licensees must substantially manufacture in the US for US sales (Drexel). In India, DBT-funded IP licences carry an irrevocable, royalty-free right for the Government of India in health, safety or security needs (Conventus Law).
What to do: know they exist and plan for them; they are rarely negotiable.
Board observer, information and reporting rights#
Universities sometimes ask for a board observer seat. Investors often don't object if it has a sunset (Orrick). Reporting obligations (royalty reports, progress reports, audit rights) are normal; keep them proportionate.
Benchmarks from published startup terms#
Use these as evidence of what is normal. The right comparison depends on your field and the strength of the IP.
| Source | Equity | Royalty | Anti-dilution | Other |
|---|---|---|---|---|
| Gunderson Dettmer analysis (university licences, median) |
4.2% (in 57% of deals) |
2.38% with anti-stacking; 1.75% without |
Median threshold $10 million (in 83% of equity deals) |
Upfront fee median $17,500; maintenance fee median $25,000 |
| Carnegie Mellon standard startup terms |
6% exclusive, 5% non- exclusive (patented); 3% (software) |
2% (patented); none (software) |
$2 million | Open-source clearance option: 1% |
| USIT guide | 10–25% for IP- | 0–5% | — | Lower equity for |
| (UK) | heavy spinouts | software (playbook 2.11) |
||
| University of Missouri |
None | — | — | 33% of non- royalty |
| express | sublicence income |
| Source | Equity | Royalty | Anti-dilution | Other |
|---|---|---|---|---|
| IIT Kharagpur | 3% (own technology); 5% (other institute technology) |
— | — | Equity-based licensing |
| Guide to | Under 5% | Around 0.5% | Through a | Stacking relief |
| Biotech | if the patent is | reasonable | up to 50% | |
| Startups (typical biotech) |
peripheral | level of funding |
Where a cell is empty, the source doesn't state a figure.
Modelling a licence: a worked example#
Always model what the licence costs year by year before you sign. The example below is illustrative; replace the numbers with your draft terms and your own plan.
Draft terms: 5% equity with anti-dilution until $2 million is raised; $15,000 upfront fee; $40,000 of past patent costs repaid over three years; annual maintenance fee of $10,000 rising to $25,000, creditable against royalties; 2% royalty on net sales; 20% of sublicence income.
Company plan: pre-revenue in years 1–2; $1 million of sales in year 3, $5 million in year 4 and $15 million in year 5; a $2 million upfront payment from a sublicensee in a second field in year 5.
| Year | Sales | Fees and patent costs | Royalty (after crediting maintenance fee) |
Sublicence share |
Total paid to university |
|---|---|---|---|---|---|
| 1 | $0 | $15,000 upfront + $13,333 patent costs + $10,000 maintenance |
$0 | $0 | $38,333 |
| 2 | $0 | $13,333 patent costs + $10,000 maintenance |
$0 | $0 | $23,333 |
| 3 | $1M | $13,333 patent costs | $20,000 (covers the $15,000 maintenance fee) |
$0 | $33,333 |
| 4 | $5M | — | $100,000 | $0 | $100,000 |
| 5 | $15M | — | $300,000 | $400,000 | $700,000 |
National phase filing costs (playbook 2.04) come on top, usually in years 2–3.
What the model shows#
- Early cash costs are modest but real for a pre-seed company: about $60,000 in the first two years. Negotiating to spread or defer them matters more than the headline royalty at this stage.
- Royalties dominate once you sell. Doubling the royalty to 4% would add $300,000 in year 5 alone.
- The sublicence share can be the biggest single payment. In this example it exceeds a year's royalties. If partnering is central to your business, negotiate the share hard, and ask for it to step down as the company adds value.
- Equity is the long-term cost. At 5% with anti-dilution to $2 million, the university keeps 5% through the first round, then is diluted with everyone else. Model it alongside founder and investor stakes (playbook 5.19).
What investors will ask you to change#
Lead investors and their lawyers review the licence in diligence (playbook 14.11). These are the requests they make most often. Raising them with the TTO before an investor does saves time.
| Investor concern | Typical request |
|---|---|
| Field too narrow for the plan | Broaden the field, or add a right of first negotiation on adjacent fields |
| Acquisition needs university consent | Free assignment on change of control |
| Diligence dates the company can't meet | Dates tied to funding, plus an extension process and cure period |
| University anti-dilution beyond the seed round |
A lower threshold, or anti-dilution that ends at the first priced round |
| High sublicence share | A lower share, or one that steps down over time or after milestones |
| No stacking relief | An anti-stacking clause with a reasonable floor |
| Improvements not included | Automatic inclusion or a right of first negotiation for a fixed period |
| University can terminate easily | Termination only for material breach after a cure period; sublicences survive |
| Investor concern | Typical request |
|---|---|
| Unclear net sales definition | Narrow it to the licensed product; deduct |
| standard items |
Two ways to handle this well:
- Share the draft licence with your likely lead before signing. It's easier to change a draft than a signed contract.
- Use published standards as evidence. TTOs move more easily when you show terms other institutions have accepted (playbook 2.03).
Red-amber-green review checklist#
Mark each clause in the draft. Fix every red before signing; take ambers to your lead investor.
| Clause | Green | Amber | Red |
|---|---|---|---|
| Exclusivity | Exclusive in core fields for the patent life |
Exclusive for a limited period |
Non-exclusive in the core field |
| Field of use | Covers beachhead and next markets |
Covers beachhead only | Defined by the current prototype |
| Territory | Worldwide | Major markets | One country |
| Improvements | Included for a set period |
Right of first negotiation | Excluded |
| Sublicensing | Allowed without per-deal approval |
Allowed with consent not unreasonably withheld |
Not allowed |
| Royalty | Low single digits, clean net sales |
Mid single digits | High, or charged on whole systems |
| Stacking | Up to 50% offset with a floor |
Small offset | None |
| Sublicence share | At or below ~20%, stepping down |
20–33% | Above a third, flat |
| Upfront and patent costs |
Small; past costs spread |
Moderate | Large cash at signing |
| Clause | Green | Amber | Red |
|---|---|---|---|
| Equity | At or below published norms |
Slightly above | Well above norms |
| Anti-dilution | Ends at a modest threshold |
Ends at the Series A | Continues indefinitely |
| Diligence | Tied to funding; extensions; cure period |
Fixed dates with cure | Fixed dates, automatic termination |
| Change of control | Free assignment on acquisition |
Consent not unreasonably withheld |
University veto |
| Termination | Material breach plus cure only |
Broader grounds with cure |
At will, or without cure |
Mistakes and how to recover#
| Mistake | Why it hurts | How to recover |
|---|---|---|
| Negotiating one term at a time |
You win the royalty and lose on fees, sublicence share and diligence |
Negotiate the whole package against a model |
| Signing an option with terms "to be agreed" |
Less leverage later | Attach a term sheet to the option |
| Accepting diligence dates without funding |
Risk of losing the licence | Renegotiate now, linking dates to funding |
| Ignoring change of control | University can block or tax an acquisition |
Ask for free assignment before signing; fix it in the first amendment if missed |
| A field defined too narrowly | Next markets need a new licence |
Ask for a field amendment or first negotiation rights |
| No model of total cost | Surprises when sales or sublicences arrive |
Build the year-by-year model before signing |
| Not involving investors before signing |
Renegotiation after signing | Share drafts with your likely lead |
| Mistake | Why it hurts | How to recover |
|---|---|---|
| Signing without a lawyer | Common traps missed | Get a specialist review of |
| who knows university | the final draft | |
| licences |
Existing playbooks to read, and sources#
Read next#
- Gunderson Dettmer's university licence presentation (2024), with benchmark data on fees, equity, royalties and anti-stacking (Gunderson Dettmer).
- Orrick's guide to university technology licensing for startups (Orrick).
- US-BOLT life science term sheet template (NVCA).
- Real licences filed with the SEC. Companies that go public often file their university licences, partly redacted. Stanford's licence to Bolt Biotherapeutics is a readable example (Justia).
- CASRAI's licence agreement guides (CASRAI).
Related playbooks in this library#
2.01 Who owns your invention · 2.03 Working with the TTO · 2.08 Royalties, milestones and anti-stacking · 2.09 Sublicensing and change of control · 2.10 Negotiating with a university · 2.11 Spinout equity norms · 5.19 Cap table modelling · 12.03 Deeptech pricing · 14.11 Technical due diligence
Sources#
- Gunderson Dettmer: university licence presentation (2024)
- CASRAI: patent licensing
- CASRAI: MFN and anti-stacking clauses
- CASRAI: biotech vs standard tech licensing
- CASRAI: IP licensing agreements hub
- Orrick: navigating university technology licensing (2024)
- Orrick Tech Studio: licensing technologies from a university
- Mondaq: university licences to startups vs pharma
- Guide to Biotech Startups: licensing from universities
- Justia: Stanford–Bolt exclusive licence
- SEC: Bolt Biotherapeutics 10-Q (2022)
- SEC: Bolt Biotherapeutics S-1 exhibit
- Carnegie Mellon: startup licence terms
- University of Missouri: express licence guidelines
- University of Nebraska: NUexpress
- WIPO/AUTM: standard agreements and Carolina Express
- Cambridge Enterprise: the USIT Guide
- IIT Kharagpur entrepreneurship policy 2026
- Drexel: Bayh-Dole Act
- Conventus Law: DBT IP Guidelines 2023
- NVCA: US-BOLT term sheet