What is giving up early equity really worth?

Senua AI has accepted no startup funding of any kind. At the time of writing, it remains completely bootstrapped.

What has changed is interest. We have been in discussions with several interested parties. One example of that interest is an offer of A$250,000 for a flat 10% of the company, and I use it throughout this post.

No accelerator or incubator where equity is given up has accepted us at the time of writing, including the local Australian based Startmate or the San Francisco based Y Combinator. I look at their terms here because they are the standard yardsticks every Australian founder compares against, and they are published openly. Startmate's are A$120,000 for about 8%. Y Combinator's are US$500,000 for 7% and a further SAFE on top. Those are the terms any company in their batches receives, not offers to us.

So we are deciding the next step. Before I sign anything, I want to understand what I would actually be giving up, and how the people across the table arrive at their numbers. This post is that working, done in public. It is a little critical in places, because the deeper I looked into how early stage companies are valued, the less the numbers looked like calculations and the more they looked like conventions.

The offer and the yardsticks, in plain numbers

Every offer of money for a share of a company implies a price for the whole company. Divide the money by the share and you get the value of the company just after the money arrives, called the post money valuation. Take the money back out and you get the pre money valuation: what the company is said to be worth today. The most useful single figure for comparing offers is the price of 1% of the company.

Australian dollar figures use about 69 US cents to the Australian dollar, the rate in early October 2026 [3].

Three things stand out straight away.

• Three of the five rows price the company at almost exactly the same figure: about A$25,000 for each 1%, or roughly A$2.5 million for the whole company. That is not a coincidence. It is the going rate for a very early company, and it barely depends on the company.
• Startmate's standard terms are the most expensive in equity: 8% for A$120,000 at a fixed A$1.5 million cap. But Startmate's own terms say that a founder who has already closed a round of at least A$250,000 from venture or angel investors (not friends and family) gets that round's valuation instead [1]. For a company that took an offer like ours first and was then accepted, the same A$120,000 from Startmate would cost 4.8% rather than 8%. The order in which you sign changes the price.
• Y Combinator is, by a distance, the best price per dollar, because three quarters of its money comes on terms set later, by the market. More on that below.

How investors actually decide

I expected to find a rigorous method behind these numbers. The best evidence says there mostly is not one.

Paul Gompers, Will Gornall, Steven Kaplan and Ilya Strebulaev surveyed 885 venture capitalists at 681 firms about how they make decisions, and published the results in the Journal of Financial Economics in 2020 [4]. Their findings are summarised by the authors [5]:

• Few venture capitalists use discounted cash flow or net present value, the methods taught in every finance course for valuing an asset. The most common yardstick is simply how many times their money they expect to get back, followed by the internal rate of return.
• 9% reported using no quantitative measure at all to evaluate a deal.
• 20% said they do not forecast cash flows when they invest.
• The management team was named as the most important factor by 47% of firms. The business itself (its product, market and way of making money) came first for only 37%.

That is not a criticism of investors. At this stage there are no cash flows to forecast, so a forecast would be fiction with decimal places. But it does mean that an early valuation is a negotiated price, shaped by the investor's target return and the going rate, and then dressed up as a calculation afterwards.

I asked an AI assistant to value my company

To see what the dressing looks like, I asked one of the big AI chat assistants to value Senua AI using the standard startup methods. It came back, confidently, with four methods, four answers, numbered citations, and a "consensus" of A$12 million to A$20 million before the money. Here is what it produced, and what is wrong with each part.

The venture capital method deserves a closer look, because it is the one investors actually think in. Keep the assistant's own guess of an A$180 million exit in five years. An investor who wants ten times their money values the company today at A$18 million. One who wants thirty times values it at A$6 million. Allow for half the company being sold in later rounds before that exit, and those become A$9 million and A$3 million. One exit guess, two reasonable return targets, and an answer anywhere from A$3 million to A$18 million.

Its numbered citations had no sources attached to them. It also described our technology as "a lightweight machine learning engine". That is precisely what Senua AI is not:

We do not use models. No LLMs. No SLMs. Of any kind.

The real lesson is not that one assistant got its sums wrong. Each method gives an answer only once you supply the inputs, and the inputs are the valuation. Averaging four guesses gives a fifth guess, not a consensus. Meanwhile the example offer and the accelerators' published terms price a company like ours at about A$2.5 million at most: a fraction of every answer the assistant gave.

What Senua AI is worth by the standard methods, done properly

So I ran the same methods myself, by their authors' own rules, with every input written down so anyone can disagree with it. Where an input is my judgement rather than a published figure, the table says so.

By the three methods investors actually use, Senua AI is worth between about A$2.2 million and A$9.1 million before the money. The scorecard, the one most tied to what Australian investors are paying right now, says about A$6.3 million on the plainer baseline. The example offer values the company at A$2.25 million before the money, and the accelerators' published terms imply A$1.4 million to A$2.4 million: at or below the bottom of that range.

I have left cost to duplicate out of that range on purpose. It counts effort, and investors do not pay for effort, only for what the effort makes likely. A formula from 1981 that converts lines of code into years of work also knows nothing about modern tools. It is a reminder of what was built, not a price anyone will pay.

Notice what the range really says. Change two of my judgements and the answer moves by millions of dollars. That is not a flaw in my working. It is the nature of valuing a company before it has revenue.

The headline number is not the price

Even a carefully agreed valuation is less solid than it looks. Will Gornall and Ilya Strebulaev studied 135 American unicorns, private companies reported to be worth more than US$1 billion, in the Journal of Financial Economics [9].

• Their reported valuations averaged 48% above fair value.
• 65 of the 135 were not worth a billion dollars at all, once corrected.
• The reason is simple. A headline valuation takes the price of the newest shares and applies it to every share in the company. But the newest shares usually carry protections the older ones do not: the right to be paid back first, guaranteed returns at listing, vetoes. The founders' ordinary shares carry none of these, and were the most overvalued of all.

So two offers at the same valuation can be very different deals. These are the terms worth reading before the number:

• What instrument it is. Priced shares now, or a SAFE (a simple agreement for future equity) that converts into shares at the next priced round.
• Who is paid back first, and how much, if the company is sold.
• Board seats and vetoes.
• The option pool. Y Combinator's 7% is measured on a company that already includes any existing pool of shares for future staff, so the founders alone carry that pool's cost. A pool added at the next round dilutes everyone [2].
• Rights in later rounds. Pro rata rights let the investor keep their percentage, and "most favoured nation" terms give them the best terms anyone else later gets.

Accelerators do not value your company at all

Here is the part I find most clarifying. Startmate and Y Combinator do not value the companies they invest in. They publish one price, and every company in the batch pays it.

Startmate

Startmate invests A$120,000 on a A$1.5 million post money cap for first time raisers, as a SAFE or in the founder's priced round, whichever the founder chooses [1]. That is 8%. A company that has already closed a round of at least A$250,000 from investors gets that round's valuation instead.

Y Combinator

Y Combinator invests US$500,000 in two parts [2]:

• US$125,000 for a fixed 7%, on a post money SAFE.
• US$375,000 on an uncapped SAFE with a "most favoured nation" term. It has no fixed percentage. It converts on the best terms given to any other SAFE investor between the start of the batch and the next priced round, or at that round's price.

So the "further 4 to 5%" is not a fixed number. It is whatever the market says the company is worth next:

If the batch goes well and the next round is priced high, three quarters of the money ends up costing very little. That is why it is the best price of the three. The catch for an Australian company is structural. Y Combinator invests only in companies incorporated in the United States, Canada, Singapore or the Cayman Islands. Anyone else must create a new parent company in one of those places, with the Australian company beneath it, which Y Combinator itself calls "a significant effort" [10]. That move has legal and tax consequences in Australia that need proper advice before anyone signs.

What these programmes really sell is not capital. It is the network, the signal to later investors, the discipline of the batch and, for Y Combinator, Demo Day. Those can be worth far more than the cheque. They can also be worth less, for a company that already has what they offer. The equity is the fee, and the question is whether the programme is worth the fee.

What the network is worth depends on who you are

Accelerators and investors sell a network as much as money, and the research says the network matters. It also says it matters most to the founders who have the least of one.

• Networks predict outcomes. Yael Hochberg, Alexander Ljungqvist and Yang Lu studied American venture firms from 1980 to 2003 in the Journal of Finance [23]. Companies backed by better connected venture firms were significantly more likely to survive to their next round and to a sale or a listing.
• The cash and the desk are not what helps. Juanita Gonzalez Uribe and Michael Leatherbee studied Start Up Chile, a government accelerator, in the Review of Financial Studies [24]. They found no evidence that its basic services, funding and a shared office, improved performance on their own. Entrepreneurship schooling bundled with them did.
• Some accelerators help, not all. Benjamin Hallen, Susan Cohen and Christopher Bingham compared ventures accepted, and almost accepted, by top accelerators in Organization Science [25]. Some, but not all, substantially sped up the ventures' development. The main driver was what the authors call broad, intensive and paced consultation. Part of the apparent advantage is simply which companies get in.
• Experience is a network already. Pierre Azoulay, Benjamin Jones, J. Daniel Kim and Javier Miranda used United States Census data in American Economic Review: Insights [26]. The mean age at founding of the fastest growing 1 in 1,000 new companies was 45. Prior experience in the specific industry "predicts much greater rates of entrepreneurial success".

Read together, the same programme is worth very different amounts to different founders:

• A young or first time founder gains the most: the learning, the introductions, the signal to later investors. For them an accelerator's percentage is often cheap tuition.
• A founder with years in business already has much of what a batch teaches, and a network in their own industry, which the research says counts for more. What a programme adds is narrower: access to investors, a new market, peers at the same stage. The equity should be priced against that narrower value.
• A deep tech founder needs a network that reaches technical buyers, research partners and sector specialists. A network built around consumer apps and software sold online is worth less to them, however large it is.
• An older founder weighs the risk differently. Hall and Woodward found that savings make a startup's lottery of outcomes easier to carry [11]. Every year of building also gives up a senior salary.

For Senua AI the question is not how big a network is, but whether it reaches the people who buy physical AI and cyber security.

What early equity costs over time

Early percentages compound. Suppose a company takes an offer like ours (10%), is then accepted by Startmate at the matched valuation (4.8%) and then by Y Combinator (7% plus 2.5% at a US$15 million next round), and then sells a fifth of the company at seed and another fifth at Series A. Those numbers are illustrative, not a forecast. The founders would then own about half the company before any pool of shares for staff, and about 44% after a pool of 10%. Every early point is a point of every later outcome.

Two pieces of research put that in perspective.

• Most founders get nothing. Robert Hall of Stanford and Susan Woodward studied the payoffs to entrepreneurs in venture backed companies over twenty years, in the American Economic Review [11]. The reward to the founders was "zero in almost three quarters of the outcomes". The average payoff of US$4.4 million comes almost entirely from a few great successes. By their estimate, a founder with a moderate aversion to risk (a coefficient of two, in their terms) and less than US$0.7 million in assets would have been better off in a salaried job.
• Control has a price too. Noam Wasserman studied 6,130 American startups in the Strategic Management Journal [12]. Each extra level of control a founder keeps, whether the board or the chief executive's chair, lowered the company's pre money valuation by about 17% to 22%. This is the founder's dilemma Wasserman is known for: be rich, or be king. You can keep the company, or grow its value with other people's money and power, and it is very hard to do both.

How founders lose control, and how to guard against it

Most investors are not predators, and most of the time founder and investor want the same outcome. But the contracts give investors powers a founder only notices when they are used. The cases where they were used badly are on the public record.

• Founders are replaced, often. Thomas Hellmann and Manju Puri found that venture backed companies in Silicon Valley were more likely, and quicker, to replace the founder with an outside chief executive, in changes that looked adversarial as well as agreed (Journal of Finance, 2002) [27]. In Wasserman's study of 212 American startups, half of the founders were no longer chief executive by the company's third year. Fewer than a quarter led their company to a public listing [28].
• Preferences can leave founders with nothing. In In re Trados (Delaware Court of Chancery, 2013), a software company was sold for US$60 million [29].
  • The preferred shareholders, whose representatives dominated the board, took US$52.2 million, less than their US$57.9 million preference.
  • Management took US$7.8 million under a bonus plan.
  • The ordinary shareholders received nothing.
  The court found the directors' process unfair, but awarded nothing, because the ordinary shares had no value before the sale.
• Consent rights can be used as a weapon. In Basho Technologies (Delaware Court of Chancery, 2018), a minority investor used ordinary consent rights over new financing to block every other source of money [30]. It sabotaged alternative offers and froze management out of the negotiations. It held back money it had promised under a loan, and then gave the board less than a day to accept its own terms, which gave it control of the board. The court called the process "decidedly unfair" and ordered the investor and its manager to pay US$20.3 million. Basho was liquidated in 2017, and its technology, the Riak database, was sold to the betting company bet365 [31].
• The leaver clause decides what an ousted founder keeps. Australian founder agreements commonly say that a "bad leaver" forfeits unvested shares, and may have to sell vested ones back at a discount, at cost or at nominal value. The board often decides who is a bad leaver [32]. Put that beside a board the investors control, and a founder can be removed and lose their equity in the same step.
• Once the technology belongs to the company, it belongs to whoever controls the company. Investors rightly require all the intellectual property to sit in the company. Basho shows the other side of that: when control moved, the technology went with it.

None of this calls for paranoia. It calls for reading the terms and choosing the people. Before I sign anything, I want:

1. Board control with the founders at this stage, and the board's make up for later rounds written down.
2. A liquidation preference of no more than the money invested, one times and not participating.
3. Consent rights limited to genuine protections, not a veto over every future financing.
4. A narrow bad leaver definition: fraud or serious misconduct, never simply being asked to leave.
5. A deliberate decision on where the intellectual property sits, and on what terms, before the money arrives.
6. References from founders the investor has backed, including those whose companies failed or who were replaced. How an investor behaves when things go wrong is the reference that counts.

The question that comes before the price: how much money do you actually need?

Every founder's real decision is not which valuation to accept. It is how much outside money the business genuinely needs, and therefore how much of the company has to be sold at all.

For many AI companies the answer is "a great deal", because their AI means renting graphics processors and paying for tokens. Every dollar of that is equity sold to pay for hardware someone else owns. Senua AI was built the other way:

No GPU. No data centre. No AI token costs, ever. Senua AI runs on ordinary, low end CPUs.

That changes the arithmetic. Our running costs are small, so the money we raise goes on people and customers, not compute. It also means we can choose when to raise, and from whom. There are also ways to fund research in Australia that cost no equity at all:

• The R&D Tax Incentive. For a company with turnover under A$20 million, it refunds the company tax rate plus 18.5 percentage points of eligible research spending: 43.5% for a company on the 25% rate, paid in cash even when the company makes a loss [13]. The government has announced changes from 1 July 2028; until then the current rules apply [17].
• The early stage innovation company rules. These give investors in a qualifying company a tax offset of 20% of what they invest, capped at A$200,000 a year for sophisticated investors, and no capital gains tax on shares held between one and ten years [14]. That makes a round more attractive to investors without the company giving up more of itself.

The exit that fits deep tech: the acquihire, and the licence and hire

Most writing about startups assumes one ending: a giant round, then a listing or a huge sale. For small deep tech teams, a different ending is far more common, and I think it is the most likely pathway for Senua AI. A larger company buys the team, or licenses the technology and hires the people who built it.

How a classic acquihire works

In an acquihire, a larger company buys a startup mainly to hire its people, and the product is often closed. John Coyle and Gregg Polsky studied how these deals are put together, interviewing entrepreneurs, venture capitalists, buyers' representatives and lawyers, in the Duke Law Journal [33]. The buyer's money is split into two pools:

• a purchase price paid to the startup's shareholders;
• packages for the people it hires: signing bonuses, and stock that vests only if they stay.

They found no settled rule for how much goes into each pool, and predicted that a "money back for the investors" norm would develop. In plain terms, investors get their money back first, out of the purchase price, and what the founders receive comes mostly as employment, earned by staying.

Staying is the hard part. Weiyi Ng and Toby Stuart matched acquisitions of venture backed companies over two decades to more than 30 million résumés, in the Strategic Management Journal [34]. Acquired employees left at much higher rates than comparable direct hires. The gap was widest for senior and highly educated staff, and in executive, technical, business development and sales roles. That is why the money in these deals is tied to staying.

The newer form: licence and hire

Since 2024 the largest technology companies have used a variant, sometimes called a reverse acquihire. They take a licence to the startup's technology, usually nonexclusive, and hire its founders and key staff. The startup itself carries on, and the licence fee pays its investors. The deep tech cases are instructive:

Two older deals are closer to home:

• Where 2 Technologies (Australia). A small Sydney company founded by Lars and Jens Rasmussen. Google bought it in October 2004, and the team launched Google Maps in February 2005 [43].
• Xnor.ai. A spinout of the Allen Institute for AI that made AI run on small, low power devices instead of in the cloud. Apple bought it in January 2020, for a reported US$200 million [42].

Two warnings come with this form:

• Regulators now look at these deals. The UK's Competition and Markets Authority cleared Microsoft and Inflection, but treated it as a merger [35]. The US Federal Trade Commission made inquiries about Amazon and Adept [37].
• Not everyone is paid. At Windsurf, staff who had joined in the previous year reportedly received nothing from Google's deal, until Cognition's purchase gave every employee a share [40].

Why this pathway fits Senua AI, and what it means for funding now

• The value is in a small team and its technology. Senua AI is its own language, its own engine and its own products, built by a small team. A team and technology like that is exactly what these deals buy.
• Raising little keeps the outcome ours. In an acquihire, investors' money comes back first. A company that has raised nothing has no one to pay back first. Every dollar raised before such an exit is a dollar repaid ahead of the founders, often with a preference on top.
• Venture money is aimed at a different ending. A fund needs a few very large exits to return its money. An acquihire rarely does that, so a venture investor may push for a bigger, riskier path than the one that suits the company. Angels, revenue and the R&D Tax Incentive fit this pathway better.
• The technology must be clean to be licensed. A classic acquihire pays for people. The licence and hire deals above also paid for technology, and the licences were often nonexclusive, so the company could keep selling it. That needs the intellectual property clearly owned, documented and ready to license.
• A company that stands on its own commands a better price. Windsurf's licence came after a planned US$3 billion sale to OpenAI fell through [39]. Buyers pay more when there is competition, and when the company could carry on without them. So we will keep building Senua AI as a business that stands on its own, whatever its ending turns out to be.
• Everyone who builds it should share in it. Windsurf showed how the newest staff can be left out. Equity terms should be set now so that everyone participates.

Which path is best for a bootstrapped deep tech company?

Senua AI has been bootstrapped so far. The technology is built and running, and the question is the next step. Putting everything above together, this is my answer for a company like ours: deep tech, low running costs, a working product, and no pressing need to spend on hardware.

1. Turn the product into revenue first, and let the R&D Tax Incentive fund the research. Neither costs any equity. Revenue also changes how the company is valued: from judgements like the ones in my table to multiples of real sales. Every percent sold afterwards is worth more.
2. If cash is needed sooner, raise the smallest round that reaches the next proof point. Raise it from angels, at a price inside the honest range rather than below it, on a SAFE with a cap or on simple priced terms. The early stage innovation company rules make that round more attractive to the investor without costing the company more.
3. Choose investors for what the money cannot buy, and check how they treat founders. In deep tech that means customers, access to a sector, and patience. An offer at the going rate that brings none of those is just the going rate. Ask the founders an investor has backed, including the ones it replaced, before taking its money.
4. Take an accelerator for what only an accelerator gives, and take it in the right order.
   • Startmate after an angel round costs 4.8% rather than 8%.
   • Y Combinator is the best money per percent of the three, and the strongest signal, but only worth the move to a foreign parent company if the next market is the United States.
5. Build for the exit that fits. For a small deep tech team, that is most often an acquihire, or a licence and hire, by a larger company. Raising little keeps that outcome with the people who built the company, and clean intellectual property lets the technology be licensed as well as the team hired.
6. Leave the institutional venture round until there is revenue. A venture fund needs a very large exit within the life of the fund. Deep tech takes longer than most software to prove itself. Raising from a fund before the proof sells the most equity at the lowest price, to the investor with the least patience.

Applied to where we actually are:

• An offer like the example, A$250,000 for 10%, is the going rate, at the very bottom of what the standard methods support. It is worth taking only if the investor brings something beyond the cheque, on clean terms.
• An accelerator would mean applying and being accepted. Nothing of the kind has happened. If we ever took that path, Startmate's published terms are worth it for the network and cost less after an angel round. Y Combinator's are the best priced money, at the cost of moving the company's legal home.
• Staying bootstrapped remains a real option, because our running costs are small.

Where to get advice that is on the founder's side

Almost every guide to raising money is written by someone with a stake in how you raise it: an investor, an accelerator, a law firm or a platform. That does not make the guides wrong, but it is worth knowing whose interests they were written in. These are the resources I found most useful, grouped by who runs them.

No stake in your company: government and regulators

• business.gov.au [17]. The Australian Government's site for businesses: grants and programs, and an eligibility check for the R&D Tax Incentive.
• The Australian Taxation Office [13]. The authority on the R&D offset rates and on the early stage investor rules, which every adviser's summary is drawn from.
• ASIC's guide to crowd sourced funding for companies [18]. Another path altogether: raising from many small investors through a licensed platform, with the rules set out by the regulator.
• The US SEC's Office of the Advocate for Small Business Capital Formation [19]. An independent office of the American regulator, for anyone whose path leads to the United States, as Y Combinator's does.

Independent research

• The Kauffman Foundation [20]. A private, nonpartisan foundation that researches how entrepreneurs actually get capital, and who misses out.
• The research cited in this post. Gompers and colleagues, Gornall and Strebulaev, Hall and Woodward, and Wasserman are all peer reviewed, and all readable by a founder.

Free data from inside the industry

• Cut Through Venture's reports [21]. The best free numbers on Australian rounds and valuations, published quarterly and yearly. The yearly report is published with Folklore Ventures, a venture fund. The data is good; read the commentary knowing who wrote it.

Written by investors: useful, and read with that in mind

• Y Combinator's deal and SAFE documents [2] and Startmate's published terms [1]. Clear, public and the same for everyone, which is itself founder friendly.
• AirTree's Open Source VC [22]. Templates and guides from an Australian venture firm.

On your side because you pay them

• Your own lawyer and your own accountant. Not the investor's, whose lawyers act for the investor. Have your own read every term sheet before you sign it, and have them check the R&D Tax Incentive and the early stage investor rules for your company in particular.

What I am weighing

I have not decided yet, and I will write about the decision when I make it. These are the questions I am asking of every option, including staying bootstrapped:

1. The price of 1%, worked out the same way for every option, in one currency.
2. The terms behind the number: instrument, preferences, board seats, vetoes, option pool, later rights.
3. The order. Signing one offer can change the price of the next.
4. What comes with the money: customers, credibility, people, and whether we need those or already have them.
5. How the investor has treated founders before, from the founders themselves.
6. Whether the money keeps the likely exit open. In an acquihire or a licence and hire, who is paid first, and how much is left for the people who built the company?
7. What it costs in control, and whether that is a price worth paying at this stage.
8. How much we actually need, given that our technology does not need a data centre to run.

The honest conclusion of this deep dive is that early stage valuation is not a measurement. It is a convention, an agreed price on an agreed day with terms attached. Founders cannot change the convention. They can understand it, compare offers on the same terms, and decide how much of their company they really need to sell.

References

Peer reviewed research

4. P. A. Gompers, W. Gornall, S. N. Kaplan and I. A. Strebulaev, How do venture capitalists make decisions?, Journal of Financial Economics 135(1) (2020) 169 to 190.
5. The same authors' summary, How do venture capitalists make decisions?, Harvard Law School Forum on Corporate Governance, 20 August 2019: the percentages quoted above.

9. W. Gornall and I. A. Strebulaev, Squaring venture capital valuations with reality, Journal of Financial Economics 135(1) (2020) 120 to 143. Working paper: NBER Working Paper 23895. The figures above are the journal version's.

11. R. E. Hall and S. E. Woodward, The burden of the nondiversifiable risk of entrepreneurship, American Economic Review 100(3) (2010) 1163 to 1194. The quoted words are from its working paper, NBER Working Paper 14219 (2008).
12. N. Wasserman, The throne vs. the kingdom: founder control and value creation in startups, Strategic Management Journal 38 (2017) 255 to 277.

Programme terms, methods, market data and tax rules

1. Startmate, Startmate accelerator investment terms 101, 28 August 2025, read 18 October 2026.
2. Y Combinator, The Y Combinator deal, read 18 October 2026.
3. Exchange Rates, Australian dollar to US dollar, 1 October 2026: 0.6928.

6. D. Berkus, The Berkus method: valuing an early stage investment, Berkonomics, updated November 2016.
7. B. Payne, Scorecard valuation methodology, rev 2019: establishing the valuation of pre revenue start up companies, Angel Capital Association, 21 October 2019. The factor weights and the choice of baseline used above are from this revision.
8. Cut Through Venture, Australian startup funding, Q2 2026, 22 July 2026: weighted average investor reported valuations for Australian deals by stage.

10. Y Combinator, Frequently asked questions, "Incorporating", read 18 October 2026.

13. Australian Taxation Office, Rates of R&D tax incentive offset.
14. RSM Australia, Early stage innovation: an overview of the ESIC tax concession. The rules and current limits are the Australian Taxation Office's; check them there before relying on them.
15. B. W. Boehm, Software Engineering Economics, Prentice Hall (1981): basic COCOMO, organic mode, effort = 2.4 × (thousands of lines)^1.05 person months.
16. Senua AI, Cyber pricing: Shield, A$1,799 a month including GST.

Founder resources

17. Australian Government, Research and Development Tax Incentive, business.gov.au, read 18 October 2026: changes announced in the 2026 to 2027 Budget start on 1 July 2028.
18. Australian Securities and Investments Commission, RG 261 Crowd sourced funding: guide for companies, June 2020.
19. US Securities and Exchange Commission, Office of the Advocate for Small Business Capital Formation, annual report for 2024.
20. Ewing Marion Kauffman Foundation, Access to capital for entrepreneurs: removing barriers, 2023.
21. Cut Through Venture and Folklore Ventures, State of Australian startup funding 2025.
22. AirTree Ventures, Open Source VC, read 18 October 2026.

Peer reviewed research: networks, accelerators, founder age and founder replacement

23. Y. V. Hochberg, A. Ljungqvist and Y. Lu, Whom you know matters: venture capital networks and investment performance, Journal of Finance 62(1) (2007) 251 to 301.
24. J. Gonzalez Uribe and M. Leatherbee, The effects of business accelerators on venture performance: evidence from Start Up Chile, Review of Financial Studies 31(4) (2018) 1566 to 1603.
25. B. L. Hallen, S. L. Cohen and C. B. Bingham, Do accelerators work? If so, how?, Organization Science 31 (2020) 378 to 414.
26. P. Azoulay, B. F. Jones, J. D. Kim and J. Miranda, Age and high growth entrepreneurship, American Economic Review: Insights 2(1) (2020) 65 to 82. The quoted words are from the abstract of NBER Working Paper 24489.
27. T. Hellmann and M. Puri, Venture capital and the professionalization of start up firms: empirical evidence, Journal of Finance 57(1) (2002) 169 to 197.

Other research, court records and reporting

28. N. Wasserman, The founder's dilemma, Harvard Business Review 86(2) (February 2008), from a study of 212 American startups. A practitioner journal, not peer reviewed.
29. In re Trados Inc. Shareholder Litigation, 73 A.3d 17 (Del. Ch. 2013), opinion, 16 August 2013.
30. Basho Technologies Holdco B, LLC v. Georgetown Basho Investors, LLC, C.A. No. 11802 VCL (Del. Ch. 6 July 2018). Summary: Willkie Farr and Gallagher, Delaware Chancery Court holds minority shareholder liable as controlling shareholder for abusing contractual consent rights, 7 August 2018.
31. Computerworld, bet365 swoops in to save open source Riak database, 2017.
32. Sprintlaw, Bad leaver clauses: what Australian startups and SMEs need to know, read 18 October 2026.

Acquihires and licence and hire deals

33. J. F. Coyle and G. D. Polsky, Acqui hiring, Duke Law Journal 63 (2013) 281. Peer reviewed legal scholarship, from interviews with entrepreneurs, venture capitalists, buyers' representatives and lawyers.
34. W. Ng and T. E. Stuart, Acquired employees versus hired employees: retained or turned over?, Strategic Management Journal 43(5) (2022) 1025 to 1045.
35. TechCrunch, UK regulator greenlights Microsoft's Inflection acquihire, but also designates it a merger, 4 September 2024.
36. Verdict, UK regulators investigate Microsoft's Inflection AI acquihire, 2024: the reported US$650 million licence.
37. GeekWire, Amazon hires founders from well funded enterprise AI startup Adept, June 2024; Computerworld, FTC is looking into Amazon's deal with AI startup Adept, 2024.
38. Amazon, Amazon and Covariant, August 2024: the nonexclusive licence and the people who joined.
39. CNBC, via NBC, Cognition to buy AI startup Windsurf days after Google poached CEO in $2.4 billion licensing deal, July 2025.
40. Maginative, Cognition acquires Windsurf after Google's talent raid left it gutted, July 2025.
41. DatacenterDynamics, Nvidia hires staff from networking firm Enfabrica and signs tech license in $900m deal, September 2025, reporting CNBC and Reuters. Nvidia had not confirmed the terms.
42. TechCrunch, Apple buys edge based AI startup Xnor.ai for a reported $200M, 15 January 2020.
43. Keystone SDA via blue News, From your desk to your pocket: 20 years of Google Maps, 2025. The price Google paid for Where 2 Technologies was not disclosed.