Just because you’re building with tech doesn’t mean you need VC money. People have made this point before, usually as a defence of bootstrapping, or as a dig at founders who raise a seed round to sell candles.
But neither version gets at the part that is more interesting: why does that statement keep becoming truer?
One way to answer that is to look at what technology used to signal. In 1999, two men raised $7 million to build a company called Computer.com. They had a website, a very good domain name and, apparently, a great deal of confidence. Six weeks before the Super Bowl, they spent roughly half of the money on advertising.
A Weekend America account from the period tells the story in the founders’ own words. The ads ran. The company launched. Ten months later, Computer.com was sold to Office Depot for an undisclosed amount.
One of the founders later remembered meeting dentists who were talking about starting hedge funds and suddenly thinking, essentially, okay, maybe everyone has decided they are a financial genius now.’
There is something almost too perfect about a company called Computer.com spending millions telling people to visit Computer.com. But the joke only works because the name meant something then. The Internet was new, and the technology was difficult. A company with a .com in its name looked like it had arrived early to somewhere everyone else was going.
That feeling was not unique to Computer.com. In a study of 95 publicly traded companies that announced Internet-related name changes in 1998 and 1999, Cooper, Dimitrov and Rau found cumulative abnormal returns of roughly 74% around the announcements over a ten-day window. The effect was not tightly related to how dependent the underlying businesses actually were on the Internet. The label was doing some of the work.
That is the interesting part. A signal can be useful without being the thing it is supposed to tell you about. A name containing “.com” did not make a company good. It just gave investors something new to believe about what the company might become. The paper is old, but the problem it points to is not.
For a while, building software worked in much the same way. Getting from an idea to a real product that people could actually use required engineers, infrastructure, money and patience.
It did not prove the business would work, but it told you that somebody had crossed a difficult hurdle. The cost of building the technology was part of what made the signal useful. Then the hurdle moved.
Cloud infrastructure became easier to access. Development could be bought from almost anywhere. No-code tools took work out of the engineering queue. AI can now write a surprising amount of software. None of this is bad. Quite the opposite. It is probably one of the best things to happen to the cost of building technology.
It just creates a strange side effect. When almost everyone can produce the signal, the signal tells you less. A website tells you there is a website. An app tells you someone built an app. A dashboard tells you that somebody connected a database to a front end. At some point we started treating a login screen as though it came with a moat.
And this is where the question at the beginning starts to answer itself. If the technology can be built more cheaply, then the mere fact that technology is involved is a weaker reason to assume the company needs venture capital. The question shifts to what remains expensive, difficult or scarce.
The problem with cheap signals
Michael Spence made a version of this point in his 1973 paper, “Job Market Signaling”. His setting was the labour market, not startups, but the basic idea is simple: a costly signal can tell you something because it is not equally cheap for everyone to produce. The signal has information because the cost of producing it is part of the information.
That is why the fall in the cost of software matters in a way that is easy to miss. The product can be better and cheaper while the product itself becomes a weaker signal. Spence gives us a language for something the startup ecosystem often experiences as intuition.
Claude Shannon was doing something completely different when he wrote “A Mathematical Theory of Communication”, but information theory gives us another useful way into the same problem. Information depends partly on surprise. If I tell you the sun rose this morning, I have not really told you anything. You expected it. If I tell you the sun did not rise, now we have a problem.
Shannon’s paper formalised this intuition more rigorously: common events carry less information than surprising ones.
The same thing happens with software. When only a relatively small number of companies could build and operate credible digital products, seeing one told you something. Now a digital layer is often just the beginning. The product may still solve a real problem. It may even solve it very well. But the fact that it exists is no longer enough to tell you how the economics will behave at scale, what is really difficult about the business, or whether anyone really cares.
And once the signal gets weaker, something else tends to happen. People start looking for another signal.
Charles Goodhart was writing about statistical relationships becoming unreliable once policymakers started using them as targets. Marilyn Strathern later gave the idea its famous formulation, “When a measure becomes a target, it ceases to be a good measure,” in her 1997 paper, “Improving Ratings”.
The point was not that measurement is bad; it was that once a measure becomes the thing people optimise for, it starts changing the behaviour it was meant to observe.
The startup version is not difficult to recognise once you see it. Building software was never supposed to be the target. It happened to be a decent proxy for ambition because it was expensive and difficult. Once it became cheaper, the proxy weakened. The ecosystem, however, did not suddenly throw away the habits built around the old proxy.
So we get the language. Platform. Marketplace. Network effect. AI-native. Infrastructure. Every one of those can describe an extraordinary business. They can also make a very normal business sound like it has been waiting patiently for a Series A.
The words are not the problem. The problem is when they start doing work the underlying business has not done yet. The market is huge. The product is sticky. The network will scale. There is an AI layer. Why now? All perfectly reasonable things to discuss. But a deck can answer all of them and still leave the basic economics untouched.
Nobody has to sit in a room and agree to this. The incentives are enough. Once everyone knows what a “good startup” is supposed to sound like, people learn the language because it works. The person pitching learns it. The person listening learns it. The accelerator teaches it. The template has a box for it. Eventually the box becomes part of the business model (at least rhetorically).
That is how these things usually happen. Not through fraud. Through repetition. A phrase gets rewarded often enough that it stops feeling like a shortcut and starts feeling like the natural way to describe the business. Eventually the language is doing some of the analytical work before anyone notices.
There is a small thought experiment here. Take the technology out of the sentence and see what survives. “AI-powered logistics optimisation platform” sounds impressive. “We help distributors reduce empty truck capacity” sounds less impressive, but now we have something to work with. How much money does the customer save? How often does the problem happen? What happens when the customer base doubles? Is the edge in the software, the data, the distribution or simply in being very good at the work?
Sometimes the answer will be technology. Sometimes it will not. That is the point. Technology can be the answer to the economic question. It should not become the substitute for asking one.
The internet got mature. The bottleneck moved.
Part of the reason this feels different now is that the Internet itself has matured. As Noah Smith has written, most businesses in mature markets already use the Internet and some form of business software. New software can still be valuable, but much of it is now displacing other software rather than replacing a completely offline process.
That changes the nature of the opportunity. When the Internet was young, getting an industry online was itself a large part of the job. Now the Internet is infrastructure. The question becomes: what is the hard part after the Internet arrives?
Eliyahu Goldratt’s Theory of Constraints is useful here because it gives a name to the answer without telling you what the answer has to be. Every system has a limiting factor. Improve everything else, and, unless the constraint moves, you may simply have built a faster way to wait.
Think about e-commerce. You can build a beautiful marketplace in months, onboard thousands of merchants, optimise the checkout and get customers to place orders. But if the goods still have to move through a fragmented delivery network, the website is not the bottleneck. The order can be placed in thirty seconds and still spend two days trying to get from one side of Lagos to the other.
Jumia ran into this exact problem. It built a logistics network because fragmented logistics was one of the harder parts of making its marketplace work and later opened that network to third parties. The point is not really about Jumia. It is that the software can be finished while the system underneath it is still waiting. The difficult part was everything the website could not make disappear.
You can see the same thing in the AI coding boom. A September 2026 revision of an NBER paper, “Writing Code vs. Shipping Code”, uses data on more than 500,000 GitHub developers and their AI usage. The authors find large increases in coding activity across successive generations of AI tools, but those gains shrink sharply as you move from commits to projects and then to actual releases. Coding got much faster. The whole system did not speed up by the same amount.
You can write the product tonight. You cannot make someone want it tonight. You can automate onboarding, but you cannot automate trust. You can ship twenty features before lunch, but lunch was never the bottleneck.
The bottleneck might be distribution. It might be sales, procurement, regulation, working capital or customer support. It might be manufacturing. It might be getting a person to change a habit. Sometimes it is simply that the market does not care enough yet. There is no software update for that.
And sometimes the bottleneck is even more stubborn. Some things still require a person to do the thing. Baumol’s work on unbalanced growth and the cost disease is usually taught as an explanation for why some services stay labour-intensive even as other sectors become dramatically more productive. For this argument, the useful bit is simpler: putting a digital layer around a service does not automatically remove the human time that makes the service exist.
Take tutoring. Put the tutors in an app, add scheduling and payments, maybe throw in an AI assistant for good measure, and the business can become much easier to run. But the student still needs a tutor.
Sometimes the same confusion happens one layer earlier, before we even get to the business model. We talk about whether a market is “digital” as though it were a switch: either the technology has arrived or it has not. It is much messier than that. GSMA’s 2026 Mobile Economy Africa report says that 63% of Africa’s population lived within mobile broadband coverage in 2024 but was not using mobile internet, compared with a 9% coverage gap.
In Nigeria, GSMA’s smartphone research puts smartphone ownership at 27% of the population in 2024 and the mobile-internet usage gap at nearly 60%. So coverage, having a smartphone and actually using the internet are three different things.
The same distinction shows up at the business level. Research ICT Africa’s survey of 718 Nigerian microenterprises found that more than 80% owned a mobile phone, but only 28% had a smartphone and 13% used the Internet for business activities. That is a useful data point, but it is not the whole picture.
In a separate 2023 survey of 250 Nigerian MSMEs already using e-commerce, GSMA found that more than half relied solely on social platforms such as WhatsApp, Facebook and Instagram, while only 14% used social platforms, e-commerce marketplaces and their own websites together.
None of this means these businesses are “not digital.” It means digital adoption has layers. A business can be on WhatsApp, take a bank transfer and still run most of its inventory, procurement and operations much as it always has.
That is why “digitalisation” is such a slippery word. Software can make a business better without making it structurally larger. A digital layer can remove friction without changing the thing that ultimately limits how much the business can produce.
What changes when the business gets bigger?
That brings us to scale. What actually changes when the business gets bigger? Not just whether revenue can go from one million to ten million, but whether something about the economics is different at ten million.
Capital can matter, but only if it changes something underneath the headline growth.
There is a small thought experiment here too. What changes if the business has ten times the money? If the answer is mostly ten times as many people doing roughly the same thing, the business may become larger without becoming fundamentally different. If the answer is that capital lets it build infrastructure, create distribution density, finance working capital, secure a licence or reach a volume at which unit costs change, then the money is doing something much more important.
There is another trap here. A company raises $5 million, hires, expands and grows. The easy story is that the $5 million caused the growth. Sometimes it did. Sometimes the business was already doing something that made the financing available in the first place.
Jang and Kaplan’s 2025 NBER paper on venture capital selection looks at more than 8,000 sourced startup opportunities and finds that selected companies subsequently performed better than evaluated but uninvested companies, while also showing how noisy the selection process can be. The financing event and the business outcome are not the same observation.
The visible parts of a startup story are easy to copy. The round is visible. The valuation is visible. The headcount is visible. The announcement is visible. The boring thing the money actually fixed is usually harder to see.
And that brings us back to the original question. Just because a company is being built with technology does not tell you whether it needs venture capital. What matters is what the capital would actually change: the constraint, the unit economics, the speed to scale or the thing that becomes possible only with it. When the easy things become easier to produce, we have to get better at looking for the things that did not become easier.
Rented ambition
This is why the question of whether something is a “tech company” is less useful than it sounds. A shop can use WhatsApp. A restaurant can take online orders. A distributor can use inventory software. An accountant can automate repetitive work. The fact that the technology is ordinary is exactly what makes it useful.
What has changed is the meaning of the signal. Building with technology used to be reasonable evidence that a company was attempting something ambitious because building the thing was hard enough to filter the field. That proxy has weakened because the act of building has become easier. The business still has to prove itself, but it can no longer borrow as much credibility from the fact that it has software.
There is something quite freeing about that. A perfectly ordinary business does not need to dress itself up as a platform. A service business does not become more interesting because we call it a marketplace. A product does not acquire a moat because an AI model sits behind the login screen. And a company does not become venture-scale because its deck has learnt the right vocabulary.
The real question is much more awkward: what actually gets better when the business gets bigger? What gets cheaper? What compounds? What becomes harder to copy? What remains scarce? What still requires trust, years, capital, relationships, infrastructure or some annoying combination of all five?
That is the distinction the whole argument keeps pointing back to. Some technology changes the economics of the business as it grows. Some technology just makes the existing economics easier to live with. Both can be useful. But if the economics do not change, the presence of technology alone is a weak reason to assume venture capital is necessary.
That is what I mean by rented ambition. Not fake businesses or bad businesses. Often not even badly run businesses. It is what happens when the visible machinery of scale becomes cheap enough to rent before the economics underneath it have caught up.
Computer.com rented ambition through a domain name and a Super Bowl ad. The modern version can do it with a slick interface, an AI wrapper and a deck full of arrows pointing up. The irony is that the technology itself is genuinely doing something wonderful: it is making it easier for almost anyone to build.
Which means building is becoming less interesting as evidence. When everyone can build the product, the product tells us less about the company.
The interesting part is what did not get easier: what is still difficult, what gets better with scale, what gets cheaper, what compounds, what becomes harder to copy, and what technology has not made ordinary? Because that is probably where the business is, and more often than we admit, that is where the ambition is too. The rest may just be rented.
Adefolajuwon Ijaiya is an investment professional working across venture capital, debt financing, corporate development and transaction advisory in Africa. His work sits at the intersection of capital, strategy and execution, helping businesses at different stages become investable, fundable and scalable.
He has worked on equity and debt transactions, fundraising and M&A mandates, and portfolio monitoring across different tech verticals. Based in Lagos, he writes on technology, startups, emerging-market investing and the African business ecosystem.
Last updated: September 29, 2026
