Opinion: AI is tearing up the old SaaS rule book and pushing venture capital towards a barbell of mega-bets and tiny cheques. Australia needs to pay close attention, argues Daniel Petre
Venture capital, lazily defined, is capital invested into high-risk and hopefully high-return companies that disrupt tired market sectors with genuinely new products.
That’s been the deal for 40-odd years. It’s not the deal anymore, or at least not in the way most of us learned it.
The SaaS and e-commerce era is over
Over the past ten years the US has poured close to US$1.92 trillion into roughly 142,000 venture deals. Everything from angel cheques to nine-figure Series E rounds. Australia, over the same period, managed somewhere around US$22 billion (call it AUD$31 billion) across about 3,200 deals.
Different scale, obviously, but for almost the whole of that decade both markets behaved the same way. Money spread reasonably evenly across stages, and SaaS and e-commerce hoovered up most of it.
The rule of thumb was blunt but reliable. Get your SaaS company past maybe US$20 million of revenue and you were more or less destined for a good outcome.
That rule is dead. Actually, scratch that – it’s not dead, it’s been murdered, and I want to talk about the weapon.
Call it the SaaS apocalypse, or ‘SaaSpocalypse’ as it has come to be known. The term is overblown, but the underlying threat is real. The moats around scaled SaaS businesses, many built over a decade or more, have turned out to be far shallower than investors had assumed.
AI models can now expand into vertical use cases almost overnight. A three-person team with API access can spin up something that is competitive with a supposedly defensible incumbent in months, not years.
I am not suggesting every SaaS company will be taken out by some vibe-coded app. But some will, and all SaaS companies will have to work a lot harder to defend their turf.
The Cursor warning
Growing 30 to 40 per cent a year used to be the golden ticket. Not anymore.
Cursor is the cleanest case study going. It grew from roughly US$400 million in annualised revenue in 2024 to US$4 billion annualised run rate in 2026, raising US$3.4 billion along the way. By early 2025 it was being called the dominant player in AI coding assistants. Full stop, no argument.
Then Anthropic shipped Claude Cowork and rapidly improved the coding capabilities of its models. Within months Cursor’s ‘dominant’ narrative looked shaky. Cursor wasn’t clearly ahead anymore, arguably behind on functionality, and – more unsettling for anyone holding the stock – if Anthropic could ship something that good, that fast, what happens when Google or OpenAI properly turn their attention to it?
Cursor tried to raise at US$50 billion in Q2 2026, 1.6 times its late-2025 mark. The market said no, despite revenue still growing fast. It ended up selling into Musk’s SpaceX/Grok stable for shares, at a US$60 billion valuation.
Read that sequence again. A company that grew revenue tenfold in two years still couldn’t raise capital at just 1.6 times its previous valuation.
The challenge facing mid-stage SaaS companies
SaaS valuations have been substantially downgraded in recent years. While there are examples of companies that have recovered some of the valuation downgrades, most, if not all, are valued far lower than they were two years ago.
These companies may exhibit strong revenue growth and user retention, but most investors regard them as having less of a moat than before.
If you were a VC willing to pay 15–25 times revenue for a SaaS company growing at 35 per cent a year, on the assumption that its moat would protect continued revenue and profit growth, chances are you see that investment very differently now. Love the company, but I am not sure it is worth that much anymore.
As for new SaaS companies? Maybe they can grow for a while but are they really building something defensible? Unless a company can show how it will withstand competition from LLMs and AI-native applications – in other words, unless it has a credible AI story – investors may simply avoid mid-stage, non-AI startups that are performing well but cannot explain why AI won’t eventually erode their advantage.
If you’re a VC watching situations like the SaaS apocalypse play out, as well as the expansion and visible acceleration in frontier model capability out of both the US and China, what do you actually do? You don’t do what you did in 2019. You can’t.
Worrying signs from US VCs
Here’s the number that should worry anyone who thinks the current US boom is healthy: US$340 billion was invested by American VCs in 2025 – up from US$214 billion in 2024 and closing in on the 2021 peak of US$360 billion.
But 50 per cent of that US$340 billion went into just 0.05 per cent of the deals done. Of 16,709 total investments, roughly 480 deals, or 2.9 per cent of the count, soaked up 70 per cent of the money.
The sub-US$100 million deal volume hit a 14-year low in Q4 2025. In Q1 2026, 72 per cent of all VC dollars went to AI companies, big and small. Early-stage companies held their ground in terms of number of deals done but, unless you were a fast scaling AI native app or platform, everything from Series A onwards struggled to get funded – regardless of how impressive the growth or retention metrics were.
Australia can’t produce an OpenAI or Anthropic at anything like their current scale. We don’t have companies capable of absorbing cheques of that size, nor VC funds that can compete with the largest US players for meaningful positions in late-stage LLM and other major AI rounds. But the same instinct is visible here on a smaller scale: 61 per cent of Australian venture capital in 2025 went to companies that could credibly wave the AI flag, woo-woo included.
So the picture in the US is a barbell: enormous, valuation-agnostic bets on the handful of companies everyone’s already decided are winners, and pixie dust sprinkled thinly across everyone else on the off-chance one of them turns out to be the next big thing. In Australia we have a barbell of sorts – big deals at one end and small at the other, with little in between. However, our big deals aren’t that large on a global scale.
Neither end of that barbell requires much skill. Writing a cheque into a round everyone else is also fighting to get into isn’t deep analysis, it’s queueing. And sprinkling small cheques across hundreds of companies isn’t a strategy, it’s a lottery ticket dressed up as a portfolio.
Where we are headed
So where is this going? First, the scale players – the top handful of general-purpose US VC funds – will move fast to fund late-stage, high-growth companies that exhibit AI resilience/functionality, globally. Their fund size gives them a dominant position in this market. Australian firms are not at the table or even in the room.
US firms will also keep using their global sourcing networks to sprinkle pixie dust, backed by massive brands and tonnes of capital to deploy. Expect them to compete even harder for allocation into fast-scaling AI-native or AI-ready non-US founders, Australians included, and expect that to squeeze the mid-sized generalist fund everywhere, including here. They have the size to play both ends of the barbell.
Second, non-traditional capital – sovereign wealth, including the mooted new mega-funds coming out of the UAE; superannuation funds investing directly; private equity; hedge funds – will keep pushing further into late-stage rounds. Why wouldn’t they? Writing a growth-equity cheque into an already anointed winner requires no special VC skill, and those pools of capital dwarf anything the venture industry can bring to the table.
Third, and this is the one that actually gives me some optimism for Australia: the historical data has always shown bigger funds return less, not more, than smaller funds. That should, eventually, push serious capital back toward smaller, deeply specialised funds: deep tech, drug discovery, hardware-software combinations with a genuine moat, the kind of long-dated, patience-testing bets that don’t fit an AI hype cycle. These companies need investors with genuine domain expertise and the patience to stay with them.
The Australian story
In Australia our VC landscape is dominated by general-purpose ‘we can do everything’ VC firms with only a handful of deep tech/sector specialist firms. Given Australia punches above its weight in many areas of global R&D, the fact that we do not have many deep tech-focused VC firms is genuinely concerning.
Over time it may well be that smaller, deep tech or sector-specific funds outperform the larger firms and, at the end of the day, when all the smoke has cleared, it is all about investment returns.
The next five years, I suspect, will start rewarding patience and genuine specialisation again because eventually the pixie dust runs out, and someone has to pick winners based on actual analysis.
Hopefully the institutional funders of VC in Australia become a little more engaged with helping take our R&D expertise and commercialising it by funding smaller specialist VC firms that exhibit clear domain knowledge and focus. This shift in the Australian VC landscape is both coming and needed.
Daniel Petre AO is the co-founder and Partner Emeritus of AirTree Ventures, one of Australia’s leading venture capital firms. He has spent more than three decades in technology, investment and innovation, including senior roles at Microsoft.
Want to see more Forbes articles on your feed? Tap here to make Forbes Australia a preferred source on Google.
Look back on the week that was with hand-picked articles from Australia and around the world. Sign up to the Forbes Australia newsletter here or become a member here.
