Will AI kill SaaS?
Cost pressures are temporary. SaaS will face disruption driven by changing market needs. Focus on your customers to develop your strategy.
I am still involved in many SaaS businesses. SaaS has attracted a fair bit of comment in recent weeks, around one basic question. Will AI disrupt the SaaS business model?
Many people think the answer is yes. Based on one key argument. The cost of compute for a product using AI. The argument runs:
SaaS is based on ultra low cost cloud infrastructure.
In this model, the cost of compute is largely irrelevant and gross margins are high.
This means the most common revenue model is per seat pricing. The more users, the more company the money makes.
AI based on LLMs is expensive to run and that cost grows depending on usage.
Therefore SaaS companies must move to a usage based pricing model.
This kind of linear logic has never made sense to me. It doesn't align with my experience of how the world works or my amateur knowledge of business and economic theory.
In this case, there is one simple reason the argument doesn't add up. The cost of compute is not going to stay high forever.
Partly based on Wright's Law which shows that the more times an operation is performed, the lower the cost. Research shows this leads to an exponential curve where cost reduces rapidly over time.
Experience shows this to be true in the tech industry. Moore's Law is a famous example of this principle in action.
We can already find evidence that this will happen for AI. In this paper from [[azeem azhar]], the authors conclude:
"Instead of being anxious about a lack of computing power, executives should be gearing up for an abundance of compute."
In reality the questions SaaS companies should be asking about compute cost are:
How fast will the cost fall?
Will investors hold their nerve while this unfolds or start making panic changes now?
Back to the basic question. I still think the answer is Yes. AI will lead to disruption in SaaS. Cost of compute will not be the main driver of that change.
Thinking about how to handle disruption is hard enough without being distracted by irrational extrapolation effects that won't last. Here are a few better options:
There may be good reasons to rethink your pricing. Per seat pricing is common in SaaS but there are plenty of other models around, including some based on usage. AI might change how your product delivers value. It may even open up new, more value based models.
There will be winners and losers. No doubt some big failures but a new growth phase for others. And inevitably some great new names will emerge from the current wave of startups.
Customers will drive who wins and loses not costs. The big changes come from new customer needs, new user behaviours and new markets.
So disruption yes but destruction no. If you want to build a new SaaS business or drive growth in your existing business, look at your customers for the strategic answers, not your cost base.
Doing the next right thing
AI is going to lead to disruption. If you want to explore what that means for your business and you need a sound board, a challenge, a foil, that’s what I do every day. Please get in touch.
Thanks for reading.