Why AI Won't Kill SaaS — It'll Just Make It Cheaper
Last week I had a confidence crisis regarding my SaaS/AI thesis. When I first saw ChatGPT a few years ago, I thought that AI was going to kill software given enough time. Just as software greatly dematerialized our world, AI was going to kill most of SaaS. Only permissioned companies (banking, for example), companies that make a physical goods, or companies controlling/owning a network are safe from AI.
My original thesis was updated last year: instead of killing SaaS, AI will unravel the premiums seen on SaaS companies. Premiums will drop, both on the price that they charge for software and their stock price. The new thesis is that there will be one-man shops (or a few people) that pop up like mine (Athena) and we will compete heavily on price. We won't have all the bloat that traditional SaaS companies have.
My original thesis was updated last year: instead of killing SaaS, AI will unravel the premiums seen on SaaS companies. Premiums will drop, both on the price that they charge for software and their stock price. The new thesis is that there will be one-man shops (or a few people) that pop up like mine (Athena) and we will compete heavily on price. We won't have all the bloat that traditional SaaS companies have.
Last week I was thinking about how easily companies can already vibe code their own alternatives and my confidence in the thesis waivered. With Claude's Opus 4.6, it is becoming a lot easier than it ever has. My fear was that instead of paying for an external solution, practitioners will want to spin up their own custom-made solutions.
After a call to a good friend, and some more thought about the space, I am back to the original thesis.
While some companies may replace some of their tools with home-built solutions, I think it doesn't make sense for most. Most people don't have enough spare time at work to spin up a their own tools, even if LLMs are getting really good. There are also some companies will just be slow to adopt AI. At the end of the day, the decision will come down to how much time and effort it will take to make your own solution.
For example, if you are paying tens of thousands of dollars for a lease tracking tool, having the controller, or someone on the accounting team, create that tool in house would definitely save the company money. If there are tools that only cost a few thousand dollars per year, it wouldn't make sense to build your own tools.
To be competitive, existing players (or new players) will have to make their offerings high quality products and cheap enough that it doesn't make sense to spend precious company time to make those tools. If a solution is only a few thousand dollars a year, when you factor in the cost of the AI tools (e.g., hosting, cost of time. opportunity costs, AI subscription costs, security concerns), the decision to build and maintain the tool in house will not pan out.
To be competitive, existing players (or new players) will have to make their offerings high quality products and cheap enough that it doesn't make sense to spend precious company time to make those tools. If a solution is only a few thousand dollars a year, when you factor in the cost of the AI tools (e.g., hosting, cost of time. opportunity costs, AI subscription costs, security concerns), the decision to build and maintain the tool in house will not pan out.
Building your own tools does not help the company advance its own products or generate more revenue, and it does not always help practitioners do their jobs more effectively. The reason that a company will pay hundreds of thousands of dollars to external companies on their CRM instead of building it in house is that their developers are more valuable building tools their clients can use.
A company should focus on what they are best at, not trying to do everything themselves. It wouldn't make sense for a company to build their own version of Excel, Word, and PowerPoint so that they can get rid of their Microsoft 365 subscription; the math just doesn't add up. In fact, Microsoft will probably add more Microsoft 365 subscriptions and raise prices over the coming years.
Software companies have had a lot of bloat in their costs that require them to charge exorbitant fees. There is such a premium we have placed on software that they are able to get away with inefficiencies and thick premiums. So, for every tool that a company will try to vibe code in house, it really comes down to what the alternatives in the market are costing. If there are no good options in the market, then you might make it in house if you have the right people overseeing the project (it needs to be built to the right specs and not have security holes). More likely, for most tools, you will see new companies popping up that can offer the product for a much more affordable price.
A company should focus on what they are best at, not trying to do everything themselves. It wouldn't make sense for a company to build their own version of Excel, Word, and PowerPoint so that they can get rid of their Microsoft 365 subscription; the math just doesn't add up. In fact, Microsoft will probably add more Microsoft 365 subscriptions and raise prices over the coming years.
Software companies have had a lot of bloat in their costs that require them to charge exorbitant fees. There is such a premium we have placed on software that they are able to get away with inefficiencies and thick premiums. So, for every tool that a company will try to vibe code in house, it really comes down to what the alternatives in the market are costing. If there are no good options in the market, then you might make it in house if you have the right people overseeing the project (it needs to be built to the right specs and not have security holes). More likely, for most tools, you will see new companies popping up that can offer the product for a much more affordable price.