Software Needs A New Business Model
Software needs a new business model!
What is crystallizing for me is that we may be seeing a regime change in the software industry.
The past decade+ saw the huge success of asset-light SAAS companies, with low capital intensity and high human capital requirements. Under those conditions, a bigger team was an advantage so companies grew headcount quickly: all else being equal, more developers meant more features, more configurability, etc. which meant higher revenues. The marginal cost for serving each additional customer was minimal on a relative basis, so development costs were front-loaded.
AI turns that on its head. Inference at runtime is and will continue to be expensive as we perform progressively more sophisticated activities using AI, so OpEx increasingly moves from primarily salaries to compute resources; not to mention, CapEx for data center buildouts which someone has to eventually pay for. This turns software companies into high capital intensity businesses (much like an industrial or mining company); at the same time, AI is a multiplier for the actual product development, so a smaller, leaner team would be a bigger advantage in this situation.
In other words, over the next decade a bigger headcount will likely change from being primarily an asset to a liability for software companies. The implications for employment in this sector are troubling; but as we've seen in other industries in the past, broad-based changes like this cannot be stopped, only adapted to.
