If work software cost what it costs to run: a thought experiment
A thought experiment. It's 2031. Every tool your company uses (project management, CRM, calendar, analytics, docs, chat) is open source. Teams that want to self-host do. The rest pay a hosted version by usage, close to what the infrastructure costs.
Two trends already point there: AI tools made writing software much cheaper, and cloud infrastructure has always had a public price list.
The software budget shrinks
Today a 50-person company pays about $30 per user per month for a project tool, $50 for a CRM and $12 for a scheduler. Together that's easily the cost of one more hire, every year.
In this scenario the line shrinks to its real cost: servers, backups, bandwidth. The money moves to what the tools can't do: people who understand the business, clean data, security and automations built for how the company works.
Software becomes a utility
Nobody debates the water bill in a board meeting. When work software gets cheap enough, it joins that list. Companies end up running more of it, not less: the tools that "didn't justify the license" get installed anyway.
Where the value moves
When copying costs nothing, value collects in what can't be copied:
- Data and trust. A customer relationship can't be forked.
- The physical world. Chips, sensors, robots, energy.
- Models trained on your own data, running on your own infrastructure.
- Judgment. People who decide well with tools everyone has.
Vendors selling seats have two options: charge for running the software, or defend a margin the market no longer supports. Music, newspapers and cable TV show how the second option tends to go.
Where we are
zerosoft is our bet on that scenario: eleven open source apps under AGPLv3, $0 to self-host, and a cloud billed per workspace by usage (clicks, bookings, emails, storage) with free seats. In 2031 nobody says "we bought a CRM". They say "we use one", the way they say it about email.