I don't believe in the future of lock-in business models.
Siemens just showed the playbook at VivaTech, with Roland Busch on stage: heavily discounted engineering software for startups, an Industrial Copilot that writes and validates code, all under the Industrial AI revolution banner. But a discount that steep is less strategy than signal. When a startup can prompt a working tool in an afternoon, the old enterprise license is hard to justify.
The whole bet rests on the lock-in holding later, and that is getting thin from two sides. AI lets a competitor build in weeks what took years, and the incumbents are too slow to react, not because the people are weak but because big orgs decide at big-org speed. By the time the discount becomes lock-in, smaller teams have shipped around it.
There is a catch the closed vendors miss too. An AI agent is only as good as the code it can read. On open source it understands and extends the system. Behind a closed binary it is guessing. The better agents get, the more source access decides what they can do, which is exactly what a locked platform withholds.
So we bet the other way. Open instead of closed, AI on the digital twin over an open MCP server, not inside one vendor's garden. When everyone can build faster, open is the durable choice. Whoever is fast and open does not need lock-in.






