Anton Osika 5.0 6 ideas

"The legacy software enterprise software companies like SAP, like Oracle, like Workday... I would not be surprised if the amount of profits from these companies continues to decline as we're seeing the 99% for everyone being able to recreate what took them ten years to build in the past in one week." Legacy enterprise SaaS companies rely on high switching costs, massive development barriers, and monopolistic data silos to maintain high margins. AI coding agents completely destroy this moat by lowering the cost of custom software creation to near zero. Instead of paying expensive recurring licensing fees for rigid software, businesses will generate their own bespoke tools, leading to severe margin compression and customer churn for legacy providers. SHORT because the fundamental business model of charging rent for generic, siloed enterprise software is being disrupted by cheap, hyper-customized AI generation. These legacy giants successfully pivot by integrating AI into their own platforms, leveraging their massive existing distribution networks and entrenched enterprise relationships to actually increase prices and retain market share.

"Then they now build like we're doing at Lovable build out their own technology stock where they make sure that this is the only data that all of these tools that people live in... can share all of that data seamlessly." If every enterprise shifts from buying off-the-shelf SaaS to building and running their own custom AI-generated software stacks, the underlying infrastructure requirements will explode. Companies will need massive amounts of cloud compute, database hosting, and AI inference power to run these bespoke internal systems. The hyperscalers act as the ultimate toll collectors for this transition from "renting software" to "computing custom software." LONG because hyperscalers will capture the infrastructure spend that is redirected away from legacy SaaS licensing fees. AI coding efficiency drastically reduces overall enterprise compute bloat, or companies shift toward localized, on-premise AI models to protect their proprietary data, bypassing public clouds.