Mistral CEO says AI could replace 50% of enterprise software

Mistral AI CEO Arthur Mensch predicts that over 50% of current enterprise software could be replaced by AI, sparking concern among investors about the impact on software stocks.

Mensch said that enterprises with proper infrastructure could quickly develop AI-driven applications to automate workflows, including procurement and supply chain processes.

He added that many businesses are considering replatforming their IT systems to leverage AI for increased efficiency and cost reduction.

While systems of record software are unlikely to change, Bipul Sinha believes workflow software could be significantly disrupted by AI.

The company plans to open its first office in India this year, partnering with local firms to establish infrastructure and develop models tailored to Indian languages and data regulations.

Mensch emphasized that AI could enable faster software development and new operational models for enterprises.

🔗 Source: CNBC

🧠 Food for thought

Implications, context, and why it matters.

AI shifts enterprise work toward automation

Arthur Mensch expects software to replace software. Many teams may instead use AI apps to take on tasks handled by employees.
New products act as Systems of Work. They run tasks on their own, such as drafting reports or answering customer service requests 1.
The prize goes beyond software spend. AI agents can stand in for roles across the $11 trillion U.S. labor market 2.
Mensch expects in house builds. Andreessen Horowitz (a16z, a Silicon Valley venture capital firm) found over 90% of surveyed enterprises testing third party AI apps for customer support, plus one public fintech stopped an internal build to buy a third party tool 3.

AI layers sit above systems of record

Core Systems of Record such as a central CRM (customer relationship management system) will likely stay in place.
Many startups build over the top workflow layers. These layers use existing databases then automate work that used to happen inside those tools 1.
Legacy platforms can slide into a data storage role. More value moves to AI systems where the work runs.
Pricing can shift from per seat licenses to usage or outcome based fees tied to AI labor 2.
Adoption remains uneven. Nearly 30% of pilot AI projects get dropped, plus skill gaps still slow rollouts 4.