# The Great AI Consolidation: Why the Next Winners May Be Established Companies

Author: Norbert Biedrzycki
Published: 2025-11-15
Language: en
URL: https://norbertbiedrzycki.pl/en/the-great-ai-consolidation-why-the-next-winners-may-be-established/
Summary: In the early years of every technology wave, the startups set the pace. Later, the incumbents with customers, data and distribution catch up. In AI, that moment may be arriving now.

**In the early years of every technology wave, the startups set the pace. Later, the incumbents with customers, data and distribution catch up. In AI, that moment may be arriving now.**

Three years after ChatGPT, access to a capable language model is no longer a competitive advantage. Strong models are available from several American providers, from open source projects and from Chinese labs, at prices that fall every few months. Customers increasingly expect AI capabilities to be part of the software they already use, not a separate product they need to buy.

When the core technology becomes widely accessible, the competitive advantage moves elsewhere: to distribution, customer relationships, proprietary data and deep knowledge of specific industries. These are exactly the assets that established companies have, and that most startups lack.

**The consolidation wave**

2025 has been a year of large AI related transactions led by incumbents. In March, Google agreed to acquire the cloud security company Wiz for $32 billion, its largest deal ever. ServiceNow bought Moveworks to strengthen its agent offering. In May, Salesforce agreed to acquire Informatica for around $8 billion, explicitly to strengthen the data foundation for its AI agents. In June, Meta invested $14.3 billion for a 49 percent stake in Scale AI and brought its founder in to lead a new superintelligence effort. In July, the race for the coding startup Windsurf ended with Google paying around $2.4 billion to license its technology and hire its leadership, and the rest of the company going to Cognition.

The pattern is clear. Large companies are buying data, talent and capabilities that make their existing distribution more valuable.

**The adoption gap**

At the same time, research from MIT published in August found that around 95 percent of enterprise generative AI pilots were delivering no measurable impact on the P&L. The study also noted that solutions purchased from specialised vendors succeeded considerably more often than those built internally, and that the biggest returns were often in back office functions rather than in the highly visible sales and marketing pilots.

I find this consistent with what I observe in portfolio companies. The obstacle is rarely the model. It is integration into workflows, data quality, change management and ownership. These are areas where companies with existing operations, customer trust and industry expertise have an advantage.

**What it means for private equity**

For private equity, this strengthens a strategy I have been thinking about for several years. Instead of competing for expensive AI native startups at venture style valuations, a fund can acquire established businesses with strong market positions and proprietary data, then transform them with AI.

A regional insurance broker, a specialised software provider for a vertical industry, a business services company with recurring contracts or an industrial distributor with decades of transaction data may all be better AI investments than a startup with impressive technology but no distribution.

The value creation thesis is not that these businesses become technology companies. It is that AI improves their margins, deepens their customer relationships and strengthens their defensibility, while they are acquired at multiples that reflect a traditional business.

**The risk on the other side**

There is, of course, a risk. Not every incumbent will adapt. Some will be disrupted by AI native competitors that reimagine the service entirely. The skill for investors is to distinguish between incumbents whose assets become more valuable with AI and those whose business model AI will hollow out.

*Investment question: Could the greatest AI investment returns come from transforming existing businesses rather than creating new AI companies?*
