Why Insight Partners Is Spreading Bets While Others Focus on AI Labs
Insight Partners co-leader Devin Parekh defended a diversified investment strategy while other venture firms have concentrated capital in frontier AI labs, the companies building the largest AI models. Parekh, who has co-run the $90 billion firm for 26 years, spoke in a sit-down interview with TechCrunch at its StrictlyVC event in New York on Sept. 13, 2026. TechCrunch
Insight Partners is a global software investment firm. Columbia Business School Parekh serves as a managing director with a remit covering e-commerce and application software businesses on a global basis. Insight Partners
That software focus includes sustained work in data infrastructure, the systems that store and manage large amounts of data. Insight Partners has led and co-led numerous funding rounds for Databricks. TechCrunch
On buyouts, the pace has slowed. Insight Partners has not completed a major buyout since 2024. TechCrunch
Legal technology illustrates the applied side of that breadth. Legora is a collaborative AI platform for lawyers. Legora It develops artificial intelligence software to support legal professionals with tasks such as document review and drafting. Insight Partners Its $80 million Series B round was led by ICONIQ and General Catalyst. Legora General Catalyst co-led the investment. General Catalyst
The broader context here will be familiar from past platform shifts. Capital concentrates at the infrastructure layer first, the base models and computing power. Application software looks diffuse by comparison. It moves with procurement cycles, integration work, and trust in regulated workflows. Enterprise buyers rarely rip out systems of record, the core software they use to run the business. They add copilots, agents, and retrieval layers, assistants that sit on top and pull in the right information.
In my view, that is why Parekh's position matters for builders. Diversification looks cautious when attention centers on a few model providers. It is also how enterprise value has usually accrued. Data platforms, vertical tools, and profession-specific assistants do not replace infrastructure bets. They compound them, turning raw model capability into usable systems with permissions, audit trails, and domain constraints.
Worth flagging for technical teams is what this means for architecture choices. If significant capital continues to support applied software and data infrastructure, engineering leaders retain leverage. They can route workloads across APIs, open weights, and specialized tools. They can optimize for inference cost, latency, and data governance, the price, speed, and control of running models, rather than a single vendor roadmap. Complexity rises. Optionality rises with it. That keeps data gravity, access controls, and evaluation harnesses central to deployment decisions.
My own experience watching younger users adopt new tools is that utility outlasts novelty. Frontier models matter. What changes daily practice is the layer around them. Search that respects context. Drafting that tracks precedent. Review that holds up under scrutiny. A $90 billion software investor choosing breadth is a bet that durable gains will come from that layer. It is a quieter thesis. It ages well.
