Private Markets Community

Hello Community!

PitchBook's 2026 Outlook reports show that AI accounted for 65% of total VC deal value in 2025, and this momentum is expected to continue shaping VC deployment in 2026. This includes enterprise SaaS, defense tech, carbon tech, mobility, fintech, and cybersecurity. The report is attached here or you can download from our library.

AI is now the lion's share of global VC deployment - 65% of VC deal value last year. As we enter 2026, AI remains the defining theme for venture investors across sectors from climate to cybersecurity. What does this concentration mean for diversification and risk?

Attachment(s)

2026-artificial-intelligence-outlook-the-great-competitio...pdf   1.16 MB 1 version

Good question. At BXR Group (bxrgroup.com), we have spent a lot of time asking ourselves what this concentration means for diversification and risk, ultimately deciding to make our latest VC AI investment into Energetico (energetico.com), whose technology and commercial solutions are revolutionizing HVAC for the Energy & AI Era, but have broader use cases than just AI / datacentres (thus increasing diversification and reducing risk).

------------------------------
Brendan James Smart
Principal
------------------------------

Thanks for sharing, Aya. To me, the 65% headline is less about "AI is a category" and more about correlation (a lot of VC is now moving together). When so much money is in AI, many portfolios can end up exposed to the same few things: whether IPO/M&A markets reopen, how expensive it is to build and run AI, and what a small number of big platforms decide to allow or charge.

PitchBook also notes that a big share of AI funding is going to a relatively small group of core AI/model companies, and only a few models(LLMs) are really available at large scale. So even if you're investing across climate, fintech, and cybersecurity, you can still be indirectly relying on the same underlying platforms.

I think diversification is still possible, just not by "AI vs non-AI" labels. It's more about mixing exposures: compute-heavy vs lighter inference use cases, infrastructure/enablers vs app layer, different moat types (data/regulatory/distribution), and being careful about the most crowded "copycat" areas.