Biotech vs AI vs Climate Tech: How Research Momentum Differs Across the Three Largest Innovation Verticals

The three verticals that attract the most capital, talent, and policy attention in technology investment are artificial intelligence, biotechnology, and climate tech. Each produces enormous volumes of scientific preprints. But the shape of that output, the acceleration patterns, the geographic distribution, and the internal theme diversity differ in ways that carry direct implications for how investors should time their entry and where they should focus diligence.

The Finch Innovation Index tracks all three verticals across its 73 investable technology themes, providing a structured view of how momentum scores, keyword emergence, and publication geography vary not just between verticals but within them.

Volume Is Not Momentum: Why Raw Preprint Counts Mislead

AI dominates raw preprint volume among the three verticals. Repositories like arXiv receive tens of thousands of AI-related submissions per month across machine learning, computer vision, natural language processing, and robotics subfields. AI preprint volume on arXiv has grown at roughly 25 to 30 percent annually over recent years. But volume alone is a poor proxy for innovation momentum. A vertical can produce enormous output while decelerating in novelty, as measured by the rate at which new keywords, methods, and subtheme clusters appear.

Biotech preprint volume is lower in absolute terms but distributed across a wider array of distinct subthemes. Biotech research spans genomics, synthetic biology, drug delivery, protein engineering, immunotherapy, and dozens of other areas, each with its own publication cadence. This means biotech's momentum signal is more fragmented, requiring theme-level resolution rather than vertical-level aggregation.

Climate tech preprint output is the smallest of the three in absolute volume but has shown the steepest rate of acceleration in recent years. Climate tech preprint output has shown the steepest percentage growth rate among the three largest innovation verticals. This acceleration is driven by subthemes like carbon capture, grid-scale storage, green hydrogen, and advanced photovoltaics, each of which has seen surges in new methodology papers. As we discuss in our piece on how momentum scoring captures acceleration rather than size, percentage growth and keyword novelty often matter more than raw counts for identifying emerging investment windows.

Internal Theme Diversity and Where Momentum Concentrates

AI research momentum is heavily concentrated in a small number of subthemes at any given time. Large language models and generative AI have dominated AI momentum scores since 2022, pulling attention and publication energy away from other AI subthemes like reinforcement learning and symbolic reasoning. AI research momentum is heavily concentrated in a small number of subthemes, particularly large language models and generative AI. This concentration creates a paradox: the vertical looks uniformly hot at the surface, but most of the acceleration comes from two or three theme clusters.

Biotech momentum is more evenly distributed across its constituent themes. No single biotech subtheme captures more than 15 to 20 percent of the vertical's total momentum score in a typical quarter. Biotech momentum is more evenly distributed across subthemes than AI, with no single area exceeding roughly 20 percent of total vertical momentum. This breadth makes biotech harder to track with simple keyword filters but more resilient to hype cycles in any one area. Rising keyword detection, as described in our analysis of theme emergence signals, is particularly valuable here because new biotech clusters can form quickly around specific targets or modalities.

Climate tech sits between these two extremes. Climate tech research momentum is increasingly bifurcated between hardware-oriented energy themes and software-oriented climate modeling. Energy storage and hydrogen production themes carry strong momentum, while adaptation-focused research grows more slowly.

Geographic Patterns Reveal Structural Differences

The geographic distribution of research output differs markedly across the three verticals. AI preprint output is dominated by the United States and China, which together account for more than 60 percent of high-impact AI publications. The US and China together account for more than 60 percent of high-impact AI preprint output. Biotech is more geographically distributed, with significant contributions from the UK, Germany, Japan, and South Korea alongside the US and China. Climate tech research shows the strongest European presence of the three verticals, reflecting the EU's policy-driven investment in decarbonization research. Climate tech research shows the strongest European geographic presence among the three verticals, reflecting EU policy-driven R&D investment.

These geographic patterns matter for sovereign wealth funds and multinational corporates assessing where research-to-commercialization pipelines are forming. The Finch Innovation Index surfaces country-level publication patterns as part of its geographic intelligence layer, enabling users to track shifts in national research competitiveness across all 73 themes.

What This Means for Investment Timing

Each vertical demands a different analytical approach. AI investors need to monitor theme concentration risk: when momentum narrows to one or two subthemes, late-stage capital crowds in and early-stage signal degrades. AI investors should monitor theme concentration risk because momentum narrowing often precedes late-stage capital crowding. Biotech investors benefit from broad theme scanning at higher resolution, looking for acceleration in specific modalities or targets before they consolidate into named fields. Climate tech investors should weight percentage growth rates more heavily than absolute volume, since the vertical's smaller base means high-momentum subthemes can shift from niche to investable faster than in AI or biotech.

The common thread is that vertical-level analysis is insufficient. Momentum scoring at the theme level, combined with keyword emergence and geographic tracking, provides the resolution needed to distinguish signal from noise across all three verticals.