How Will Artificial Intelligence Impact Adaptation Investing?

The shift in society moving from the typewriter to a digital world was massive. The AI revolution will be just as disruptive and happen in a fraction of the time. As I focus on the implications of this when investing in adaptation startups I land on the best opportunities being in fast followers of AI, not first movers or AI software companies.

The two critical pathways for adaptation (resilience and demand-driven responses) will both see their impacts and return profiles elevated by quickly adopting artificial intelligence. Using AI to drive company efficiencies and scale faster will be a key for the winners going forward.

But I don’t think abnormal returns will come from pureplay AI software.

There will be a much better chance of creating defensible moats when there are physical world challenges, tangible services, and people involved in your processes. Having physical products and people in your value chain creates mess. This mess traditionally scares away VCs. I view this mess as an opportunity! Solving the mess creates a defensible moat that you don’t have in AI software.

There are several issues to unpack here. The first is how businesses will create value in a world where artificial intelligence is ubiquitous, and the second, equally important, consideration is what this will mean for investors trying to deploy capital most effectively.

AI Will Disrupt the World, But Who Will Capture the Value?

Folks in the venture industry have invested hundreds of millions of dollars into AI-for-X companies pitching software-based solutions both vertically within industries and horizontally across them. In fact, investment in AI companies drove over 70% of all VC activity.

However, my "hot take" is that there won't be a mega disruptor or trillion-dollar AI software company. Infrastructure (e.g., Nvidia, data centers, etc.) will do well. Data will create value (for a while). Consumers will have great tools. And the software will get commoditized.

Don’t get me wrong - AI will be a massive value creator and some founders have, and will, become rich in AI software. I just think the current valuations are a bubble and will not last.

The scale of this bubble becomes clear when examining recent trends: In the first half of 2025, AI startups received 53% of all global venture capital dollars, according to new data from PitchBook. And that percentage actually jumps to 64% when look at the U.S. alone. Yet this massive capital deployment faces fundamental challenges.

The Climate Adaptation Counterexample

Tellingly, the one area where AI applications maintain both funding momentum and defensible moats is where they intersect with physical world challenges.

Consider climate adaptation: despite representing 12% of all climate tech companies, pure play adaptation startups only received 3% of total funding. Yet even within this underfunded sector, investments overwhelmingly went towards Digital Solutions & AI (44%) and Earth Observation & Sensors (27%). The propensity to privilege digital solutions over the physical remains a persistent problem even in the spaces where the solutions will most likely be intersectional.

This pattern reveals something crucial: pure software AI faces commoditization precisely because it lacks the friction and complexity that creates moats. Research from leading venture firms like Andreessen Horowitz confirms this concern: these startups have weaker defensive moats due to the commoditization of AI models and challenges with data network effects. The report from Andreessen noted that gross margins for these businesses are often in the 50-60% range – well below the 60-80%+ benchmark for comparable SaaS businesses.

The commoditization is already evident in model performance. As Raphaëlle d'Ornano noted in a recent substack, top models—GPT-4o, Claude 3.7 Sonnet, and Gemini 2.5 Pro—achieve near-identical scores (within ~5%) on MMLU and other benchmarks, reflecting commoditization of general intelligence. At most, I think any company in the market that's building a software platform will have a few weeks of traction before a competitor architects their solution with a smaller team and at a lower price point.

The real value will collect around infrastructure, proprietary data lakes (for a short period of time), and with end users who leverage AI within their own businesses—particularly where those businesses interface with irreducible physical world complexity.

The Shift from Software to Tangible Services

As an example of how being a fast follower in adopting AI is beneficial, businesses are already expanding their services with artificial intelligence rather than staff — replacing headcount with bespoke digital agents, assistants, and apprentices. These entry-level automatons create opportunities to rapidly increase volumes of business services that in the past they would have simply been the software product itself.

This transformation is accelerating rapidly. By 2028, Gartner predicts that 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, with at least 15% of day-to-day work decisions being made autonomously through AI agents.

What this means is that service businesses can now operate at scales that was previously the purview of software. The shift from digital tools to artificial intelligence means companies can offer more than just a toolkit, they can perform more functions and do more for customers.

Implications for Investors

For venture investors, the threat of margin compression is compounded by the early-stage bets in tools that were built for a software-as-a-service business model that no longer resonates with what's transforming into on-demand services built on agentic artificial intelligence.

The concentration of capital in AI reveals underlying structural challenges. Rather than create the software that provides services, businesses can now… just provide the services. The risk of compression in enterprise software returns – the foundation for venture capital – is palpable. And while valuations skyrocket with companies reaching millions of dollars in revenues in the near-term, the long-term opportunities may be for the fast followers that can adapt products to specific business cases.

Companies that leverage AI internally rather than sell AI as their product likely have a better chance of monetization without the risk of commoditized pricing for a particular software application. As Kimberly Tan, Marc Andrusko, and Olivia MooreOlivia Moore point out: AI itself is not a moat: it is a way to deliver value to customers. In the fiercely competitive AI market, moats still matter.

Advantages in Building for the Physical World: The Climate Adaptation Moat

As I noted earlier, one of the last domains where competitive moats will disappear is in the intersection of physical and digital services. Here, in the spaces where physical industries have been harder to digitize, the leap to AI commoditization will be much more difficult. It's in these spaces, which are most critical for adaptation to climate change, that investors should be able to find real value.

Market Structure Creates Natural Moats

Climate adaptation represents a perfect case study in how physical world complexity creates defensive barriers that software alone cannot replicate. The market structure itself creates moats through three key dynamics:

Regulatory and Compliance Complexity: Physical infrastructure must navigate complex regulatory frameworks that vary by geography, creating jurisdiction-specific knowledge moats. Unlike software that can be deployed globally with minimal modification, climate adaptation solutions must be deeply customized for local regulations, climate patterns, and infrastructure legacy systems.

Integration Complexity: Physical systems require deep integration with existing infrastructure that took decades to build. Unlike software APIs that can be swapped in weeks, physical system integrations create switching costs measured in years and millions of dollars.

Physical Infrastructure Creates Urgent, Non-Discretionary Demand

Unlike software purchases that can be delayed or substituted, climate adaptation often represents existential business requirements. Data center hubs around the world are all in the top 20 locations for climate risk by 2050, with 20-64% of those hubs projected to be at high risk of physical damage from climate change hazards by 2050, according to a report from the Climate Risk Group’s Cross Dependency Initiative.

Companies facing these risks cannot simply choose to "wait for a better solution"—they must act, creating committed customer bases that software companies rarely achieve.

The Growth Trajectory Favors Early Movers

Looking forward, adaptation investors predict a 20-fold increase in the amount of early-stage and innovation capital for adaptation and resilience companies over the next decade..

This anticipated 20x growth in a currently underfunded market means early entrants can establish market position before competition intensifies. Unlike AI software where late entrants face entrenched, well-funded competitors, adaptation markets remain wide open for companies that can navigate the physical world complexity.

Why Physical World Complexity Equals Competitive Advantage

AI improves adaptation strategies, like identifying vulnerable agricultural regions through drought forecasting, but that data alone won’t be enough. Instead, it’s data that can be used in tandem with interventions like OlsAro, which makes drought-tolerant grains, or ALORA, whose ocean agriculture systems are being developed to address climate-related challenges to food systems globally.

The value isn't in having better AI algorithms—it's in using those algorithms in tandem with physical data and services to create both forecasting tools and opportunities for interventions.

This bundling of AI with physical world complexity, local knowledge, and infrastructure integration creates moats that pure software companies cannot replicate by simply improving their algorithms. The defensibility comes not from the AI, but from the ability to make AI that’s useful in the physical world.

What investments win in a warmer world

The data on investment trends reveals a massive disconnection between what the world needs and what markets currently value. While venture investors compete with big dollars for businesses with historically low margins investing tens of billions of dollars, climate adaptation represents a $1.4 trillion annual market that receives only 3% of climate tech funding.

Some venture investors and analysts are beginning to recognize the disconnect, and there will be much more investment in these spaces in the coming years. I think returns will begin to flow ot the companies that combine AI's power to tackle the complexity of physical world challenges—particularly in climate adaptation where massive investment needs meet genuine technological barriers to commoditization.

In climate adaptation, this combination of technological leverage and physical world complexity creates the defensive moats that the pure software AI world increasingly lacks.

The opportunity to accelerate growth through being an AI fast follower allows physical companies to have growth rates closer to software and still be VC investible.

I think early movers in this space will build the infrastructure businesses of tomorrow, while avoiding the commoditization trap that awaits pure software AI companies.