Why We Backed Presage Labs
The shift from human-driven to agent-driven compute is no longer theoretical. As AI agents take over corporate workflows, financial operations, and infrastructure management, a clear gap has emerged: the cloud was never built for this.
Presage Labs is building the predictive infrastructure layer this transition requires. Here is why we invested.
The Problem: Infrastructure Built for Humans, Not Agents
Cloud architecture was designed around predictable, human-paced request patterns. AI agents break that model entirely. Operating at machine speed, they trigger billions of micro-transactions simultaneously creating load spikes that have no historical precedent.
Traditional monitoring tools are reactive by design: they tell you what broke, not what is about to break. When agent networks shift load in microseconds, reactive autoscaling is already too late. What is needed is accurate prediction, not faster reaction.
Compounding this is a structural supply constraint on compute itself. Three converging pressures make brute-force capacity expansion increasingly unworkable:
- Fabrication limits: Advanced silicon supply remains tightly constrained by global manufacturing capacity.
- Power grid constraints: Data centers face a 19 GW deliverable power shortfall. Large clusters cannot be powered on schedule.
- Regulatory friction: Local zoning and environmental approvals are slowing physical expansion for major cloud providers.
The answer is not more compute, it is using existing compute better. That requires understanding not just what a system is doing, but what it is about to do.
The Solution: Causal World Models
Statistical models trained on historical data cannot predict the behavior of novel, emergent agent networks. You need a model that understands the underlying causal dynamics of the system. One that can simulate what happens next, not just extrapolate from what happened before.
Presage Labs is building Causal World Models purpose-built for this problem: systems that forecast load behavior, model risk, and optimize resource allocation inside hyper-complex, high-throughput environments. Unlike monitoring dashboards or statistical anomaly detectors, Presage intervenes before failure, not after.
Investment Rationale
- Deep technical moat: Presage is doing foundational algorithmic work, not building on top of existing tooling. This creates durable defensibility that commodity infrastructure cycles cannot easily erode.
- Large addressable market: Healthcare, financial services, and cloud computing represent a multi-trillion-dollar infrastructure base. All three are early in their transition to agentic architectures.
- Right timing: The infrastructure gap Presage addresses is becoming apparent now. Getting in early means building relationships and product-market fit before the problem becomes a crisis.
The Founding Team
Presage Labs was co-founded in Paris by Benjamin Rey and Arthur Chevalier. Benjamin brings operational and business-building experience with a clear focus on scaling technical products. Arthur is the technical lead, with deep expertise in machine learning, predictive analytics, and causal simulation.
Together they combine scientific rigor with the pragmatism needed to turn research into product. That combination is rare at the seed stage, and it is the primary reason we moved quickly.
Portfolio Fit
Presage joins our AI infrastructure portfolio alongside Gensyn (decentralized compute coordination) and Pathway (post-transformer frontier model). Each company addresses a distinct layer of the same emerging stack:
- Gensyn coordinates the supply of distributed compute.
- Pathway is building AI architectures and models that autonomously and continually learn, evolve, and reason.
- Presage Labs ensures those networks remain stable under pressure.
We are backing Benjamin, Arthur, and the Presage team at the ground floor, and we are excited about what they are building.