Alembic raises $145M for causal AI and supercomputer

2 hours ago7 min read

In a significant departure from the conventional AI funding narrative, Alembic Technologies has secured a formidable $145 million Series B and growth funding round, achieving a valuation approximately thirteen times greater than its previous round—a meteoric rise from roughly $50 million to $645 million. This substantial capital infusion, spearheaded by Prysm Capital and Accenture with participation from Silver Lake Waterman and others, signals a pivotal strategic shift within the enterprise AI sector.While the industry at large remains captivated by the scaling of large language models, Alembic is placing a contrarian bet on causal AI, a fundamentally different paradigm that moves beyond identifying statistical correlations to definitively establishing cause-and-effect relationships. This approach, as articulated by founder and CEO Tomás Puig, posits that the true, defensible competitive advantage in artificial intelligence will not stem from marginally better chatbot performance, but from proprietary data and the specialized systems capable of interpreting it to answer unique, high-stakes business questions.The company's journey to this point is a testament to iterative discovery; initially a resource-constrained startup focused on marketing analytics and running simulations on an 'army of Mac Pros,' its post-Series A experimentation revealed its causal model possessed unexpected generality, functioning as a foundational model applicable across virtually any business domain with time-series data, from supply chain logistics to financial forecasting. This technological breakthrough is now being supercharged by a massive infrastructure investment: the deployment of a proprietary Nvidia NVL72 superPOD, a liquid-cooled supercomputer that Alembic claims is one of the fastest privately owned systems globally.The rationale for such a capital-intensive move is twofold. Technically, the company's spiking neural network architectures require 'online and evolving' computation, continuously permutating through billions of data combinations to isolate causal signals—a process so demanding it famously melted standard GPUs.Economically, owning this 'F1 car' infrastructure is drastically more cost-effective than relying on cloud providers, with Puig noting that an equivalent AWS setup would cost an estimated $62 million annually. Perhaps more critically, this strategy addresses the stringent data sovereignty requirements of its Fortune 500 clientele, which includes Delta Air Lines and Mars, as many enterprises in regulated industries like finance and consumer packaged goods have contractual or legal prohibitions against using hyperscale clouds.The company's relationship with Nvidia, which began after CEO Jensen Huang read about its Series A and initiated contact, has been instrumental, evolving from a technical partnership into a foundational enterprise customer relationship. The practical applications are demonstrating transformative potential; Delta quantified the revenue impact of its Olympics sponsorship within days, while Mars can now attribute sales lifts to viral social moments and specific product changes, moving from directional hunches to precise, causal understanding.This positions Alembic not merely as another analytics vendor, but as a potential central nervous system for the enterprise, a layer of intelligence that understands not just what is happening, but why. In an AI landscape where public model capabilities are rapidly converging, Alembic's bet is that the future belongs not to the most powerful general-purpose intelligence, but to the most insightful private one.

#causal AI

#enterprise analytics

#supercomputing

#proprietary data

#Nvidia Blackwell

#featured

Stay Informed. Act Smarter.

Get weekly highlights, major headlines, and expert insights — then put your knowledge to work in our live prediction markets.