The Rise of the Alembic Supercomputer: Melted GPUs and Causal A.I. Convergence

Alembic’s AI systems focus on identifying cause-and-effect relationships rather than just correlations, which requires a high level of computing power and data processing capabilities. The company’s investment in the Nvidia NVL72 superPOD, one of the fastest privately owned supercomputers ever built, will enable it to power its enterprise-grade causal AI models.

The recent funding round of $145 million in Series B, led by Prysm Capital and Accenture, reflects the growing importance of proprietary data and causal reasoning in the field of artificial intelligence. Alembic’s strategic direction aims to leverage private corporate data to provide answers to questions that generic models cannot.

The company’s transformation from a marketing technology vendor to a provider of enterprise-wide AI solutions was driven by a breakthrough in its technology after the Series A funding round. Alembic discovered that its causal models could work across various business domains with time-series data, leading to a fundamental shift in its market positioning.

Causal AI, as opposed to correlation-based analytics, provides executives with more accurate and actionable insights for decision-making. Alembic’s platform has been adopted by several Fortune 500 companies, including Delta Air Lines and Mars, to gain a unified view of their marketing effectiveness, operational efficiency, and strategic investments.

The decision to invest in a liquid-cooled supercomputer like the Nvidia NVL72 superPOD reflects Alembic’s commitment to meeting the technical demands of its AI models and the data sensitivity requirements of its enterprise clients. By owning its computing infrastructure, the company can ensure the security and efficiency of its operations, setting it apart from others in the industry. Alembic’s system differs from large language models in that it uses “online and evolving” models built on spiking neural networks, brain-inspired architectures that continuously learn as new data arrives. This continuous learning occurs at a massive scale, with Alembic’s system automatically permutating through billions of possible combinations of how data could be analyzed when new data is brought in by customers. To support this level of computation, Alembic writes custom CUDA code and low-level GPU kernels optimized specifically for causal inference workloads, which are not possible on standard cloud configurations.

The move to liquid-cooled systems has addressed the problem of melting down GPUs due to pushing them beyond their thermal limits, and has also enabled Alembic to run workloads more cost-effectively compared to cloud platforms. This supercomputer strategy also ensures data sovereignty for customers, particularly in regulated industries, who have contractual prohibitions against using cloud platforms like Amazon Web Services, Microsoft Azure, or Google Cloud.

Alembic’s collaboration with Nvidia, its founding enterprise customer, supercomputing partner, and technical collaborator, has been instrumental in supporting the startup’s technical ambitions. Nvidia has provided Alembic with the necessary computing resources, including arranging for Equinix to provide a private cage with sufficient power capacity and sourcing Alembic’s first H100 GPU cluster. The partnership has deepened over time, with Alembic utilizing Nvidia’s AI Enterprise software suite and benefiting from Nvidia’s support in obtaining next-generation liquid-cooled systems.

Alembic’s customer roster has grown rapidly, with enterprises seeking ways to measure AI and marketing investments that traditional analytics cannot capture. Alembic’s platform has been used by companies like Delta Air Lines, Mars, Fortune 500 technology and financial services firms, and Texas A&M University’s athletics program to measure various aspects of their businesses. The ability to measure previously unmeasurable activities has transformed decision-making for these companies, allowing them to connect marketing efforts to tangible outcomes and make data-driven decisions. Alembic’s revolutionary Causal AI technology represents a significant breakthrough in the field, enabling us to move beyond mere correlation and delve into the intricate details of how organic conversations translate into direct impacts on sales. The platform boasts the capability to forecast revenue, close rates, and customer acquisition trends up to two years in advance with an impressive 95% confidence level, as per Puig’s statements.

Puig highlights the unique advantages that set Alembic apart from competitors such as Nielsen, Google, Meta, and other emerging AI analytics startups. The company’s causal models are built on proprietary mathematics developed over years, safeguarded by patents, making replication a monumental task. Additionally, the substantial computing requirements pose a significant barrier, with Alembic operating at a foundational model level of compute far beyond typical cloud services like AWS Sagemaker. Furthermore, the focus on data sovereignty requirements for enterprise clients opens up opportunities for neutral third parties, a niche that hyperscale cloud providers struggle to address. Lastly, Alembic’s expertise in handling messy and fragmented data, honed over years of engineering efforts, sets it apart in the industry.

Alembic’s strategic focus on private data marks a departure from the trend of building larger language models dominating the AI landscape. While other players compete on chatbot capabilities or trivia knowledge, Alembic’s infrastructure caters to a different form of intelligence—one that comprehends cause-and-effect relationships within the intricate, proprietary data of individual companies.

The company’s journey from a bootstrap startup to operating a top-tier private supercomputer mirrors the evolution of enterprise AI as a whole. With a vision to become the central nervous system of enterprises, connecting cause and effect across various business functions, Alembic’s expansion into robotics and GPU-accelerated database products showcases its versatility and ambition.

As Alembic navigates complex enterprise landscapes with long sales cycles and integration challenges, its $145 million funding round and proven track record with notable clients underscore its unique position in the market. While competitors may pose threats, Alembic’s specialized approach to understanding cause-and-effect relationships in data sets it apart in a world where general-purpose AI solutions abound.

In essence, Alembic’s emphasis on specialized systems that uncover hidden cause-and-effect relationships within data sets positions it as a pioneer in the realm of enterprise AI. With a focus on private intelligence engines over universal chatbots, Alembic stands out as a trailblazer in leveraging data insights for sustainable competitive advantages. Transform the following:

“Tomorrow is a new day with no mistakes in it yet.”

into:

“Tomorrow holds endless possibilities with no regrets attached.”