From Anzo to Siemens: the fate of a semantic graph platform
Intelligence Center X. Not very catchy… wait, it’s Anzo? No, it’s Rapidminer. Ah, it’s Siemens?
Cambridge Semantics built Anzo, a semantic knowledge graph platform grounded in RDF, SPARQL, and in-memory MPP architecture, with AnzoGraph at its core. One of the few enterprise graph systems that took W3C standards seriously at scale, with deployments at Merck, Novartis, Bosch, and the FDA.
In 2024, Altair acquires Cambridge Semantics, folds Anzo into RapidMiner. The pitch was something like: knowledge graphs as the context layer for data science and AI pipelines, inside an existing analytical platform with a million users.
Then Siemens acquires Altair for some billions. RapidMiner Graph Studio lands inside the Siemens Xcelerator portfolio, alongside PLM, digital twin, and simulation tooling. There you go, the money tower. Sorry, the ideal graph platform 🤔
What’s telling is why Siemens wanted this. Their pitch isn’t “we bought a graph database.” It’s that a knowledge graph spanning PLM, ERP, MES, and CRM is the connective tissue for industrial AI agents. Something that can answer complex cross-domain questions across the digital thread without replicating data.
How does it compare? The semantic RDF segment is well-contested. Stardog competes directly on the enterprise knowledge graph platform story. Query-time reasoning, data virtualization, and a polished governed layer. Ontotext GraphDB is the standards-purist choice, deep OWL reasoning and linked data interoperability, popular in life sciences and public sector. Both are serious competition.
Where AnzoGraph was always architecturally distinct is raw throughput at scale: the in-memory MPP design (I think originally derived from SPARQL City’s engine) was built for analytical workloads across tens of billions of triples, not just transactional graph traversal. That’s a different performance profile than Stardog or GraphDB, which are more optimized for reasoning latency than bulk analytical load.
Neo4j occupies a different position entirely. It’s not a semantic layer; you build meaning on top of it. RapidMiner Graph Studio’s native RDF-star support, by contrast, encodes meaning in the graph itself, not in the application layer.
The Graphmart paradigm is their differentiator on the tooling side. It’s essentially CDC-aware semantic materialisation packaged for data engineers rather than ontologists.
The broader signal: industrial software stacks are no longer treating knowledge graphs as analytics accessories. They’re positioning them as the semantic substrate that makes AI agents safe, traceable, and grounded. That Siemens is making this bet is the clearest validation yet that the RDF-first, standards-compliant approach has a serious future in enterprise. Not just in research or public sector, where it already had one.
▶ Graph Studio: https://www.siemens.com/en-us/products/rapidminer/graph-studio/