SambaNova is a leader in next-generation AI infrastructure, delivering a full-stack inference platform for customers worldwide. At the core of SambaNova's technology is the RDU (Reconfigurable Dataflow Unit) — a chip built on a dataflow architecture rather than the traditional GPU model. Its decode performance is especially strong for agentic workloads like multi-turn agents, code generation, and long-running applications. RDUs are packaged into SambaRack, rack-scale hardware that lets customers deploy state-of-the-art models with better performance, greater energy efficiency, and faster time to value.

SambaNova is a leader in next-generation AI infrastructure, delivering a full-stack inference platform for enterprises, AI labs, service providers, neo-clouds, and sovereign AI initiatives worldwide. SambaStack combines rack-scale hardware, optimized open-source models, and a serving platform to enable customers to deploy state-of-the-art models with superior performance, lower total cost of ownership, and rapid time to value. Built on SambaNova’s Reconfigurable Dataflow Unit (RDU), its dataflow architecture delivers low-latency, high-throughput inference on large models, with industry-leading decode performance that is especially impactful for agentic workloads such as multi-turn agents, code generation, and long-running applications.

Responsibilities

SambaNova is hiring a Principal Product Manager, Product Strategy and Competitive Intelligence to turn market intelligence into decisions that shape our product and business. The role translates third-party research, competitor disclosures, and benchmark data into roadmap, pricing, and positioning decisions, while building AI-powered tooling to make that intelligence faster, deeper, and more scalable. The ideal candidate can build a TCO model from primary inputs, connect architecture specifications to financial performance, and go hands-on with AI tools to automate and continuously improve their own research pipeline. This individual will be responsible for:

Market intelligence to product inputs

Own the pipeline from third-party research (SemiAnalysis TCO, Tokenomics, Accelerator, and Datacenter models; competitor disclosures; benchmark data) to decision-ready analysis.
Maintain the house model of where RDU inference wins on cost per million tokens, at which latency tiers, for which workloads, and against which competitor configurations.
Convert TCO, tokenomics, and data center data into roadmap language: feature priorities, pricing moves, and segment bets.
Open each product review with a summary of market changes since the last quarter and their implications for SambaNova.

Competitor specifications and roadmaps

Track chip-level specifications for NVIDIA, AMD, Google TPU, AWS Trainium, Cerebras, Groq, and emerging accelerators: compute, memory hierarchy (SRAM, HBM, DDR), interconnect, power, process node, and packaging.
Track rack- and cluster-level design: scale-up and scale-out topology, networking (NVLink-class fabrics, Ethernet and InfiniBand, optics), storage, cooling, and rack power density.
Track roadmaps: announced parts, credible leaks, and supply signals such as HBM allocation, CoWoS capacity, and foundry node ramps.
When a spec or roadmap change moves the TCO or latency frontier, quantify the effect on SambaNova's positioning within a week.
Maintain a versioned spec comparison matrix with a changelog.

AI tooling for intelligence

Build agentic pipelines that monitor research releases, competitor announcements, InferenceMAX and ClusterMAX updates, conference disclosures (Hot Chips, GTC, OCP), and regulatory and supply-chain filings.
Extract specs and claims into the comparison matrix automatically, with human review on anything that feeds a decision.
Automate the first draft of the weekly brief from monitored sources.
Treat the tooling as a product and iterate on precision, coverage, and detection latency.

Segment and demand theses

Use data center capacity and power data to pressure-test the SambaManaged pipeline: which operators have stranded power and cooling suited to air-cooled RDU racks, where, and on what timeline.