Experience RENEGADE Summit 2026

At RENEGADE Summit 2026, Furiosa emphasized that the surging token demand created by agentic AI exposes the limits of existing compute architectures. RNGD was presented as the AI engine for the agentic era, and the company announced that the entire RNGD product line has already reached deployment-ready status.

• June Paik explained that AI data centers are shifting from training-centric to inference-centric operations. The key criterion was not performance itself, but true performance per energy and TCO.
• LG AI Research, Samsung SDS, LG U+, Upstage, and MegazoneCloud took the stage together to share their direction on sovereign AI infrastructure, cloud-scale inference services, and production AI systems.
• Hanjoon Kim introduced the Tensor Contraction Processor (TCP) architecture, explaining that by making tensor contraction the core abstraction between HW/SW, it increases data reuse, reduces memory movement, and enables compiler-driven global optimization.
• Jeehoon Kang unveiled the compiler stack, Tensor Contraction Language (TCL), and Furiosa Virtual ISA. TCP and the SDK were co-designed from the ground up to apply new models and optimization techniques more flexibly than the existing CUDA-centric GPU stack.

At the close of the event, it was announced that RNGD has officially transitioned to the mass production stage. The message was clear: Furiosa is elevating inference chips for data centers into more efficient and sustainable infrastructure, turning it into something deployable now rather than a future roadmap.