AI4VIETNAM
Enterprise

FPT Launches AI EOS, GPU Kubernetes and Token Factory at TRANSFORM 2026

October 19, 2026 · 6 min read · Enterprise

FPT Launches AI EOS, GPU Kubernetes and Token Factory at TRANSFORM 2026

FPT Corporation used its FPT TRANSFORM 2026 event in Hanoi on October 17 to launch three products aimed at helping Vietnamese enterprises move AI from pilots into live operations: FPT AI EOS, FPT Kubernetes Engine with GPU, and FPT Token Factory, the company said in a press release.

Held at InterContinental Landmark 72 under the theme “Ship AI Live,” the programme featured more than 50 sessions, ten technical workshops and more than 200 solutions on show, with organisers saying more than 1,700 tickets had sold before the day. FPT Smart Cloud chief executive Le Hong Viet said the additions let each organisation choose how much AI autonomy it wants, from applications and models to infrastructure, so technology and data stay more under local control.

FPT AI EOS is pitched as an orchestration layer for AI agents that can plug into customer-support, contract and sales systems, with digital workers preparing quotes and contracts while humans still approve risky terms and signatures. FPT Kubernetes Engine with GPU adds a managed Kubernetes layer for training and serving models inside FPT’s AI Factory stack; in company test conditions, clusters spun up in about five minutes. FPT Token Factory offers a single gateway to FPT-run and third-party models for inference, with data staying on FPT infrastructure when FPT hosts the model.

What it means in practice

FPT Launches AI EOS, GPU Kubernetes and Token Factory at TRANSFORM 2026 — contextual photo

The launch extends the company’s shift described in FPT’s move toward outcome-based AI and its Nvidia Nemotron Personas Vietnam expansion, and fits Vietnam’s updated 2030 ASEAN AI-hub strategy. For CIOs, the practical choice is whether to buy orchestration, GPU ops and model access as a bundle or keep mixing foreign clouds.

FPT’s own test numbers are not a service-level promise, and Token Factory still depends on third-party policies when traffic leaves FPT-hosted models. The harder proof will be multi-month deployments that cut ticket times or contract cycles without creating shadow agents that bypass compliance.

What to watch: which of the three products Vietnamese banks and manufacturers buy first, whether GPU cluster spin-up times hold outside lab conditions, and how Token Factory prices compare with direct hyperscaler APIs.

← Back to AI4Vietnam