Netweb Technologies announced the launch of its new ‘Make in India’ AI Supercomputing solutions, the Tyrone Camarero Spark and the Tyrone Camarero GB200 systems. These are powered by the latest NVIDIA Grace Blackwell platforms, signaling a major boost for sovereign AI development. The Spark system is highlighted as one of the world’s smallest AI supercomputers, delivering 1 Peta Flop (FP4) performance in a compact, low-power desktop form factor, while the GB200 targets trillion-parameter models.
Netweb Unveils New ‘Make in India’ AI Supercomputing Portfolio
Netweb Technologies India Limited has announced the launch of advanced AI infrastructure solutions designed and manufactured in India, supporting the nation’s Sovereign AI Development mandate. The press release, dated February 18, 2026, details two primary product lines: the Tyrone Camarero Spark and the Tyrone Camarero GB200 systems, both leveraging the cutting-edge NVIDIA Grace Blackwell architecture.
The Tyrone Camarero Spark: Compact Powerhouse
The Tyrone Camarero Spark is introduced as one of the world’s smallest AI supercomputers, packaged in a compact desktop form factor. Key specifications and features include:
- Delivers a petaflop of AI performance (FP4) and 128GB of unified memory.
- Operates with a low system power supply of just 240 Watts.
- Capable of running inference on models up to 200 billion parameters and fine-tuning models up to 70 billion parameters locally.
- Combines NVIDIA Blackwell GPUs, NVIDIA Grace CPUs, NVIDIA Networking, and the NVIDIA CUDA-X libraries.
Sanjay Lodha, Managing Director & CEO of Netweb, stated that the Spark system revolutionizes AI compute in India by offering 1 Peta Flop (FP4) performance, compared to a previous generation system requiring 3200 Watts for a lower 170 Tera Flops (FP16) performance.
Tyrone Camarero GB200 for Trillion Parameter Models
Netweb is also scaling its offerings with the Tyrone Supercomputing Systems designed for trillion parameter models, built on the NVIDIA GB200 NVL4 platform.
- This architecture delivers up to 2x performance for scientific computing, AI model training, and inference over prior generations.
- The GB200 system utilizes four NVIDIA Blackwell GPUs connected via NVLink bridge and two NVIDIA Grace CPUs connected via NVLink-C2C.
- It is designed for the most compute-intensive, rack-scale workloads, integrating 20 Tyrone GB200 MGX systems per liquid-cooled rack, totaling 80 NVIDIA Blackwell GPUs and 40 NVIDIA Grace CPUs.
Vishal Dhupar, Managing Director, South Asia, NVIDIA, emphasized that these systems deliver powerful, efficient infrastructure necessary to accelerate innovation for generative AI and sovereign compute demands.
Showcase Event
Netweb confirmed that a complete range of the Tyrone AI product line, reflecting its ‘Make in India’ strength from edge to data center computing, will be displayed at the India AI Impact Summit in New Delhi between 16th-20th February 2026.
Source: BSE
