Svashasan Logo
Labs Newsroom / Press Release
Research & Training

Svashasan Labs unveils 'Param-1' supercomputing cluster for end-to-end training

February 7, 2026

Param-1 Data Center

Inside the Param-1 facility: Liquid-cooled racks housing over 10,000 custom AI accelerators.

BENGALURU, India — Svashasan Labs today announced the activation of 'Param-1,' one of the world's most powerful custom supercomputing clusters dedicated solely to the development of autonomous vehicle neural networks.

As Svashasan moves toward widespread deployment of its Supervised Pilot fleet, the volume of real-world driving data has grown exponentially. Param-1 was engineered from the ground up to ingest, process, and train models on these billions of frames of anonymized video data with unprecedented efficiency.

Custom Silicon at Scale

At the heart of Param-1 are 10,240 proprietary AI accelerators, designed by Svashasan's hardware engineering team. These chips are optimized specifically for 4D-Transformer architectures, allowing the cluster to achieve a training speed that is 2.4 times faster than industry-standard GPU clusters.

"Param-1 represents a quantum leap in our ability to process real-world driving data. By removing the compute bottleneck, our engineers can now test and iterate on new safety features in days rather than months." — Dr. Anjali Sharma, Head of AI Research

Energy-Efficient Intelligence

Innovation at Svashasan Labs extends beyond performance to sustainability. Param-1 is housed in a facility that leverages a proprietary immersion cooling system, reducing cooling-related energy consumption by 40% compared to traditional air-cooled data centers.

The facility is powered 100% by renewable energy sources, aligning with Svashasan’s mission to not only revolutionize transportation but to do so with the smallest possible environmental footprint.

Hardware closeup

Proprietary logic boards designed for high-bandwidth data movement between compute nodes.

Accelerating the Future

The first models trained on Param-1 are already undergoing internal validation. These models show a significant reduction in disengagement rates in complex urban scenarios and improved predictive capabilities in adverse weather conditions.

This infrastructure investment ensures that Svashasan remains at the cutting edge of AI, providing our customers with a driving experience that gets smarter every single day.

Media Inquiries: press@svashasan.com

Related News