Bengaluru: A Bengaluru-based software developer has gone viral on social media after claims that he built a successful artificial intelligence (AI) business using decommissioned banking hardware purchased for just ₹3.5 lakh at a government auction.

The viral post, shared by 100x Engineers, alleges that the 27-year-old developer, identified only as Raj, now earns nearly ₹27 lakh a month by fine-tuning AI models for overseas clients. However, the claims have not been independently verified.

From discarded servers to AI infrastructure

According to the post, Raj purchased a decommissioned hardware lot from a State Bank IT auction, where government institutions and enterprises periodically dispose of unused equipment.

The lot reportedly included enterprise-grade storage systems, networking hardware and around 75 terabytes of enterprise storage drives, equipment said to have originally cost more than ₹1.5 crore.

Instead of reselling the hardware, Raj is said to have converted it into an AI computing cluster inside his one-bedroom apartment.

AI models for global clients

The post claims Raj expanded the setup by adding two custom-built computers powered by RTX 3090 graphics cards and 128 GB RAM each.

Using this infrastructure, he reportedly fine-tunes open-source AI models, including Llama 3.3 70B and Qwen, for clients, particularly software companies in the United States through platforms such as Hugging Face.

The post further claims the system enables him to avoid expensive cloud computing costs, with a monthly electricity bill of around ₹8,000.

Claims remain unverified

The social media post concludes that India already possesses significant computing infrastructure through surplus enterprise hardware, arguing that innovation lies in repurposing such equipment rather than investing heavily in new systems.

However, there has been no independent verification of the developer’s identity, the reported earnings, the hardware specifications or the business model described in the viral post.

Despite this, the story has sparked widespread discussion among technology enthusiasts, entrepreneurs and AI developers about the potential of refurbished enterprise hardware for building cost-effective AI infrastructure