Bharat1.ai inaugurated Ba AI Superpark on April 19 in Bengaluru as the first phase of an AI-native city project designed to house founders, researchers, and AI practitioners in a purpose-built 500,000 square-foot facility. The project is led by Bharat1.ai co-founder and CEO Umakant Soni, alongside co-founders Subhashis Banerjee (AI Foundry CEO) and Sireesh Kupendra Reddy (real estate developer). Unlike traditional smart cities layered onto existing infrastructure, this facility is being designed and built from scratch, with AI, humans, and humanoid robots intended to operate together as a single system.
Facility and Ecosystem
Ba AI Superpark currently houses approximately 25 startups and is equipped with 400 gbps connectivity to four AI clouds—Neysa, E2E, AWS, and Google Cloud’s GCP—enabling users to train, fine-tune, and test large language models. Bharat1 has partnered with universities including IIT Kanpur, IIT Ropar, IIT Bombay, and BITS Pilani.
Umakant Soni told Tech in Asia that the facility is not simply a co-working space but aims to aggregate “the entire ecosystem needed to build an AI economy.” The project targets eventually housing more than 10,000 founders and AI researchers. Soni aims to bring together over 100 startups working across the AI stack, from energy systems (solar and nuclear microreactors) to robotics, sensors, and autonomous systems.
Robotics startups at the facility include one building cargo drones with a 1,000-kilometer range, another that has developed a cooking robot capable of preparing everyday dishes, and a third creating robotic quadrupeds for multiple use cases. Bharat1 is also incubating a nuclear sector startup in stealth mode.
Technology Infrastructure and Partnerships
Nvidia is a “core technology collaborator” in enabling the city’s infrastructure, according to Soni, providing access to advanced GPU systems and software frameworks through cloud partnerships with Bharat1. The US chipmaker has previously collaborated with Artpark, the robotics incubator formerly led by Soni.
Energy infrastructure is a critical component of the project. The city plans to rely on a hybrid model combining solar energy with nuclear microreactor ecosystems. Bharat1 is already in talks with a nuclear startup that could supply a small modular reactor to meet the city’s energy needs.
Funding and Scale
The estimated investment for the full AI city project stands at approximately $650 million over multiple phases. Bharat1’s co-founders have already “invested millions of dollars” in the project, though they did not provide specifics. Bharat1 plans to launch a $100 million fund in the second half of 2026 to support startups within the ecosystem.
By 2030, the city aims to host 25,000 researchers, startups, and AI practitioners, alongside thousands of humanoid robots and AI agents. Bharat1 is exploring land parcels in and around Bengaluru for expansion.
Governance and Accountability Framework
Soni proposes linking each AI agent to a real-world human identity, potentially through India’s Aadhaar system, to establish accountability for AI systems. Under this model, if an AI agent acts maliciously—whether through fraud, deepfakes, or misinformation—responsibility would trace back to a human operator. Access to AI agents would be a privilege tied to responsible use, with misuse potentially resulting in loss of access, similar to professional license revocation.
The Bharat1 team has released a whitepaper about Aadhaar linkage and plans to test it in phases within the AI city to study its impact on accountability before broader implementation.
Soni describes current AI systems as “jagged intelligence,” capable of narrow domains but unreliable outside specialties, with hallucinations remaining a common problem. He views the AI city as a controlled testing ground where AI systems can be observed in real-world conditions before large-scale deployment, generating data on traffic, energy, robots, and human behavior to improve future AI systems.
The project also includes personal AI agents managing residents’ schedules, errands, and daily decisions. At the core is what Soni calls “cognitive infrastructure,” where every building, system, and individual will potentially have a digital twin feeding into a central AI “brain” managing traffic, utilities, energy, and operations.
Replication and Global Expansion
Soni sees the Bengaluru project as a blueprint applicable elsewhere, both within India and globally. He notes that India is expected to need 25 to 30 new cities over the next two decades as the country shifts away from agriculture, and argues that building them using traditional models would replicate existing inefficiencies.
Once the AI city’s systems—digital twins, coordination models, and governance frameworks—have been tested and refined, they could be replicated for other projects, either retrofitting existing cities with AI-driven infrastructure or building new greenfield AI cities from scratch. If the Aadhaar-linked agent model proves effective, it could become part of a broader framework that other countries adopt for AI accountability, though success would depend on a market’s identity infrastructure and regulatory environment.
Bharat1 is being selective about partners, with Soni expecting Nvidia to play an important role in helping take the model to regions with challenges similar to India, such as Southeast Asia, Africa, and West Asia. Bharat1 has already received interest from potential partners in West Asia, though further details were not shared.
Regarding funding sources, Soni stated that Bharat1 is not seeking any government funding for the project. However, venture capitalist Ashish Taneja, founding partner and CEO of venture fund GrowX, cautioned that India still lacks sufficient risk capital for such ambitious bets and noted that the project would likely benefit from coordinated backing from central and state governments.
Sharad Sharma, co-founder of nonprofit think tank iSpirt Foundation, sees merit in the AI city concept, noting that India is “rich in problems” and that AI can help the country become better problem solvers. Sharma added that Bharat1’s project goes beyond traditional incubation models and builds on Soni’s track record.
Soni’s background includes founding Pi Ventures, one of India’s early deeptech venture firms, co-founding venture studio AI Foundry, and serving as former CEO of robotics incubation center Artpark, with backing of several AI and robotics startups.
