ON THE second day of the India AI Impact Summit 2026, held with considerable fanfare at Bharat Mandapam in New Delhi, a professor from Galgotias University introduced the world to "Orion." The quadruped robotic dog, she told DD News, had been developed by the university's Centre of Excellence. Within hours, internet users had identified Orion as the Unitree Go2, a commercially available product manufactured by China's Unitree Robotics, sold on the open market for approximately $1,600. By the next morning, the university had been ordered to vacate its exhibition stall.

The Galgotias incident is not merely a story of individual dishonesty. The professor was thrust before the cameras to represent an institution that had received a ₹42 lakh research grant from the Technology Innovation Hub on IoT at IIT Bombay just weeks earlier, under the National Mission on Interdisciplinary Cyber Physical Systems implemented by the Department of Science and Technology. Galgotias is an almost perfect symbol of India's wider technology strategy: a large PR apparatus, thin technical substance, and political proximity standing in for genuine productive capacity.

This is not the first time Galgotias has occupied the centre of national controversy. In 2020, a researcher at the institution published a paper investigating whether the vibrations produced by thalis and ghantis could cure COVID19, a specimen of the Hindutva pseudoscience that has flourished under BJP patronage. That a university with this track record could receive DST grants and secure a pavilion at a summit attended by the heads of state and representatives of nearly twenty countries speaks plainly to how the Modi government operates in the technology space: rewarding proximity to power over competence. When IT Minister Ashwini Vaishnaw personally shared the video of Orion on social media before the backlash erupted, he was not acting carelessly. He was following a well-established script in which the appearance of technological prowess substitutes for its reality.

The Galgotias incident was an opening act. The summit itself told a more consequential story.

First, there is the question of what kind of investment the summit actually attracted. The headlines were staggering. Reliance Industries pledged ₹10 lakh crore, approximately $110 billion, over seven years to build AI-ready data centres. Adani Group committed $100 billion over a decade to build renewable-energy-powered hyperscale facilities. Microsoft affirmed it was on track to invest $50 billion across the Global South by 2030. At first glance, this looks like a genuine inflection point.

Look closer, and the shape of these investments tells a different story. The Ambani and Adani pledges are, at their core, real estate and energy plays. Multi-gigawatt data centres are energy-intensive civil infrastructure projects that require land, power, and cooling - all domains where these conglomerates already hold dominant positions, and where Modi government policy has delivered consistent regulatory and tax advantages. The Union Budget contains explicit tax giveaways for data centre construction. The India-US trade framework under negotiation includes provisions to deepen American AI firms' position in Indian infrastructure. What is being built is not Indian AI. It is the Indian landing pad for American AI. The compute will be chips purchased from American companies. The models running on it will also be American.

Second, there is the structural exposure of India's software sector. India's IT industry employs approximately 5.7 million people and contributes over 7.5% of GDP. It was built on one comparative advantage: the large-scale, low-cost supply of English-speaking technical labour to serve Western outsourcing needs. That advantage is now under systematic assault. TCS cut over 12,000 jobs in 2025 - the largest redundancies in its history. Infosys reduced headcount by an estimated 20,000 between fiscal year 2023 and 2025. Nasscom has acknowledged that "workforce rationalization" will continue as AI automates the routine tasks - application maintenance, software testing, basic coding - that have historically been the bread and butter of Indian IT.

The stock market delivered its own verdict. In the week immediately preceding the summit, the Nifty IT index shed 9.4%, wiping approximately ₹4 lakh crore in market capitalisation. The trigger was the release of "agentic" AI tools from Anthropic. These tools are designed to automate high-volume back-office functions, the very services Indian IT companies provide. The irony of Indian IT stocks collapsing days before a summit designed to announce India's AI ascendance was largely set aside in favour of headline investment numbers.

Third, and most critically, where are India's own foundational models? The country with the world's second-largest internet user base and 100 million weekly ChatGPT users has produced no frontier AI model of its own. There is no Indian equivalent of DeepSeek, Qwen, or Kimi. The IndiaAI Mission, launched in 2024 with a budget of ₹10,372 crore, is oriented primarily toward subsidised compute access and dataset curation, not toward funding sovereign model development. Sarvam AI, arguably India's most credible domestic AI company, unveiled new multilingual models at the summit. It is promising work. It is also vastly outgunned.

The summit's deepest admission of dependency came from TCS and Infosys, which announced partnerships with OpenAI and Anthropic, respectively, to help their clients adopt American AI platforms. India's two largest IT firms are becoming implementation partners for the very systems displacing their workers. Services revenue from AI integration consulting will not offset the structural loss from automation of the software delivery work that currently sustains those firms.

Meanwhile, China has taken a fundamentally different approach. China poured over $125 billion into AI-related infrastructure in 2025 alone, while simultaneously developing frontier model capabilities through Baidu, Alibaba's Qwen series, and the remarkable DeepSeek, which produced internationally competitive large language models at a fraction of the assumed compute cost, while operating under severe US export controls on advanced chips. Unitree Robotics, the Chinese firm whose robot became India's most-discussed AI product of the week, has moved from imitation to genuine innovation through state-supported industrial incubation. India's summit featured a Chinese robot and proclaimed it a demonstration of Indian innovation. China's ecosystem is building the robots the rest of the world is rebranding. The difference is not talent. It is state direction, long-term capital, and industrial discipline.

The historical parallel runs through the 1990s software boom. India's first technology revolution was also built on foreign dependency, processing the back-office work of American corporations rather than building the applications that those corporations sold. The model delivered growth and created a middle class, but it did not build sovereign technological capacity. AI is now threatening to automate precisely the layer of work that this model relied upon. Rather than using this disruption to build something genuinely innovative, the Modi government is reproducing the 1990s dynamic at a higher level of abstraction: Indian land and energy, American intelligence.

The billions pledged at the summit are not trivial. Data centre infrastructure, if built at scale, will give Indian developers faster and cheaper access to compute - a genuine need. But none of this adds up to the leap from consumer to builder that technological sovereignty requires. That leap demands public investment in foundational research, universities that produce AI engineers and PhDs rather than repackage Chinese robots, and an industrial policy willing to protect domestic AI companies against the predatory pricing of American platforms. None of that was on the agenda at Bharat Mandapam.

India brims with tech talent. What it lacks are the companies, institutions, and public investment architecture that would allow that talent to build things the world needs, rather than remain trapped in low-value service roles while others capture the surplus. The robots at the AI summit were Chinese. The models powering the data centres being announced will be American. And the workers who built India's IT industry are being told to reskill for a future being designed elsewhere. That is not an AI strategy. That is a continuation of dependency by more expensive means, dressed up as national ambition.