The story so far:The AI Impact Summit from February 16-20 attracted lakhs of visitors, with high-profile AI executives and heads of state visiting New Delhi. On Saturday (February 21, 2026), 88 countries and international organisations signed the New Delhi Declaration on AI, which stresses that the technology must be democratised to make a difference.
Since when are AI summits being held?
Since 2023, dozens of countries have participated in annual gatherings to discuss AI. In its first instalment in Bletchley Park, the U.K. hosted a small dialogue on safety, where India was represented by then Minister of State for Electronics and Information Technology Rajeev Chandrasekhar. There is no international organisation that convenes these gatherings as of now — participating countries have generally passed on the baton for the following year’s summit. 2024’s edition was held in Seoul, and the last AI Action Summit was held in Paris in February 2025, with Prime Minister Narendra Modi co-chairing the edition with President Emmanuel Macron. In the Paris summit, the multilateral AI dialogue saw a significant shift, with U.S. Vice-President J.D. Vance vocally rejecting a safety-first approach, instead emphasising innovation and unrestrained investment into frontier research and commercialisation of the technology.
What were the India-hosted summit’s goals?
For India’s turn, the government advanced a key set of priorities: that AI’s capabilities must be available to as many people as possible; more work must be done to make it relevant in the Global South, such as by expanding representation for languages that have been under-represented in the training of western LLMs (large language models); and that the technology should be “safe and trusted”.
On the domestic front, the government sought to project India as an attractive destination for AI infrastructure and research, drum up investor sentiment for putting money into AI, and for encouraging the technology’s already-enthusiastic adoption in India, especially in fields such as healthcare, agriculture and education. The names of the working groups convened for this purpose reflect some of these priorities: human capital, inclusion for social empowerment, safe and trusted AI, resilience, innovation, and efficiency, science, democratising AI resources, and AI for economic development and social good.
What were the outcomes of the summit?
The government says that the summit attracted over five lakh visitors, a record that easily surpassed the attendance of the G20 summit in 2023. The government also touted $250 billion in investment commitments, as well as $20 billion in commitments for frontier deep tech research. The event also hosted over 500 individual discussions, with speakers from around the world. During the summit, India joined the U.S.-led Pax Silica initiative, which seeks to build a network of like-minded countries opposed to concentration of power in electronics manufacturing and critical minerals. India also achieved broad consensus among 88 countries and international organisations for the New Delhi Declaration on AI, which was signed by the U.S., China, France, and several other countries who are key for the development and deployment of AI at present.
The event saw the long-anticipated launch of India’s first domestically trained multi-billion parameter LLMs by Sarvam AI. Sarvam is a Bengaluru-based firm that enjoys millions of dollars in private equity investment as well as government support in subsidised access to computing resources under the IndiaAI Mission. The firm says its models are efficient, and beat comparable models at many benchmarks. It has announced that its models will be open source, and after the summit, launched a chatbot interface in beta to interact with its LLMs.
The event was also plagued by a series of setbacks almost every day. On the opening day — three days before the plenary with the heads of state and keynotes by figures like OpenAI CEO Sam Altman and Google CEO Sundar Pichai — organisers were overwhelmed by an unexpected number of participants. Security and logistics issues abounded, with significant traffic snarls delaying even some speakers from making it to their discussions. On the second day, an exhibitor at the summit’s concurrent expo, Galgotias University, passed off a Chinese-made robodog as their students’ creation. The wide publicity of this caused significant embarrassment to the organisers, who ordered the university to vacate the premises. And on Friday (February 20, 2026), a group of Indian Youth Congress members stormed the expo, revealing hidden T-shirts and chanting, “Mr. Modi is Compromised,” before being apprehended by the police. The Delhi Police has apprehended the protesters and is investigating the protest.
What were the investment commitments?
Reliance Industries Ltd. announced commitments of ₹10 lakh crore in domestic AI, only slightly more than the Adani Group’s similar commitment. Google gave a few fresh details about its existing $15 billion investment in data centre and AI projects in India, such as a subsea cable system that would directly connect India and the U.S. The event also saw high-profile pacts between OpenAI and the Tata Group, along with another agreement between Anthropic and Infosys. OpenAI said it would lease 100 megawatts of data centre capacity from Tata’s HyperVault, and provide its advanced models to the firm’s employees. The pacts were notable due to the stock price damage that both firms took after Anthropic’s latest coding LLM shocked investors with its advanced capabilities. Yotta Data Services, a d omestic data centre operator, announced $2 billion in data centre infrastructure build-out, with graphics processing units from Nvidia.
What does the New Delhi Declaration say?
The declaration is broadly in line with India’s stated multilateral priorities when organising the summit. Nearly all commitments are described in the statement as “voluntary” and “non-binding,” encouraging wider participation. These include a charter for the “democratic diffusion” of AI; a “Global AI Impact Commons,” which would serve as a database of use cases for countries to draw inspiration from; a “Trusted AI Commons,” described as a “repository of tools, benchmarks, and best practices [that supports the] development of secure and trustworthy AI systems,”; an “International Network of AI for Science Institutions” which would link technical institutes around the world; an “AI for Social Empowerment Platform”; an “AI Workforce Development Playbook and Reskilling Principles,” and “Guiding Principles on Resilient and Efficient AI.” “The summit is expected to catalyse long-term international partnerships and position AI as a key driver of economic growth,” the government said in a statement, adding that there was “broad-based global consensus on leveraging AI for economic growth and social good”.
Published – February 21, 2026 11:14 pm IST
