The India AI Impact Summit, held in New Delhi from February 16 to 20, 2026, concluded with the adoption of the New Delhi Declaration on AI Impact, endorsed by 88 countries and international organisations.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source The declaration is framed around development and shared benefit rather than binding controls, organized under seven pillars including democratizing AI resources and ensuring secure and trusted systems.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source Against that consensus, a US official, Michael Kratsios, stated plainly that Washington rejects global governance of AI, and China was effectively absent from the gathering.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source The stake: the summit made visible a fault line between a broad, development-oriented majority and a US posture that resists multilateral oversight. We assess with moderate confidence that this split, not the declaration's text, is the durable outcome, and that AI governance is fragmenting into parallel tracks rather than converging on one framework.
What the summit produced
The headline number is real breadth. Eighty-eight countries and international organisations endorsed the New Delhi Declaration, a wide coalition assembled around a document deliberately pitched at cooperation rather than constraint.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source Its seven pillars run to access and capability, democratizing AI resources, promoting economic growth and social good, expanding AI use in science, and developing human capital, alongside secure and trusted systems.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source This is a development agenda, and its breadth is a function of that framing: a text about sharing AI's benefits is easier to sign than one that imposes obligations.
The two most consequential facts are about who did not fully join. The United States, present through its delegation, used the stage to reject global governance of AI outright.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source China, described as effectively absent, did not anchor the gathering, its non-participation coinciding with the Lunar New Year.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source We assess with high confidence that a declaration the two largest AI powers do not stand behind as a governance commitment is a statement of aspiration among middle and developing powers more than a binding global settlement.
The framing choice deserves emphasis, because it explains the breadth. The declaration is titled around AI impact, not AI safety or AI governance, and its pillars foreground access, capability, and shared benefit.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source That is a deliberate pivot away from the risk-centered language of earlier international summits toward a development agenda that developing economies can endorse without accepting constraints on their own AI ambitions. We assess with moderate confidence that the pivot is what made 88 signatures possible: a text about distributing AI's benefits asks little of any signatory, whereas a text imposing binding safety obligations would have fractured the same coalition. The cost of that breadth is depth, since a document everyone can sign commits no one to much.
Second order effects and the ledger
Who gains. India gains standing as convener of the largest such gathering and as a broker for a development-focused bloc, a position that lets it shape the agenda without picking a side between Washington and Beijing.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source Middle and developing powers gain a shared text that legitimizes their claim to AI access and capacity building.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source The frontier labs gain commercial ground: the CEOs of OpenAI, Anthropic, and Google DeepMind used the summit to announce partnerships with Indian companies, converting a governance forum into a distribution event.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source
Who loses, or is exposed. The multilateral-governance project loses momentum, because a US refusal to accept global oversight signals that the most capable AI ecosystem will not be bound by a shared framework, which weakens the leverage of any such framework over the frontier.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source Middle powers named as anxious about technological dependence on the US, including European states, Canada, and India, are exposed by that same refusal: a consensus text does not reduce their reliance on US-controlled models and compute.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source We assess with moderate confidence that the summit deepened rather than closed the gap between where AI capability sits and where governance ambition sits.
The counter-case
The argument that the split matters less than it looks is that development-focused consensus is where governance realistically begins. Eighty-eight endorsements build norms and coalitions that can harden into commitments over time, and the US rejection was of global governance specifically, not of every form of cooperation.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source A reading that treats the declaration as merely aspirational may undervalue the slow accretion of shared language that earlier summit series produced. The counter to that is capability concentration: norms without the participation of the two states that build and control the leading systems have limited purchase on those systems' behavior, and China's absence plus the explicit US refusal leave the declaration without its two most important would-be adherents.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source For the fragmentation thesis to fail, the US would have to soften its posture or the coalition would have to gain leverage independent of the frontier labs, and neither is in evidence yet.
What to watch
- The US posture at the next summit in the series. If Washington maintains or hardens its rejection of global governance at the following gathering, the fragmentation read holds; a shift toward engagement would signal the split is negotiable.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source
- Whether the declaration produces concrete instruments. Watch over the next year whether the named initiatives around democratizing AI resources move from text to funded programs, the test of whether 88 signatures carry weight.2 Organiser 2026-02-21 The New Delhi Declaration on AI Impact, adopted at the February 16 to 20, 2026 summit, was endorsed by 88 countries and international organisations around seven pillars including democratizing AI resources, promoting economic growth and social good, and ensuring secure and trusted AI systems. Open source
- China's next move on AI diplomacy. If Beijing convenes or joins a competing forum, parallel governance tracks are forming; sustained absence would leave the field to the Delhi coalition and the US refusal.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source
- Middle-power dependence shifts, or does not. Track whether states worried about reliance on US models pursue sovereign compute or alternative suppliers, the concrete signal of whether consensus text changed anything about the underlying dependence.1 TIME 2026-02-19 At the India AI Impact Summit, White House official Michael Kratsios said the US totally rejects global governance of AI; China was effectively absent coinciding with Lunar New Year; and CEOs of OpenAI, Anthropic and Google DeepMind announced partnerships with Indian companies, amid middle-power anxiety over dependence on the US. Open source