On this page · 9 sections
Summary. The next global AI summit after New Delhi has fixed dates: 21 and 22 June 2027 at Palexpo in Geneva, under the theme Bridging Innovation and Trust. It is the fifth in the series that began at Bletchley Park in 2023. On 12 August 2026 the Federal Council appointed former CERN Director-General Fabiola Gianotti as its Delegate, with overall coordination. ITU's AI for Good Global Summit follows at the same venue on 23 and 24 June. Delhi ended with a voluntary declaration endorsed by 92 countries and international organisations as of 5 March 2026, including the USA, China, the EU, the UK and Russia. Thirteen model developers also signed commitments, and their one dated promise is to publish usage statistics by the next summit. For Indian firms, the next steps are practical: a Call for Inputs and participation rules by the end of 2026, a side-event call in early 2027, MeitY's AI consultation paper within weeks, and most DPDP Rules on or about 13 May 2027, 39 days before the summit.
About this article. For founders, compliance leads and policy teams at Indian AI and IT services companies. On 1 November 2026 we checked summit details against Swiss federal pages and the official site, Delhi outcomes against PIB releases and the outcome documents (endorser lists dated 23 February and 5 March 2026), and DPDP dates against the Rules in the Gazette issue dated 13 November 2025 (G.S.R. 846(E)). This is general information, not legal or policy advice, and no organisation paid for inclusion. Researched with AI assistance and reviewed by our editorial team.
What changed since February
When Swiss President Guy Parmelin announced Geneva at the Delhi summit, the federal press release of 19 February 2026 said only the first half of 2027. The plan is now far more specific.
| Item | February 2026 | Now |
|---|---|---|
| Dates and venue | First half of 2027, Geneva | 21 and 22 June 2027, Palexpo |
| Organisers | Environment and communications department (DETEC) with foreign affairs (FDFA) | Gianotti coordinates overall; DETEC leads preparations with FDFA and the economic affairs department |
| Framing | International law and fundamental rights | Bridging Innovation and Trust: prosperity for all, plus trustworthy, responsible and safe AI |
| Adjacent event | Not stated | ITU AI for Good Global Summit, 23 and 24 June, same venue; exhibition from 21 June |
| Budget | Not stated | About CHF 6 million for the core summit, roughly half from the Confederation, the rest from sponsors |
| Participation | Not stated | Curated, with limited capacity; rules due in late 2026 |
Swissinfo reported that Federal Councillor Albert Rösti announced the dates on 6 July 2026 at the first UN Global Dialogue on AI Governance in Geneva, speaking on behalf of about 40 countries, and named developing countries' participation as one focus.
The content is still moving. The Federal Council wants partnerships and joint approaches that outlast the event, and says sponsors' money will not influence the agenda or outcomes. On 5 October 2026 in New Delhi, Parmelin called India a key partner, promised continuity between the summits and said the thematic focus was still being refined, according to New Kerala.
Key dates on the road to Geneva
These dates come from the summit site and its guidelines, the UN, ITU and the DPDP Rules.
| When | What | Who it matters to |
|---|---|---|
| By end of 2026 | Call for Inputs opens on the summit site, inviting views on themes for global AI governance | Policy teams, industry bodies, labs |
| Late 2026 | Participation and media accreditation rules published | Anyone hoping to attend |
| Now to early June 2027 | Road to Geneva pre-events can be registered, subject to approval | Universities, associations, companies |
| Early 2027 | Call for side-event proposals; events run in or around Geneva, the week before or alongside the summit | Firms wanting a stage in Geneva |
| 3 and 4 May 2027 | Second UN Global Dialogue on AI Governance, New York | Policy teams |
| On or about 13 May 2027 | Most DPDP Rules in force in India | Companies handling personal data |
| 21 and 22 June 2027 | Geneva AI Summit, Palexpo | Governments, labs, industry |
| 23 and 24 June 2027 | ITU AI for Good Global Summit, Palexpo; exhibition opens 21 June | Product teams; booths via ITU |
A pre-event needs a link to one of the two priorities and an individual assessment at the organisers' discretion, it cannot promote products at odds with the summit's values, and organisers pay for it. It must take place at least a week before the summit. The site listed 195 pre-events in 31 countries on 1 November 2026, including one by Swissnex in India at Bengaluru on 31 October.
The sponsorship deadline was 15 September 2026 and later requests may not be considered, so plan on pre-events, side events and the Call for Inputs instead.
What the New Delhi summit produced
PIB gives two date ranges: 18 and 19 February 2026 for the summit and 16 to 21 February for the wider programme, at Bharat Mandapam. It reports delegations from more than 100 countries and 20 international organisations, about 6 lakh in-person attendees and over 850 exhibitors. MeitY organised it, and India's government billed it as the first global AI summit hosted in the Global South.
The declaration
The New Delhi Declaration on AI Impact rests on seven pillars, or chakras: democratising AI resources; economic growth and social good; secure and trusted AI; AI for science; access for social empowerment; human capital; and resilient, innovative and efficient AI. It closes by saying the summit will contribute to voluntary, non-binding guidelines and principles.
Endorsements grew after the summit: 88 by 21 February, 91 by 24 February after Bangladesh, Costa Rica and Guatemala joined, and 92 in the list dated 5 March 2026. ORF notes the earlier declarations drew 29 signatures at Bletchley, 11 at Seoul and 60 at Paris, and that Delhi won both the United States and China although the US did not sign at Paris in 2025.
Other deliverables and who backed them
Counts are PIB's from 2 March 2026, except where a document's own list differs.
| Deliverable | Backing | What it is |
|---|---|---|
| Charter for the Democratic Diffusion of AI | 22 per PIB; 24 on its own list (20 countries, 4 institutions) | Voluntary principles on access to AI resources |
| Guidance Note on AI Governance | 22 countries | Voluntary reference for national governance capacity |
| Trusted AI Commons | 22 per PIB before launch; its list has 21 countries and UNICEF | Repository of tools, benchmarks and best practices |
| Guiding Principles on Resilient, Innovative and Efficient AI | 20+ countries | Energy-efficient AI, plus an infrastructure playbook |
| Guiding Principles for Reskilling in the Age of AI | 23 countries | Workforce principles, alongside an ILO playbook |
| Global AI Impact Commons | 80+ stories from 30+ countries | Platform to replicate deployed use cases |
| Network of AI for Science Institutions | 19 partner countries | Scientific collaboration |
| Alliance for Advancing Inclusion Through AI | 20 countries and UNICEF | Inclusion-focused collaboration |
The coalition narrows as the documents get specific. On lists dated 23 February 2026, neither the US nor China appears as an endorser of the Charter, the Guidance Note or the Trusted AI Commons, while Switzerland appears on all three.
On money and compute, PIB expects over USD 200 billion in AI-related investment and lists Reliance's USD 110 billion over seven years, Adani's USD 100 billion by 2035 and Google's USD 15 billion AI hub in Visakhapatnam. Those three alone add up to USD 225 billion (110 + 100 + 15), so the headline is not a simple total. India also said it would add 20,000 GPUs to the 38,000 or more already provisioned under the IndiaAI Mission.
The company commitments
The commitments document lists 13 signatories as on 23 February 2026: Amazon, Anthropic, BharatGen, Cohere, Google, Gnani.ai, G42, Meta, Microsoft, Mistral AI, OpenAI, Sarvam AI and Soket. PIB names the four Indian organisations among them. The two commitments are not equally firm.
- Usage data. By the next AI summit, signatories will publish statistical insights from anonymised, aggregated usage data, directly or through contributions to international efforts.
- Multilingual evaluation. Signatories will evaluate multilingual capability on a subset of languages and contexts of their choosing, and work with local ecosystems on under-represented languages. No date or publication duty is attached.
Progress reporting is qualified too: where practicable and appropriate, through mechanisms such as the Hiroshima AI Process Reporting Framework.
What Indian AI firms should watch
1. Whether the Delhi commitments report back
Geneva is the deadline for one promise: the usage statistics. If signatories publish them, Indian product teams gain free adoption data. If they do not, the impact framing loses credibility. Do not wait for Indian-language results, because the evaluation commitment lets each signatory pick its languages and sets no date. Keep running your own evaluations and compare them with whatever is published.
2. The trust agenda becomes technical
Both Swiss priorities connect to work Indian companies already do, such as evaluations, guardrails and benchmarks. The Trusted AI Commons framework invites voluntary contributions of datasets, benchmarks and toolkits, which could give Indian-language tools an international audience, though we found no public contribution route yet. Teams with strong evaluation sets should consider the Call for Inputs or a pre-event; OFCOM names businesses among those invited to pre-events.
3. India's own AI rulebook arrives first
On 8 October 2026, Union Minister Ashwini Vaishnaw said MeitY would publish an AI regulation consultation paper within a month, focused on safety, a human-first approach, scale, inclusion and deepfake harms, with a techno-legal approach in which industry carries most of the responsibility, as MediaNama reported. On 26 October MediaNama reported him saying the coming weeks, with input from civil society, citizens and industry, and said it is unclear whether the rules will sit in a new law or under the IT Act. MeitY's document index showed no such paper on 1 November 2026.
MediaNama reads this as a change from MeitY's earlier reluctance to regulate. Our AIGEG and TPEC explainer covers the governance bodies set up in April 2026. India will reach Geneva with a domestic position under consultation, which is likely to shape what its delegates argue for.
4. DPDP compliance lands 39 days before the summit
Under the DPDP Rules, Rules 3, 5 to 16, 22 and 23 start eighteen months after publication, on or about 13 May 2027. They cover notices, security safeguards, breach reporting, retention, children's data and Significant Data Fiduciary duties. Rule 4 on Consent Managers comes first, on or about 13 November 2026. AI teams holding personal data should have consent records, retention schedules and breach processes for training and inference data ready well before the summit. Our DPDP Rules checklist sets out the evidence to prepare.
5. Compute access and the diffusion charter
The Charter treats access to compute, representative datasets, foundation models and skilled talent as essential to taking part in AI development. India's own example is the IndiaAI compute programme, plus C-DAC's request of 7 October 2026 to empanel providers for up to 10,000 AI compute units, with bids now closing on 13 November 2026, extended from 29 October. Our IndiaAI 10,000-GPU explainer covers rates, eligibility and the tender timeline.
6. Geneva's multilateral setting
Geneva is home to the ITU and hosted the first UN Global Dialogue on AI Governance on 6 and 7 July 2026; the second runs in New York on 3 and 4 May 2027, seven weeks before the summit. The organisers see the summit series as a complement to existing international processes, not a replacement. For Indian firms, standards work and UN-track discussions may matter more than the summit's final text, and ITU's exhibition, opening on 21 June, is the practical place for product teams to be seen.
7. Whether Delhi's investment pledges become capacity
PIB's USD 200 billion is a set of expectations and announcements, not completed projects. Before Geneva, check which pledges have become data-centre permits, power contracts, GPU deliveries or cloud regions customers can use. For buyers, that decides whether Indian-hosted capacity is still scarce in mid-2027, which affects pricing and data-residency options.
What policymakers should watch
- Continuity of Delhi's platforms. Will the Global AI Impact Commons, the Trusted AI Commons and the AI for Science network carry into Geneva's outcome documents? Parmelin's promise of continuity is the test.
- The coalition. Delhi's 92 endorsements included the US and China, but neither is on the three narrower lists above. A text centred on trust and safety may be harder to agree than Delhi's impact framing.
- Who shares the chair. WAM, the UAE's state news agency, says the UAE will co-chair the 2027 summit with Switzerland and host in 2028, and Swissinfo reported Parmelin saying Switzerland would work with the UAE. The August federal pages and the summit site do not mention a co-chair, so treat that as the UAE's description for now.
- Binding versus voluntary. PIB describes Delhi's deliverables as voluntary. Watch whether Geneva adds any reporting obligations, even soft ones.
What to do now, by role
| Role | Action before June 2027 |
|---|---|
| Founder or model builder | Track the Call for Inputs; prepare Indian-language evaluation results; decide on a pre-event |
| Enterprise compliance lead | Finish DPDP readiness by 13 May 2027; follow MeitY's consultation; map AI use cases to guardrails and audit trails |
| IT services exporter | Watch the trust agenda alongside EU and US customer requirements; package evaluation and governance services |
| Researchers and academia | Check the AI for Science network and pre-events; plan side-event proposals for early 2027 |
IT services teams can read this alongside our Nasscom NTLF 2027 preview.
India, EU and US notes
For Indian firms, the binding obligations through 2027 are domestic, mainly DPDP and any AI rules from MeitY's consultation. Firms selling into Europe still answer to EU law, as our EU AI Act timetable note explains, and US customers set their own contractual AI requirements. Use the summit to anticipate direction, and build to the binding rules.
How eCorpIT can help
eCorpIT is a Gurugram-based technology organisation, founded in 2021, assessed at CMMI Level 5 and MSME certified, with senior-led engineering teams working across AWS, Microsoft and Google platforms. Our AI evals and observability service builds the golden sets, scorers, traces and release gates that a trust agenda will ask AI teams to show. Talk to us at /contact-us/.
Last updated: 1 November 2026.
