India May Prioritise Domestic AI Models in Critical Sectors, Including Defence and Aerospace

India is considering a policy shift that could give homegrown artificial intelligence companies a stronger role in sensitive government operations. The Centre is likely to encourage, and potentially require, critical government sectors such as defence and aerospace to prioritise Indian-developed AI models, large language models (LLMs) and AI agents over frontier systems developed by foreign companies.

The proposal is linked to India’s ₹10,372 crore IndiaAI Mission, which has selected 20 indigenous AI model projects from 12 startups. The broader objective is to strengthen technological sovereignty, protect sensitive information and reduce dependence on foreign AI infrastructure in areas where national security and data control are particularly important.

Key Takeaways

• The government is considering prioritising Indian AI models in critical sectors such as defence and aerospace.
• Some central and state government departments could be asked to avoid foreign-developed frontier AI models and agents.
• The ₹10,372 crore IndiaAI Mission supports the development of 20 indigenous AI models.
• Twelve Indian startups have been selected to develop these models.
• Companies including Sarvam AI, BharatGen, Gnani, Gan AI and Tech Mahindra are part of the domestic AI development ecosystem.
• Government officials claim some Indian models have achieved up to 80% of the capabilities of leading frontier AI models.
• Defence organisations are already developing air-gapped AI systems that can operate without external internet connectivity.
• The proposal could create new opportunities for Indian AI startups and technology companies.

Why Does India Want Government Departments to Use Domestic AI Models?

The government wants to reduce India’s dependence on artificial intelligence systems developed and controlled by foreign companies, particularly in areas involving sensitive government information.

AI models are increasingly being used for data analysis, document processing, cybersecurity, research, intelligence support and automated decision-making.

In critical sectors such as defence and aerospace, these applications may involve sensitive operational information, infrastructure details and national security-related data.

Using foreign-developed AI systems can create questions around data handling, hosting locations, external access, jurisdiction and long-term technological dependence.

For example, if a government department uses an AI model hosted on overseas infrastructure, sensitive information may need to be processed outside India’s direct technological control, depending on the deployment arrangement.

Developing domestic alternatives could give government agencies greater control over:

• Where AI models are hosted and operated.
• How sensitive government information is processed.
• Who can access AI systems and their underlying infrastructure.
• How models are customised for Indian requirements.
• How systems operate without external internet connectivity.
• How updates, security and technical support are managed.

The proposal therefore goes beyond encouraging Indian companies to develop AI products. It focuses on creating domestic capabilities that government organisations can actually deploy.

Defence and Aerospace Could Be the First Major Beneficiaries

Defence and aerospace are among the sectors likely to receive particular attention under the proposed approach.

According to Business Standard, some defence organisations are already developing their own air-gapped large language models that can operate at the edge.

An air-gapped system is isolated from external networks, reducing the possibility of unauthorised access through internet-connected infrastructure.

Edge deployment allows AI systems to process information locally rather than depending entirely on remote cloud servers.

These capabilities can be important for military environments where connectivity may be restricted, unreliable or unsuitable for sensitive operations.

Potential applications include:

• Analysing large volumes of defence documents.
• Processing intelligence and operational information.
• Supporting maintenance and equipment monitoring.
• Assisting with cybersecurity analysis.
• Organising technical and aerospace research.
• Providing decision-support tools to authorised personnel.

These are potential application areas, not confirmation that every system is already deployed by India’s armed forces.

The government has asked relevant organisations to examine how indigenous models developed under the IndiaAI Mission could be adapted for such environments.

What Is the IndiaAI Mission?

The IndiaAI Mission is the government’s flagship programme to develop India’s artificial intelligence ecosystem.

Approved by the Union Cabinet in March 2024, the initiative has a total financial outlay of approximately ₹10,372 crore over five years.

Its objective is to strengthen India’s capabilities across the AI value chain, including computing infrastructure, foundational models, datasets, research, applications and AI skills.

The mission covers several major components:

Component	Objective
IndiaAI Compute Capacity	Provide access to computing resources for AI development
IndiaAI Innovation Centre	Support indigenous foundational AI models
IndiaAI Datasets Platform	Improve access to datasets for AI development
IndiaAI Application Development	Encourage AI solutions for practical use cases
IndiaAI FutureSkills	Develop AI-related skills and talent
IndiaAI Startup Financing	Support AI startups and innovation
Safe and Trusted AI	Address safety, privacy, bias and responsible AI development

The mission is designed to create a domestic ecosystem rather than rely entirely on foreign AI developers and infrastructure providers.

The government’s latest push towards domestic model adoption represents a potential next step: moving from funding Indian AI development to using those products in important public-sector operations.

Which Indian Startups Are Developing Domestic AI Models?

The government has identified 20 indigenous foundation-model proposals involving 12 startups under the IndiaAI Mission.

These projects include general-purpose large language and multimodal models, as well as smaller language models designed for specific applications.

Several companies have already released models or demonstrated their technologies.

Company	AI development
Sarvam AI	Indigenous large language and multimodal models
BharatGen	Multilingual foundational AI models
Gnani AI	Speech and voice AI technology
Gan AI	Generative AI development
Soket AI	EKA-1 foundation model
ZenteiQ	Dhir 2B and Dhir 5B models
Tech Mahindra	Selected for indigenous AI model development
Fractal	Selected for AI model development
Shodh AI	Selected for AI model development

Other selected participants are also working on their respective models.

The government wants to expand the capabilities of these systems so they can handle specialised requirements in government and enterprise environments.

Some companies have already demonstrated models that can operate on private servers or without an internet connection, features that could be important for sensitive government deployments.

Indian AI Models Are Closing the Gap With Global Competitors

One of the most significant claims in the latest policy discussion concerns the capabilities of India’s domestic AI models.

A senior government official told Business Standard that some Indian models had reached up to 80% of the capabilities of leading frontier AI models.

This is an important statement because global AI companies have invested billions of dollars in training increasingly advanced systems.

However, the 80% figure is an official’s assessment reported by the publication. It is not an independently verified benchmark across all AI tasks or model categories.

Performance can vary considerably depending on the application.

A domestic model may perform well in Indian languages, document processing or specialised government workflows while still trailing international frontier systems in advanced reasoning, coding or other tasks.

For government adoption, the relevant question is not necessarily whether an Indian model can outperform every international competitor.

Instead, agencies will need to determine whether a particular domestic model is sufficiently accurate, secure and reliable for its intended application.

Why Air-Gapped AI Models Matter for National Security

An important part of India’s domestic AI strategy is the development of systems that can operate without connecting to the public internet.

Traditional cloud-based AI services generally depend on remote computing infrastructure.

While these services offer significant computing power and convenience, organisations handling sensitive information may have restrictions on sending data to external platforms.

Air-gapped AI systems provide an alternative.

They can be deployed within isolated environments where access is controlled by the organisation operating the system.

This can offer several advantages:

1. Greater data control: Sensitive information can remain within the organisation’s infrastructure.
2. Reduced external exposure: Isolating systems from public networks can reduce certain cyberattack pathways.
3. Offline operation: AI tools can continue functioning in environments with restricted connectivity.
4. Customisation: Models can be adapted to specialised organisational requirements.
5. Operational independence: Agencies can reduce their dependence on overseas cloud services.

However, air-gapping does not automatically make a system secure. Hardware vulnerabilities, malicious updates, insider threats and poor access controls can still create risks.

Security assessments and regular monitoring will remain essential.

What Could This Mean for Foreign AI Companies?

If the government formalises the proposed policy, foreign AI companies could face greater restrictions when providing services to sensitive government departments.

Companies such as OpenAI, Anthropic, Google and other international AI developers currently compete in India’s growing AI market.

A requirement to prioritise Indian-developed models in certain government applications could change procurement decisions.

However, the reported proposal does not establish a nationwide ban on foreign AI products.

The focus is on critical government sectors and potentially sensitive departments.

Foreign AI companies could still serve private businesses, consumers and other organisations where such restrictions do not apply.

There may also be opportunities for international companies to provide technology through partnerships, licensed deployments or infrastructure arrangements that satisfy government security requirements.

The final impact will depend on the precise policy framework and whether it introduces mandatory domestic-model requirements or a preference-based procurement system.

Opportunities for India’s AI Startup Ecosystem

A stronger government preference for indigenous AI could create a significant commercial opportunity for Indian startups.

Government departments represent a potentially large market for specialised technology solutions.

If domestic AI companies receive access to government projects, they could gain:

• Large institutional customers.
• Long-term technology contracts.
• Opportunities to develop specialised models.
• Access to government use cases and datasets, subject to applicable safeguards.
• Greater funding and commercial visibility.
• Opportunities to expand into defence, aerospace and other regulated industries.

The initiative could also encourage collaboration between AI startups, universities, research institutions and established IT services companies.

However, government procurement will require more than a domestic development address.

Indian models will need to demonstrate reliability, security, accuracy, scalability and the ability to operate within demanding technical environments.

The Bigger Picture

India’s proposed domestic AI push reflects a growing recognition that artificial intelligence is becoming strategically important infrastructure. Countries are increasingly concerned about who develops their AI systems, where sensitive data is processed and how much control they retain over critical technology.

The IndiaAI Mission has helped establish a pipeline of domestic model developers. The next challenge is turning those models into reliable systems that can meet the demanding requirements of government departments. If the proposed policy moves forward, defence and aerospace could become important early markets for India’s homegrown AI industry.

Frequently Asked Questions

1. Is India planning to ban foreign AI models?

No nationwide ban has been announced. According to Business Standard, the government is considering asking certain critical government departments to avoid foreign-developed frontier AI models and prioritise Indian alternatives.

2. Which sectors could use domestic AI models first?

Defence and aerospace are among the first sectors being considered because of their sensitive information and national security requirements.

3. What is the IndiaAI Mission budget?

The IndiaAI Mission has an approved financial outlay of approximately ₹10,372 crore over five years.

4. How many AI models has the government selected?

The government has selected 20 indigenous foundation-model proposals involving 12 startups.

5. Which companies are developing Indian AI models?

Companies include Sarvam AI, BharatGen, Gnani AI, Gan AI, Soket AI, ZenteiQ, Tech Mahindra, Fractal and Shodh AI, among others.

6. What is an air-gapped AI model?

An air-gapped AI system operates within an isolated computing environment that is not connected to external networks. This can help organisations maintain greater control over sensitive data and access.

7. Are Indian AI models as powerful as global frontier models?

Government officials have claimed that some domestic models have reached up to 80% of the capabilities of leading frontier models. However, this is not a universal or independently verified performance measurement across all tasks.

8. Will the policy affect private companies using foreign AI tools?

The reported proposal focuses on critical government departments. It does not establish restrictions on ordinary private-sector or consumer use of foreign AI services.

Looking Ahead

The next major development will be whether the government converts its current preference for domestic AI models into formal procurement guidelines or mandatory requirements for selected departments. Defence and aerospace organisations will need to evaluate whether Indian models can meet their security, reliability and performance requirements before expanding deployments.

For India’s AI industry, the opportunity extends beyond government contracts. Developing competitive domestic models could strengthen the country’s position in enterprise AI, cybersecurity, multilingual technology and specialised software. The success of the initiative will ultimately depend on whether Indian developers can build systems that combine strong performance with security, affordability and technological independence.