For Indian enterprises, access to frontier AI is increasingly becoming a question of where the models run, not just what they can do.
Amazon Web Services has expanded Amazon Bedrock support for OpenAI's GPT-5.6 Terra and Luna models to India through Cross-Region inference, giving organizations another way to run advanced AI workloads closer to their existing cloud infrastructure. The expansion is particularly relevant for enterprises that face stricter requirements around data handling, governance and workload location.
The rollout follows AWS's earlier support for GPT-5.6 Sol, Terra and Luna through U.S. and global Cross-Region inference. Sol, the most capable model in the family, is not part of the India expansion announced so far.
What this product expansion brings
One important limitation remains: GPT-5.6 Sol, OpenAI's flagship model in the family, does not currently have an India geographic inference profile. Indian customers can access Sol through Amazon Bedrock's Global Cross-Region inference option, but those requests may be routed to supported AWS Regions outside India.
AWS has not explained why India Geo support currently includes Terra and Luna but not Sol, so attributing the decision to pricing would be speculative.
Cost nevertheless remains part of AWS's pitch. OpenAI cut GPT-5.6 Luna pricing by 80% and Terra pricing by 20% on July 30; AWS passed those reductions through to Bedrock customers. OpenAI later lowered Sol pricing as well, with AWS announcing reduced promotional Sol rates on Aug. 21.
Changing regulatory landscape
India enacted the Digital Personal Data Protection (DPDP) Act in 2023, establishing a statutory framework for digital personal data. The accompanying DPDP Rules were finalized in November 2025, establishing operational requirements across areas such as notices, security safeguards, breach reporting, and data handling.
The rollout is phased. According to law firm Fisher Phillips, several important deadlines arrive in 2026, while full compliance with the Act and Rules is scheduled for May 13, 2027.
That regulatory shift is making the location of cloud and AI infrastructure more important for enterprises deciding where to run sensitive workloads. AWS is expanding its physical footprint in India, recently breaking ground on a new data center in Hyderabad as part of an expansion of its existing AWS Asia Pacific (Hyderabad) Region.
Amazon says it plans to invest more than $21 billion in cloud infrastructure across India from 2026 through 2030, including additional data center capacity in Hyderabad and Mumbai.
That capability can be particularly important for organizations in regulated sectors. AWS specifically points to financial services, healthcare and the public sector as examples of customers that may have local data-processing requirements. Those requirements vary by organization, workload, and applicable regulations, so in-country inference should be viewed as a deployment option that can support compliance rather than as a guarantee of compliance itself.
Nitin Bawankule, head of Enterprise Sales, India at OpenAI, said: “For many organisations in India, especially in regulated sectors, local inferencing is a key part of adopting AI with confidence. Making our models available through Amazon Bedrock gives more organizations a practical way to bring that capability into production.”
For Indian enterprises, this new product offering from AWS signals that “frontier intelligence that runs where their data already lives” already meets regulatory requirements.
What eWeek found: Local AI does not automatically mean compliant AI
AWS is positioning in-country inference as an important option for regulated Indian enterprises, but the announcement does not establish that using GPT-5.6 Terra or Luna through Amazon Bedrock automatically makes an AI deployment compliant with India's data-protection or industry-specific rules.
What the expansion does provide is a way to keep GPT-5.6 inference processing within India supported. AWS says requests using its India geographic Cross-Region inference profiles are routed only between its Mumbai and Hyderabad Regions, allowing organizations to meet data residency requirements while drawing on capacity across both locations.
Businesses still need to determine whether their broader use of AI, including how data is collected, stored, logged and governed, satisfies the legal and industry-specific requirements that apply to them.
There is another notable limitation: AWS has brought Terra and Luna to India Geo inference, but not GPT-5.6 Sol. Neither AWS nor OpenAI has publicly attributed that omission to pricing, regulation, or infrastructure constraints, so it would be premature to assume cost drove the decision.
Taken together, the rollout shows that enterprise AI competition in India is increasingly about more than access to the most capable model. Deployment location, governance and cost are becoming part of the product decision itself.
Also read: OpenAI’s India expansion is accelerating, with a Tata partnership putting 100 MW of AI compute capacity on the table and an option to scale significantly higher.