The recent introduction of Bureau of Indian Standards (BIS) norms for data centre performance, cloud computing and AI ethics is an important step towards laying a framework for the standardisation of this new, rapidly developing ecosystem. The notification, issued on February 25, 2026, was made under Sub-rule (1) of Rule 15 of the BIS Rules, 2018.
One objective is to ensure that India stays aligned with norms set by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). Compliance with the BIS norms at this stage is voluntary but it signals a baseline for quality.
This is important since a lot of the hardware and intellectual property is imported. As localisation is set into motion, the BIS could issue relevant quality control orders (QCOs) to ensure compliance with high standards. In conjunction with other policy measures such as tax holidays for data centres, the BIS notification underlines the push to accelerate investments in digital infrastructure.
The notification touches upon common definitions of cloud systems and introduces performance metrics for data centres and ethical standards for AI system design. The formal but voluntary framework provides clear norms for cloud computing, data centre efficiency and ethical AI deployment. These benchmarks ensure that there is a floor in terms of acceptable quality even as capacity ramps up.
NITI Aayog projects that data centre capacity will grow from 1.5 GW in 2025 to 8-10 GW by 2030. About $80 billion has already been committed for data centre capacity expansion by companies like Google, Amazon and Microsoft. Local giants such as the Adani Group, the Tata Group and Reliance are also committed to large investments in data centres. Government of India projections indicate that such investments could soon exceed $250 billion, which would be a very significant percentage of India’s $4 trillion economy.
Not issuing QCOs at this stage allows for a degree of flexibility at a point where domestic manufacturing and research and development are just getting off the ground. At the same time, the regulations help reduce ambiguity around compliance, security and performance, which will help organisations to scale up even as they remain aligned with global standards.
The standards and technical definitions – the vocabulary – are derived from the ISO and IEC frameworks. India will use global sustainability metrics and benchmarks like power usage effectiveness, water usage effectiveness and carbon usage effectiveness to evaluate the environmental performance of data centres.
The BIS standards also establish common definitions and norms for the cloud, which will help in enabling interoperability and ensuring seamless data exchanges between different platforms and sectors. This standardisation is absolutely necessary to attract investments at the desired scale.
The BIS has adopted its own formal methodology for measuring cooling efficiency in data centres in terms of energy usage. The standards also directly address ethical considerations such as bias mitigation in the design and deployment of AI. AI design must incorporate transparency, accountability, fairness and bias mitigation to promote AI governance.
Managing energy consumption and heat mitigation in data centres are obvious concerns. Efficient cooling standards are necessary to minimise energy consumption and related environmental impacts. Interoperability is always useful. For example, consider use cases like healthcare applications, where different hospitals must access and process patient records. In such cases, cloud standardisations to ensure interoperability must be mandatory.
Ethical AI standards with robust bias prevention are necessary for public acceptance of increasing AI usage, given India’s large, demographically diverse population. For example, AI may be used for legal purposes such as case summaries, or for HR functions such as summarising and filtering job applications, or as a tool for evaluating applications and disbursing financial products like loans, insurance premiums and mortgages. In all these cases, the relevant algorithms must not be biased or prejudiced. Moreover, AI must not only be designed to be ethical and transparent, but must also be perceived as ethical and transparent in order to build public trust.
Translating standards into ground reality when it comes to AI design and deployment will be a challenge. Another challenge lies in the sheer pace of change. The technology is evolving quicker than any regulatory framework or set of standardisation norms. For example, the creation of sovereign AI models recently forced an update of cloud system definitions, even though those definitions were established only a few months ago.
Rigid mandatory BIS standards could increase compliance costs, making it harder for start-ups to negotiate the ecosystem. Keeping standards voluntary for the moment imparts flexibility, which could mitigate such entry barriers for start-ups, thereby driving innovation and entrepreneurship.
There also seems to be a gap in cybersecurity notifications. The standards focus on performance and ethics with much less emphasis on cybersecurity. India’s digital environment is already subjected to a large number of cyberattacks by bad actors, including scams and ransomware attacks. As AI penetration rises and more organisations move to the cloud, cybersecurity, or the lack of it, will become an increasing concern. This needs to be addressed on a priority basis to reduce the vulnerability of such systems.
In many developed countries, AI data centres have run into “Not in My Backyard” resistance from citizens for several reasons. For example, the state of Maine, USA, has passed a bill banning the construction of large new data centres while several other US states are considering similar legislation due to protests at the ground level. Many municipalities have refused to allow data centres to be set up.
Recent opinion polls indicate that over 60 per cent of Americans oppose data centres. One reason is the apprehension that AI will reduce employment opportunities. The other reasons for opposition arise directly from the resource-intensive nature of data centres, which are both energy- and water-intensive. This may have an immediate negative impact on the quality of life of nearby residents who experience pressure on power and water resources, and face supply disruptions, rising tariffs, or both.
India’s data centre and AI expansion is projected to raise data centres’ aggregated share in India’s power consumption from 0.8 per cent in 2025 to around 3 per cent by 2030. There could be frictions with civil society arising from such a spike in demand, especially in water-stressed regions. Policymakers may have to plan for such possible contingencies and navigate around such sensitive political issues.
These voluntary standards are, therefore, a way to nudge the sector in the right direction. They will need to be reviewed and amended to plug gaps, and at some stage the government will have to issue QCOs to make amended standards mandatory.
Those future QCOs must also include cybersecurity norms. More broadly, policy must address concerns about high resource intensity while encouraging capacity expansions. For example, green data centres that achieve high cooling efficiency ratios through liquid cooling or run on renewable energy should be rewarded through tax breaks, or some other mechanisms.
The BIS standards will also need to be reviewed periodically, perhaps every six months or annually, in order to keep up with the rapid pace of technological changes. Facilities will need to be inspected to ensure that standards are met on the ground. Training programmes for IT auditors, regulators and technology professionals will be required to skill up to ensure regular audits and reviews as the sector scales. This must be an ongoing process.
The notification of cloud computing and ethical AI standards by BIS is surely a positive step. It will encourage localisation, in conjunction with tax breaks and focused schemes designed like the production-linked incentive scheme. Setting up high-performance data centres with ethical AI, robust cybersecurity and sustainability principles is necessary for efficiency as well as for broad public acceptance.
Devangshu Datta
