Hit Ctrl + R on government tech shopping

As policymakers grapple with implications of AI, much of the debate has focused on ensuring that regulation keeps pace with innovation. Yet, regulation is only one part of the equation. An equally important policy instrument that must keep pace with tech is public procurement.

Public procurement determines what goods and services are purchased by the government and, consequently, which technologies are adopted at scale. This is one of government's most powerful instruments of industrial policy. Accounting for 20-22% of India's GDP, it shapes markets, creates demand and influences investment decisions. It's guided by Public Procurement (Preference to Make in India) Order (PPP-MII) 2017.

As India develops capabilities in emerging technologies, this objective will remain central. Until then, however, procurement of emerging tech must promote efficiency and competition by avoiding unnecessary dependence on any single tech or vendor. Well-designed procurement frameworks can strengthen domestic industry while ensuring that public agencies continue to access the best available technologies in computing performance, efficiency and sustainability.


Strategic use of procurement is not unique to India. In the US, federal procurement played a pivotal role in developing transformative technologies such as internet, GPS and personal computing. At the heart of public procurement of tech are specifications that define functional, technical and performance requirements. The challenge arises when these specs become overly prescriptive.

While this is rarely problematic for traditional infra where materials and engineering principles remain stable, tech evolves far more rapidly. By the time specifications are drafted, vetted and finalised, they may be outdated. Specs that prescribe particular technologies can, therefore, inadvertently exclude newer alternatives that offer equal or better performance, efficiency and value.

This challenge is visible in MeitY's 'model technical specifications' for procurement of desktops, laptops and tablets, which prescribe x86 processor architecture along with detailed hardware requirements. Since they were last updated, computing has evolved. Newer processor architectures based on a reduced instruction set (RISC) offer improved battery life, better power efficiency, lower cooling requirements and lower costs.

At government scale, even modest improvements in device-level energy efficiency can reduce costs while advancing sustainability goals. This is why procurement specs should focus on objective performance requirements, allowing competing technologies to be evaluated on their merits, promoting tech neutrality and wider choice for procurement agencies.

So, the way forward is not fewer specifications but less prescriptive ones that remain flexible enough to accommodate technological change. A few steps can help achieve this:

Technical specifications should be reviewed on a fixed, more frequent cycle. The US refreshes its government-wide desktop and laptop configurations roughly every 9 mths, incorporating feedback from industry to reflect technological advances. For India, regular, time-bound reviews would help ensure public procurement remains technologically current and commercially competitive.

Procurement specs should become more performance- and outcome-based. Rather than prescribing specific makes or technology architectures, they should define the functional, security and performance outcomes products must achieve. This would encourage innovation while allowing newer technologies to compete on equal terms.

The EU's 2014 Public Procurement Directive discourages technical specifications that create unjustified barriers to competition. Britain's Procurement Act 2023 requires contracting authorities to describe performance or functional requirements, instead of prescribing particular designs or technologies.

Performance- and outcome-based specifications should be supported by objective benchmarks and recognised industry standards. For example, processor performance can be measured using benchmarks such as Geekbench and Cinebench for CPUs, 3DMark for GPUs, and Procyon for AI workloads.

Alternatively, GoI could consider withdrawing model technical specifications. Prior to their introduction, procuring agencies determined specs based on their intended end-use, with technical support from agencies like NIC where required.

Returning to this approach would reduce an additional layer of compliance and give procuring agencies greater flexibility. Procurement frameworks that are tech-neutral, performance-based and regularly updated will ensure that governments can adopt best available technologies while fostering competition and innovation.