The Indian government should not invest in AI companies. It threatens fair competition

State equity creates multiple conflicting priorities. The state plays rule-maker, subsidy provider, customer, and shareholder

Sam Altman has recently offered the United States government roughly five per cent of OpenAI, modelled as a sovereign wealth fund. While this is being framed as an opportunity for ordinary Americans to share the upside of artificial intelligence, the Financial Times reports that, according to sources, the proposal is most likely intended to secure good relations with the administration and blunt political blowback ahead of the upcoming IPO.

In parallel, the Union government in India is expected to take a 1-2 per cent share in the home-grown AI startup Sarvam in exchange for the subsidy offered under the IndiaAI mission. The subsidy covered 40 per cent of the Rs 246.71 crore compute bill for six months of access to Nvidia GPUs.

While these look like similar developments, they are not. In the two cases mentioned above, the state taking equity is proposed for different reasons, and they carry very different risks. Even if major economies are doing some version of this, India must exercise restraint as the risks outweigh the benefits.

There are at least three rationales for the recent trend of nation-states taking equity in private companies, and more than one might be true for a given instance.

The first is control of strategic sectors. Markets optimising for cost might not address supply chain resilience or strategic considerations. This is evident in the case of China’s dominance in critical minerals. The state would like a say in supply chains it deems critical, and acquires one through investments in select companies. The US government’s stake in Intel, its position in the rare-earth company MP Materials, and the golden share in US Steel are examples.

The second is value capture. The state spent public money on supporting and derisking industry and wants the returns socialised rather than fully privatised. This is the equity financing model of India’s Research Development and Innovation fund, which targets early-stage deep-tech startups and private sector R&D projects. It is similar to the case of Sarvam and Intel.

The third is public wealth creation, where the state invests in private companies through instruments such as sovereign wealth funds that pay dividends to the state and citizens. This is a task that private venture capital with a higher risk appetite is better suited for. While this is Altman’s stated rationale, considering that it is the firm offering the option to the state, the intent seems to be an attempt to purchase political goodwill.

The first has a national security rationale, the second has a fiscal rationale, and the third has neither. However, there are risks with these approaches.

Semiconductors are strategic. So are critical minerals, telecom equipment, foundation models, satellite launch, batteries, pharmaceutical precursors and undersea cables. Washington’s list has grown from chips to rare earths to steel in under two years. Cumulatively, they describe a state that takes equity wherever it feels exposed. In a contested technological order, the list is ever expanding.

State equity also creates multiple conflicting priorities. The state plays rule-maker, subsidy provider, customer and shareholder. MeitY writes India’s AI rules; it runs the mission that allocates compute subsidies; government departments are among the customers for products created by companies receiving the subsidy, and the government is now a shareholder of the company. Every subsequent decision by that ministry acquires a shareholder value dimension, whether or not anyone intends it. A procurement decision favouring one company raises the value of the state’s own holding. So does a regulation that disadvantages a competitor.

Other distortions also compound the problem. A state that holds equity acquires an interest in the firm surviving, which makes failure politically sensitive. Failure is how markets move capital from the mediocre to the capable. Further, it distorts the allocation of capital by markets. Investors would read selection by the government as an implicit guarantee, thereby lowering the cost of capital for that firm at the expense of others. These distortions and conflicts of interest do not bode well for the industry at large.

The case for value capture is sound since governments subsidise the risk of innovation while private firms capture the returns. That asymmetry deserves correcting, but it is an argument for recovering public money, not for owning a piece of the company. A repayable advance or a milestone-linked grant can achieve this.

Where equity is genuinely unavoidable for strategic-control reasons, it should come with conditions. The state should hold equity through a professionally managed independent fund rather than a line ministry. It should also hold passively, without board seats, golden shares or veto rights. The ministry that regulates a sector should not manage the state’s holdings in it. Every stake should carry an exit horizon published at the point of entry, and every stake should carry performance conditions with real withdrawal provisions.

Finally, holdings and valuations should be disclosed annually. These do not make state equity a good idea, but they may make it a viable one where it is absolutely necessary.

The writer is a researcher with the high-tech geopolitics programme at the Takshashila Institution. Views are personal