India’s courts had 5.64 crore pending cases in July 2026, according to National Judicial Data Grid (NJDG) data that the Law Minister placed before the Rajya Sabha. In a reply to the Lok Sabha, the government put the figures at more than 4.99 crore in subordinate courts, 64.67 lakh in the High Courts and 92,755 in the Supreme Court. The direction is worse than the size. In July 2023, pendency in district and subordinate courts stood at 4.41 crore. That means the lower courts added nearly 60 lakh cases to their queue in three years. This is seen as a justice problem, but it is not just that. This delay harms entrepreneurs, limiting growth and the creation of new jobs and economic activity. The average pendency of tax cases is particularly acute at nearly 6 years per case; this is active capital blocked due to lethargy in judicial decision-making.An entrepreneur needs confidence that a customer who defaults can be made to pay, that a lease will hold, and that a land title will survive a challenge. Each of these actions or inactions rests on a judge in a court. But the court repeatedly fails them. New entrepreneurs, especially those not yet part of existing business networks, find it very difficult to get started because they have no assurance that signed contracts will be honored. Large organizations exploit the justice system’s sclerosis to delay payments to MSME vendors. Further inhibiting the growth of entrepreneurship.
When enforcement takes time, new entrepreneurs suffer. They can only deal with people they know; they demand money up front and avoid expanding in ways that expose them to disputes or failures, reducing their chances for expansion and growth. A large company can carry a decade of litigation. A small manufacturer goes bankrupt during this period.
Some founders choose to work alone rather than share ideas, capital, and responsibility with partners. Their ventures remain smaller than they might otherwise become.
These consequences extend to finance. If recovery is slow and uncertain, lenders may demand more collateral, charge more, or decline a loan. That leaves an entrepreneur with less capital to expand.
Intellectual property matters just as much for deep tech. Such ventures may spend years developing an invention before earning revenue. Investors need confidence that IP rights can be established and enforced within a reasonable time. When protection is uncertain, and disputes drag on, funding becomes harder to secure. Promising ideas may never reach the market, and the skilled jobs they could create may never materialise. Lack of judicial enforceability is one of the biggest unstated reasons behind poor R&D investments in the country. Entrepreneurs know they can’t protect their innovation in court, so they shy away from investing in research and development.
Slow enforcement is one reason Indian firms under-invest in research, and one of the easiest to fix. When courts and IP tribunals move faster, firms invest in innovation. After the United States sent patent cases to specialist judges, patenting by firms in those areas rose by about 7%, driven partly by higher R&D spending. China’s specialized IP courts raised invention patents by more than a fifth. In India, the industries that depend most on enforceable contracts, such as computers, telecom equipment and automobiles, grow fastest where courts are quickest. Practitioners point out to the Delhi High Court as an example of a court that has improved its capabilities to handle IP disputes. This has
Wealthy families and family offices could provide patient capital to deep tech ventures. Yet they don’t and prefer financial investments, including listed equities, to starting or backing a business that depends on enforceable contracts, partnerships and IP rights over many years. This may be one of the biggest reasons that some second-generation business families choose financial assets over new ventures.
A dependable justice system would help entrepreneurs build beyond established business circles and major cities. It would let a young firm take an order from an unfamiliar customer in another state because the contract means something in practice, not only on paper.
For instance one of the reason software services industry has grown so rapidly is because of its industry association -NASSCOM’s model was built on trust and collaboration. It helped create a thriving, export-driven technology services industry. Indian companies served customers in countries where contracts could be enforced. They also knew that failing to deliver quality would come at a high price. That pressure pushed them to improve, earn trust, win more business, and create more jobs. A rapidly moving flywheel was built.
When the justice system becomes a weapon, it can have the opposite effect on job creation. Courts are not the reason behind every decision to avoid a new venture. But weak contract enforcement makes it harder to start an enterprise, take informed risks, and grow with confidence.
The challenge is that the judicial system does not recognize this problem; it treats every intervention as interference and conflict. Judges at all levels protect their turf in a way rarely seen in any other profession; lawgivers who claim omniscience and infallibility because of a title are holding back the country’s economic development.
The Economic Survey of 2017-18 found that delays discouraged investment, stalled projects, hurt tax collections and put stress on taxpayers. In 2016, DAKSH, a legal research group, estimated the cost of judicial delay at around 0.5 percent of GDP, on a conservative basis.
More recently, DAKSH counted 3.56 lakh cases pending before commercial tribunals as of September 2025, involving Rs 24.72 lakh crore, or roughly 7.48 percent of 2024-25 GDP. That money is not lost. It is locked up, unavailable for investment until the disputes are resolved. That means 7.5% of the capital is blocked in courts and is not being productively utilised for economic activity and creating jobs.
International comparisons point the same way. In the 2025 World Justice Project Rule of Law Index, India ranks 86th of 143 countries overall, but 114th on civil justice.
Part of the problem is capacity. The India Justice Report 2025 counts 21,285 judges for 1.4 billion people, roughly 15 per million. That is well below the 50 per million the Law Commission recommended in 1987. The average district judge carries about 2,200 cases, and 33% of High Court and 21% of district court judge posts were vacant.
Appointing judges is essential, but it is slow, and it is only half the problem. Many cases are not waiting for a judge. They are waiting to become ready for one.
The government says this is due to pendency lists, frequent adjournments, and inadequate arrangements to monitor, track, and bunch cases for hearing. The 2017-18 Economic Survey estimated that nearly 30% of a project’s life was taken up by formal proceedings such as serving summons and notices.
The courts are also taking on more cases than they should. Supreme Court admits Special Leave Petitions under Article 136 of the Constitution, this increased from around 25% in 2008 to nearly 40% in 2016. In contrast, the Supreme Court of the United States of America and Canada admit 3% and 9% respectively of the cases filed before it. Similarly the HC in many cases are taking on cases which should go to the district courts
Articles 226 and 227 of the Constitution of India empower High Courts with carefully circumscribed writ jurisdiction.6 In practice, however, High Courts have expansively interpreted this provision over a period of time, resulting in a substantial increase in Article 226 cases.7 There are currently one million Writ Petitions pending at the 6 High Courts studied, constituting between 50-60% of the Court backlog, with average pendency fluctuating between 3-10 years. Increasing work loads without proper processes to handle the existing backlog seems to represent some kind amorphous greed for more cases if nothing else.
For delays, the judiciary has already built the framework within which technology can help. On 3 June 2026, the Supreme Court’s AI Committee published draft Regulations for Use of Artificial Intelligence in Courts. The draft sets a presumption in favour of responsible adoption. It also draws firm limits: no judgment, order or finding may rest on algorithmic decision-making alone, and the judge remains the determinative authority in every adjudicative decision.
The draft also settles who decides. For a High Court and the courts under it, the regulations come into force on a date that the High Court’s Chief Justice notifies. Approvals for courts under a High Court rest with that High Court’s AI Committee. This fits the Constitution, which places the district judiciary under the control of the High Courts.
So the pilot we propose should belong to a High Court. The government’s role would be to support it: funding, digital infrastructure and technical capacity, and any procedural amendments the High Court asks for.
The pilot would work in one district court, on one kind of case: property disputes. Property disputes are a sensible starting point because the facts are often documented and the precedent is well settled. These disputes are also where land, credit and investment get stuck. The pilot would run in two layers.
The first layer tackles the delays before a hearing. Across all property suits in the chosen court, AI tools would:
● track whether summons and notices have been served,
● flag unserved process,
● record the reason for every adjournment, and
● identify which cases are ready to be heard.
The draft regulations already list case management, docket prioritisation and backlog monitoring among permissible uses, subject to approval. This layer needs no opt-in because it changes how the court manages its list, not how any case is decided.
The second layer supports the judge once a case is ready. For cases where both parties agree, the AI would read the digitised file and map the facts to statutes and precedents. It would then prepare two documents: a structured case summary and a draft order.
The judge would accept, modify, or reject the draft. The judge would make the decision and provide the reasoning. Because drafting is not among the uses the regulations list, this layer would need the High Court AI Committee’s prior written approval, with reasons recorded.
Before any live case is touched, the second layer should be tested on property cases that have already been decided. The draft regulations provide for controlled environment testing, in which outputs cannot be used in any actual decision. Comparing the AI’s drafts with the judgments actually delivered would show whether the tool is reliable enough to go further. Only if it is would the pilot move to 2,000 live cases, and then to 10,000.
Fairness to litigants has to be built into the design, not added later.
● Consent. Consent to the second layer should come from both sides, in writing, and either side should be able to withdraw at any stage without prejudice to its case. Property disputes often pit unequal parties against each other, so the court should satisfy itself that consent is genuine.
● Disclosure. The draft regulations already require parties to be informed when an AI tool materially assists in managing or analysing their case.
● The record. The pilot should go further: the AI summary and draft order should be placed on the record and given to both parties. Any error can then be challenged, and an appellate court can see exactly what the trial judge saw.
The evaluation should be independent and designed before the first case. It should be commissioned by the High Court, not by the pilot’s backers. Litigants who opt in may differ from those who don’t, so the two layers should be measured separately.
● The first layer, which covers every eligible case, can be compared against a similar court nearby.
● Results from the second layer should be reported with appropriate caution about who chose to take part.
Everything should be published, including failures.
The risks are real. Earlier this year, a trial court was found to have relied on a fabricated AI-generated precedent. In July, the Supreme Court identified six such fabricated judgments in an order of the National Company Law Tribunal. Any tool in this pilot must therefore work from a closed, verified legal corpus. It must automatically check citations, and a person must verify each one before it reaches the judge.
Court data must stay under the court’s control. The draft regulations provide that where tools are developed using court data or resources, the court should retain ownership of the tool or a perpetual royalty-free license to it. The High Court should choose the vendor through its own procurement process, with any conflicts of interest disclosed.
None of this replaces the slower work of appointing judges, filling staff and forensic vacancies, and cutting avoidable government litigation. That work takes years. A well-designed pilot takes months, and it would give the judiciary, the government and the public something the debate on AI in courts has lacked so far: evidence from an Indian courtroom, gathered under the judiciary’s own rules.
Faster civil justice will create jobs if it unlocks the capital locked in cases. But the land, credit and contracts that businesses rely on all depend on it.
(Views are personal)
