On August 6, an amendment to
the Taxation and Other Laws (Amendment) Bill, 2026 was passed by the Lok Sabha.
It amends the Payment and Settlement Systems Act, 2007 to grant the
Government the power to introduce fees on select UPI and electronic payment
transactions in the future. While there is no immediate imposition of charges,
it gives the Central Government the legal framework to determine in future,
which transactions will remain free, and which could be charged as per
Merchant Discount Rate (MDR). Any decision on this aspect will come out through
a separate Government notification.
UPI (Unified Payments
Interface) is an Indian instant digital payment system, where the interface
facilitates inter-bank peer-to-peer (P2P) and
person-to-merchant (P2M) transactions. To make the business model viable there
was a Merchant discount rate (MDR), a fee that anyone paid to the payment
service providers to provide the facility.
The Payment and Settlement
Systems Act of 2007 made a change in 2020 to prevent the imposition of this MDR
on business establishments. Its revival now will benefit banks and payment
service providers, who have long argued that maintaining the UPI infrastructure
they provide free involves high costs. But there are risks.
UPI falls under the broad
framework of Digital Public Infrastructure (DPI). -
enabling secure and seamless interactions between people, businesses and governments.
India has built DPI for over 1.4 billion people at very low cost. It
is an open and accessible network, backed by regulation and a wide range of
applications that modernize the economy, reform governance and transform lives (https://www.pib.gov.in/PressReleasePage.aspx?PRID=2235812®=3&lang=2).
Before UPI was created, the
Indian payment structure relied heavily on foreign payment rails like Visa and
Mastercard. Now, the dependency on such foreign platforms have reduced
drastically due to its success. In July 2026, the total volume of transaction
on the UPI platform was Rs 29.9 lakh crore. UPI thus stands as a cornerstone of
India's low-cost Digital Public Infrastructure (DPI).
Drawing a parallel to UPI’s success, an article on AI written
by Mr Srivatsa Krishna in TOI titled “The next DPI — how India can commoditise AI, offer some useful suggestions.
The author highlights the
same risk AI is facing (that UPI faced earlier). He mentions "India serves
as the world’s digital quarry" - implying that raw talent, data and labour
are exported by India to the Silicon Valley, yet "Indian startups must
rent that intelligence back as dollar priced API tokens." The article
further mentions "Intelligence is no longer a luxury software product; it
is foundational infrastructure." To get over the hurdles India would
face while trying to win a capital-intensive race against global tech giants,
the author suggests AI can become India's next Digital Public Infrastructure.
Extending the Digital
Public Infrastructure (DPI) should it be examined if it is a compelling
strategy for India’s AI future? The Digital Public Infrastructure (DPI) now
happily sits as democratic model with transparent identity.
Rather than competing in a
capital-intensive race against US hyperscalers to build closed foundation models, the article raises a
question on whether India should focus on commoditising the
inference layer. The argument is that by driving down the cost of
running models, India can shift economic power from global model owners to
local application developers.
The proposed framework relies on three strategic pillars
● Compute & Energy: Dedicating renewable and nuclear power to compute infrastructure
to achieve sub-market GPU rates.
● Open-Source Models: Requiring state-subsidised AI to adopt open-weights licenses and
leveraging public data across 22 indic languages.
● Unified Intelligence Interface (UII): Establishing an open API gateway—a "UPI for
AI"—supported by state-subsidised token entitlements for
students, researchers, and small businesses.
The Article correctly
captures a critical economic risk: without sovereign, open infrastructure,
India remains an "extractive digital quarry"—exporting raw
data and talent only to import finished software. Commoditising inference
turns cognition into a universal public utility rather than an expensive
privilege.
From CRG some of the
perspectives that emerge are as follows:
The August 6 amendment on
imposition of charges on UPI and electronic payment transactions signals the
Government’s recognition of their economic impact and highlights the fact that
public digital utilities require sustainable financial models to survive and
scale.
The massive success of UPI
in India had a major impact:
● It reduced dependence on
international payment networks like Visa and Mastercard – enabling saving on
fees and other infrastructure charges associated with them.
However, the Government,
through its amendment, has considered the concerns of the Banks and the
FinTechs regarding the high operational costs they had to incur without getting
anything in return so far. Any service given for free will not only be
unsustainable financially in the long term, but its effectiveness and
efficiency would also be compromised, as funds would be needed for improving
the services offered by Banks and FinTechs.
As AI gains significant
importance in the Indian context (some have been actively discussed in our
Regulatory News Summary),
a parallel situation can be drawn with the situation before India adopted UPI.
In the current context, even as India has massive talent, data and labour, all
these get exported to the Silicon Valley, where new products are developed.
These products, developed by Indians, are then imported back to India by the
Indian startups at a much higher price, resulting in overall revenue loss not
just for them, but also for the Indian Economy. Similar situation happened with
India during the pre-UPI days.
Since building AI
infrastructure requires massive capital, and most Indian Companies cannot
compete against global tech giants, the author suggests AI can become India's
next Digital Public Infrastructure.
The suggestion of focusing
on lowering the cost of AI models rather than competing to build indigenous
models due to the high costs involved, seems logical as by
making execution cheap, fast, and accessible, Indian intelligence can be
positioned as a public utility (like UPI). India can shift from being a global
model owner to local application developer. Lowering costs of models would make
intelligence a public utility that would be cheap and fast. It could also
empower domestic developers and secure national AI sovereignty.
