Summary: The International Financial Services Centres Authority (IFSCA), on 18 August 2026, issued a consultation paper proposing to specify operating lease, including any hybrid of operating and financial lease, of GPUs and connected data centre equipment as a financial product in GIFT IFSC. The proposal draws upon Notification No. S.O. 5199(E) dated 14 December 2021 and would bring the activity within Section 3(1)(d)(vi) of the IFSCA Act, 2019. Classification of leasing arrangements would be based on Ind AS 116, and the activity is contemplated within the regulatory architecture of the IFSCA (Finance Company) Regulations, 2021. IFSCA noted that GPUs are critical to AI workloads because of their parallel-processing capabilities and that India’s boarded AI GPU capacity, stated at 38,000 as of October 2025, is expected to reach 1,00,000 by the end of 2026. The consultation paper estimates that deployment of 650,000-700,000 GPUs in Indian data centres over the next five years could represent a USD 23 billion investment opportunity. It states that India presently remains entirely import-dependent for such equipment and that commercial requirements of cloud operators, colocation providers, hyperscalers and enterprises need market-based financing solutions alongside public compute infrastructure under the IndiaAI Mission.
IFSCA considers leasing particularly suited to GPUs because of their high capital cost, supply constraints, rapid technology-refresh cycles and significant obsolescence and residual-value risks. An operating or hybrid lease could convert large upfront ownership costs into periodic payments while allowing lessors to manage and remarket equipment as technology and workloads change. The proposal covers leasing of identified GPU equipment, including sale-and-leaseback arrangements, but not GPU-as-a-Service (GPUaaS). Under the consultation paper’s Ind AS 116 analysis, equipment leasing requires the lessee to control the use of an identified asset, whereas GPUaaS, where the customer purchases compute capacity and the hardware may be substituted, constitutes a service arrangement rather than equipment leasing.
The proposed definition of “GPU and connected data centre equipment” is technology-neutral and covers accelerated computing processors such as GPUs, GPGPUs, AI accelerators, TPUs and NPUs, together with integrated rack-scale AI systems, servers, storage units, networking and interconnect equipment, server power supplies and associated parts and accessories. Annexure II additionally seeks stakeholder views on whether allied power, cooling and thermal infrastructure—including high-capacity fans, precision air-conditioning equipment, chillers, liquid and immersion cooling systems, power distribution units, busways and dedicated switchboards—should fall within the permitted leasing product. The HSN/CTH mapping is expressly indicative and is not intended to determine customs classification, duty, exemption or treatment under other laws.
IFSCA’s proposal would expand an IFSC leasing ecosystem that already includes aircraft, ships and oilfield equipment. The existing Framework for Aircraft Lease and Framework for Ship Leasing illustrate the regulatory model, while IFSCA had also specified operating lease of oilfield equipment as a financial product in January 2026. Separately, IFSCA has proposed a consolidated Framework for Leasing Activity in IFSC, into which GPU and connected data centre equipment leasing may subsequently be incorporated after considering stakeholder feedback.
Public comments have been invited on the draft notification, scope of covered GPU and connected equipment, indicative HSN/CTH classifications, inclusion of allied power and thermal equipment and other relevant aspects. Comments are required to be submitted to IFSCA by 28 August 2026. The consultation paper is only a proposal at this stage; operating lease, including any hybrid of operating and financial lease, of GPU and connected data centre equipment would become a financial product only upon issuance of the proposed notification and the consequential regulatory framework.
International Financial Services Centres Authority
PRESS RELEASE
IFSCA issues consultation paper on enabling leasing of Graphics Processing Units (GPUs) and connected data centre equipment as a financial product
IFSCA today issued a consultation paper (including draft notification) on enabling Operating Lease, including any hybrid of Operating and Financial Lease, of GPUs and connected Data Centre Equipment as a financial product in GIFT IFSC.
GPUs play a crucial role in Artificial Intelligence (AI) by providing the massive parallel processing power needed to handle complex computations efficiently. Unlike traditional CPUs, GPUs can perform thousands of operations simultaneously, making them ideal f or training and running machine learning and deep learning models. They significantly accelerate tasks such as image recognition, natural language processing, and generative AI by processing large datasets and neural network calculations much faster.
The global GPU market has become an industrial capital- goods market driven by artificial intelligence. It stood at approximately USD 65 billion in 2024 and is projected to reach USD 400–500 billion by 2030 (a CAGR of about 30–35%). India holds less than 5% of global AI- optimised compute power, while the United States and China together hold over 70%. India’s boarded AI GPU capacity stood at 38000 as on October 2025 and is expected to grow 1,00,000 by the end of 2026. Rising adoption of artificial intelligence (AI) could drive the deployment of 650,000- 700,000 GPUs in India’s data centres over the next five years, creating a USD 23 billion investment opportunity. India is at present entirely import-dependent for such equipment. While the shared public compute created under the India AI Mission is directed principally at startups, researchers, academia and public institutions; the commercial requirements of cloud operators, colocation providers, hyperscale’s and enterprises must be met by the market. A platform for leasing GPUs is well suited to address this gap.
Leasing is a practical and economically viable mode of accessing such equipment, given its high capital cost, acute supply constraints, rapid technology-refresh cycles and specialised nature. An operating lease, or a hybrid of an operating and financial lease, converts the large upfront capital cost of ownership into a periodic payment, transfers obsolescence and residual-value risk to a lessor better placed to price, manage and remarket it, and permits redeployment of assets across lessees as workloads shift.
The proposal is aligned with the Government’s policy direction to ensure affordable and reliable access to computing resources as an enabler of India’s AI ecosystem. Leasing provides access to computing infrastructure without requiring ownership and would complement, rather than compete with, public compute infrastructure. As with aircraft and ship leasing, enabling GPU equipment leasing at this early stage of the global AI infrastructure expansion would help attract leasing activity to the IFSC, broaden its leasing ecosystem and position India as an early participant in this emerging market.
The consultation paper has been hosted on IFSCA’s website and comments on the same may be provided till August 28, 2026.
Gandhinagar
August 18, 2026
PUBLIC CONSULTATION- ENABLING OPERATING LEASE INCLUDING ANY
HYBRID OF OPERATING AND FINANCIAL LEASE OF ‘GPU AND CONNECTED
DATA CENTRE EQUIPMENT’ AS A FINANCIAL PRODUCT
Objective
1. The objective of this consultation paper is to seek comments/views/suggestions from the public on a proposal for the Authority to specify “GPU and connected data centre equipment” in exercise of its powers derived from the Government of India notification No. SO 5199 (E) issued by the Department of Economic Affairs, Ministry of Finance, Government of India, published in the Gazette of India, Extraordinary, Part II, Section 3, Sub-section (ii), dated December 14, 20211. Once specified, operating lease, including any hybrid of operating and financial lease, of ‘GPU and connected data centre equipment’ shall be a financial product under section 3(1)(d)(vi) of the IFSCA Act, 2019. The draft notification is placed at Annex I.
2. Classification of the type of lease of such equipment undertaken is based on Indian Accounting Standards (Ind AS 116). Vide notification No. S.O. 5199(E) dated December 14, 2021, the Government of India notified operating lease, including any hybrid of operating and financial lease, of any product or equipment specified by the Authority as a financial product; such activity is permitted under regulation 5(1)(iii)(g) of the IFSCA (Finance Company) Regulations, 2021. To date, operating lease of aircraft, ships and oilfield equipment have been enabled2.
3. The equipment so leased out may be held by the financial institution in IFSC either on ownership basis or on lease-in-lease-out basis. The product is proposed to be described through functional descriptions, read with an indicative mapping to the headings of the First Schedule to the Customs Tariff Act, 1975, as set out in the HSN / CTH Classification Matrix at Annex II. The equipment proposed to be enabled for leasing under the draft notification is with the intent of covering accelerated computing devices together with the equipment/s without which such devices cannot be deployed in a data centre environment.
4. After examining the stakeholder feedback on points (1) and (2) above, if a case for doing so is found to exist, operating lease, including any hybrid of operating and financial lease, of ‘GPU and connected data centre equipment’ by a financial institution in IFSC shall be included under the Framework for Leasing Activity in the International Financial Services Centre3 (‘Leasing Framework’), with such modifications as may be necessary.
Background
5. ‘GPU and connected data centre equipment’4 refers to accelerated computing devices together with the servers, storage, networking, interconnect and power equipment used directly or indirectly in connection with the deployment and operation of such devices in a data centre, such assets being economically operable only as part of an integrated cluster.
6. Internationally, the GPU market has become an industrial capital-goods market driven by artificial intelligence. It stood at approximately USD 65 billion in 2024 and is projected to reach USD 400-500 billion by 2030. Global data centre capital expenditure of approximately USD 6.7 trillion is estimated to be required by 2030, of which about USD 5.2 trillion is attributable to AI-capable capacity, and the market for accessing compute rather than owning it is forecast to grow from about USD 21 billion in 2024 to about USD 134 billion by 20305. Supply remains highly concentrated, with a single manufacturer estimated to hold approximately 80-85% of the data centre GPU market and fabrication contracted to a single foundry, resulting in acute supply constraints and high capital costs6.
7. India holds less than 5% of global AI-optimised compute power, while the United States and China together hold over 70%7. India’s boarded AI GPU capacity stood at 38000 as on October 20258 and is expected to grow 1,00,000 by the end of 20269. Rising adoption of artificial intelligence (AI) could drive the deployment of 650,000-700,000 GPUs in India’s data centres over the next five years, creating a USD 23 billion investment opportunity10. India is at present entirely import-dependent for such equipment. The shared public compute created under the India AI Mission11 is directed principally at startups, researchers, academia and public institutions; the commercial requirements of cloud operators, colocation providers, hyperscalers and enterprises must be met by the market. A leasing market is well suited to address this gap.
8. Leasing is a practical and economically viable mode of accessing such equipment, given its high capital cost, acute supply constraints, rapid technology-refresh cycles12 and specialised nature. An operating lease, or a hybrid of an operating and financial lease, converts the large upfront capital cost of ownership into a periodic payment, transfers obsolescence and residual-value risk to a lessor better placed to price, manage and remarket it, and permits redeployment of assets across lessees as workloads shift.
9. GPU capacity is accessed through either (i) leasing of identified GPU equipment, where the lessor owns and leases the hardware to a cloud operator, colocation provider or enterprise, including through sale-and-leaseback arrangements; or (ii) GPU-as-a-Service (GPUaaS), where the provider retains the hardware and supplies computing capacity from a shared pool. The proposal covers only the first model. Consistent with Ind AS 116, a lease requires the lessee to have the right to control the use of an identified asset; accordingly, where specific GPU equipment is identified and placed under the lessee’s control, the arrangement constitutes equipment leasing, whereas GPUaaS, where hardware may be substituted and the customer purchases computing capacity rather than control of identified equipment, is a service arrangement.
10. The proposal is aligned with the Government’s policy direction to ensure affordable and reliable access to computing resources as an enabler of India’s AI ecosystem. Leasing provides access to computing infrastructure without requiring ownership and would complement, rather than compete with, public compute infrastructure. As with aircraft and ship leasing, enabling GPU equipment leasing at this early stage of the global AI infrastructure expansion would help attract leasing activity to the IFSC, broaden its leasing ecosystem and position India as an early participant in this emerging market.
11. Looking at the world statistics and international practice, GPU deployment is concentrated in the United States, China and the European Union. Jurisdictions such as Singapore, the United Arab Emirates and Saudi Arabia are actively building AI compute capacity and associated financing ecosystems, principally through sovereign and corporate balance sheets13.. The Authority has also received requests from stakeholders enquiring about the permissibility of operating lease of such equipment from the IFSC.
On examination of such requests, it is felt that enabling operating lease of GPU and connected data centre equipment as a distinct financial activity from IFSC can play a significant role in supporting India’s AI infrastructure build-out and also develop GIFT-IFSC as a regional hub for such leasing services.
12. In order to capitalise on the growth potential, the benefits to the Indian economy and developing GIFT IFSC as a regional hub for leasing of GPU and connected data centre equipment, it is proposed to notify operating lease, including any hybrid of operating and financial lease of such equipment, as a financial product.
Public Comment
13. Comments and suggestions are invited from the public on the following:
i. The draft notification at Annexure I, including the proposed scope of “GPU and connected data centre equipment”.
ii. Whether the proposed scope of “GPU and connected data centre equipment” is appropriate, including whether any equipment may require inclusion or exclusion.
iii. The indicative HSN / CTH Classification Matrix at Annexure II, including whether the listed equipment and CTH references adequately reflect the proposed scope.
iv. Whether allied power, cooling and thermal equipment specified in Part B of Annexure II should be included within the scope of the proposed product and, if so, the appropriate parameters for determining such inclusion.
v. Any other matters that may be considered relevant for the proposed framework for leasing of GPU and connected data centre equipment in the IFSC.
14. The information furnished under items (i) to (v) above must be supported by rationale in line with the principles that have been adopted by the Authority for consideration of such proposal, as can be observed in the above background note.
15. The comments may be sent by email to Mr. Sri Vishnu Charan, Manager at [email protected], Mr. Nishil Patel, Consultant, IFSCA at [email protected] with a copy to Ms. Riddhi Bhandari, Chief General Manager at [email protected] with the subject line “Comments on the draft notification to enable operating lease, including any hybrid of operating and financial lease, of GPU and connected data centre equipment in IFSC”, latest by August 28, 2026.
16. The comments/feedback should be submitted in the following format:
| Name and Designation | |
|---|---|
| Contact No. and Email Address | |
| Name of Organisation |
—
| S. No. |
Clause no. of the draft Notification / Annex | Text of the clause | Comments/
suggestion/ suggested modification |
Detailed rationale |
|---|---|---|---|---|
Annexure I
Draft Notification
INTERNATIONAL FINANCIAL SERVICES CENTRES AUTHORITY
NOTIFICATION
Gandhinagar
No. IFSCA/[•]/2026 – In exercise of the powers conferred by section 12 of the International Financial Services Centres Authority Act, 2019 (50 of 2019) (hereinafter referred as the “said Act”) read with notification number S.O. 5199(E) dated 14th December, 2021 issued by Ministry of Finance, Government of India, published in the Gazette of India, Extraordinary, Part II, Section 3, Sub-section (ii), dated the 14th December, 2021, the Authority hereby specifies an operating lease, including an hybrid of operating and financial lease, in respect of the following products or equipment as a financial product:
i. processing units, whether or not containing, in the same housing, one or two of the following types of units: storage units, input units or output units, including various types of storage units and allied parts and accessories;
ii. rack-scale artificial intelligence systems supplied and invoiced as one integrated system, including compute trays, switch trays, CPU and input/output units integrated within a single rack, and presented in the form of systems;
iii. electronic integrated circuits comprising processors and controllers, whether or not combined with memories, converters, logic circuits, amplifiers, clock and timing circuits or other circuits;
iv. server power supply units;
v. network switches, network access controllers and connected networking and interconnect equipment; and
vi. parts and accessories of the products or equipment specified in clauses (i) to (v), whether supplied along with such processing unit or system or supplied separately
Explanation 1: The aforementioned products and equipment shall collectively be referred to as “GPU and connected data centre equipment”.
Explanation 2: The descriptions in clauses (i) to (vi) are intended to be technology-neutral and shall include accelerated computing equipment and systems, including graphics processing units, general-purpose graphics processing units, artificial intelligence accelerators, tensor processing units, neural processing units and comparable computing or acceleration technologies, together with servers, storage, networking, interconnect and power equipment that are integral to, or ordinarily deployed as part of or in direct support of, such systems, across successive technology generations and irrespective of the manner in which such products or equipment are configured, presented, supplied or invoiced.
Explanation 3: For the purposes of this notification, the specification of a product or equipment shall not be dependent upon the tariff classification, heading, sub-heading or tariff item under the Customs Tariff Act, 1975 (51 of 1975), and the applicable classification of any product or equipment under the customs or other applicable law shall be determined in accordance with such law.
2. This notification shall come into force on the date of publication in the Official Gazette
K. Rajaraman
Chairperson
Annexure II
HSN / CTH CLASSIFICATION MATRIX
GPUs, GPU PARTS AND CONNECTED DATA CENTRE EQUIPMENT
The following matrix provides an indicative mapping of the products and equipment proposed to be covered by the notification to the relevant headings of the First Schedule to the Customs Tariff Act, 1975. The functional description of the equipment governs the scope of the proposed financial product, and the CTH references are indicative.
Part A – Core GPU and Connected Data Centre Equipment
| Sl. No. | Description of product/equipment | Indicative CTH | Illustrative coverage |
|---|---|---|---|
| 1. | Processing units, including storage units, input/output units and allied parts and accessories | 8471, 8473 | GPU / AI servers, compute nodes and storage units |
| 2. | Rack-scale AI systems supplied as an integrated system | 8471 | Integrated rack-scale accelerated computing systems comprising compute, interconnect and input/output units |
| 3. | Electronic integrated circuits comprising processors and controllers |
8542 | GPUs, AI accelerators, GPGPUs, TPUs, NPUs and comparable accelerated computing processors |
| 4. | Server power supply units | 8504 | Power supply units, static converters and related parts for servers and rack-scale systems |
| 5. | Network switches, network access controllers and connected networking / interconnect equipment | 8517, 8544 | Data centre network switches, network interface equipment, transceivers and copper / optical fibre interconnect |
| 6. | Parts and accessories of the equipment specified above | 8473, 8517, 8542 and other applicable headings | GPU cards, accelerator boards, modules, motherboards, riser cards, trays, mounting hardware, spares and replacement parts |
Part B – Allied Power and Thermal Equipment
| Sl. No. |
Description of
product/equipment |
Indicative CTH | Illustrative coverage |
|---|---|---|---|
| 7. | Cooling and air-movement equipment installed for leased compute |
8414 | High-capacity fans and related air-movement equipment |
| 8. | Precision air-conditioning equipment for data centre environments |
8415 | CRAC, CRAH, in-row and rear-door cooling units |
| 9. | Chillers and liquid-cooling equipment serving leased compute | 8418 | Chillers, liquid-cooling and immersion-cooling equipment, including CDUs |
| 10. | Electrical control and power distribution equipment dedicated to IT load | 8537 | PDUs, busway and dedicated switchboard panels |
Note: The CTH references above are indicative and are provided to assist stakeholders in identifying the scope of equipment proposed to be covered. The classification of any specific product or equipment under the Customs Tariff Act, 1975 shall be determined in accordance with the applicable law. The inclusion of any product or equipment in this matrix is solely for identifying the proposed scope of the financial product and shall not, by itself, determine its customs classification, duty or exemption, or affect its treatment under any other applicable law or regulatory requirement.
Notes:
1 Issued under section 3(1)(e)(xiv) of the International Financial Services Centres Authority Act, 2019 by the Ministry of Finance, Department of Economic Affairs, Government of India, and published in the Gazette of India, Extraordinary, Part II, Section 3, Sub-section (ii), dated December 14, 2021.
2 Framework for Aircraft Operating Lease (Circular No. 172/IFSCA/Finance Company Regulations/2022-23/001 dated May 18, 2022, as amended); Framework for Ship Leasing (Circular No. 496/IFSCA/FC/SLF/2022-23/001 dated August 16, 2022, as amended); and the notification specifying ‘oilfield equipment’ issued in January 2026.
3 Consultation Paper on the Draft Circular titled ‘Framework for Leasing Activity in International Financial Services Centre’, issued by the Authority in August 2026. References in this paper to the said Framework are to the draft circular as placed for public consultation.
4 GPU’ (Graphics Processing Unit) refers to accelerated computing processors, including AI accelerators, GPGPUs, TPUs, NPUs and comparable devices, deployed for artificial intelligence model training, inferencing, high-performance computing, rendering and allied workloads; ‘connected data centre equipment’ refers to the servers, storage units, networking and interconnect apparatus, power and thermal management equipment, and associated equipment used directly or indirectly in connection with the deployment and operation of such GPUs in a data centre environment.
5 McKinsey & Company, ‘The cost of compute: A $7 trillion race to scale data centres’ (2025); International Data Corporation (IDC), Worldwide AI Infrastructure Spending Guide (2025); and Analysys Mason, ‘GPU-as-a-Service (GPUaaS): worldwide forecast 2025-2030’ (2025).
6 Industry estimates and Stanford University, ‘AI Index Report’. Cross-border movement of such devices is additionally subject to the export control regime of the country of origin.
7 PSA White Paper, supra, citing Stanford University, ‘AI Index Report’ (2024).
8 Transforming India with AI, PIB press release dated October 12, 2025.
9 Abhishek Singh, Additional secretary, Ministry of Electronics and IT (MeitY), and CEO of IndiaAI Mission – The Indian Express (web edition) – February 7, 2026.
10 Data Centres: Powering India’s AI boom. Avendus Capital. May 2026.
11 IndiaAI Mission, Ministry of Electronics and Information Technology, Government of India (approved March 2024), which provides shared, subsidised compute infrastructure directed principally at startups, researchers, academia and public institutions.
12 Each successive generation (for example, H100, H200 and B200/B300 within a span of two to three years) has brought a step change in performance per watt, so obsolescence risk, rather than physical wear, is the principal determinant of asset value in this class.
13 Publicly announced sovereign AI programmes include Saudi Arabia’s HUMAIN and Stargate UAE; France, Japan, the Republic of Korea, Singapore and the European Union have announced comparable national programmes. Figures are drawn from public announcements and industry trackers anda are indicative.
