India’s bullet train shifts from Japan’s Shinkansen model to an Indian-led system, reshaping technology transfer, costs and strategic control.
On the 97-km stretch between Surat and Vapi, the civil works for India's first high-speed railway are finished, including the river bridges. The structures were drawn up with Japanese technical design input and backed by yen loans. Yet the train scheduled to run there first is not a Shinkansen. It is the B28, a Bharat-made bullet train with a test speed of 280 km/h, an eight-car set being designed and developed by BEML Limited in collaboration with Integral Coach Factory (ICF), with a European Train Control System (ETCS) Level 2 signaling and train-control system being implemented by a consortium of Dineshchandra R. Agrawal Infracon Pvt. Limited, Siemens Limited and Siemens Mobility GmbH. A railway conceived in 2015 around the Japanese Shinkansen system is set to open as a hybrid: Japanese-influenced civil engineering and Shinkansen-derived track technology, an Indian-developed train and a European signaling standard.
That hybrid is easily read as a story of delay or of a partnership gone sour. The record supports a more specific reading. The Mumbai–Ahmedabad High-Speed Rail (MAHSR) project began as India's entry into an integrated technological ecosystem, one in which the train, the signaling, the operating rules and the safety case are designed together and proven together. India's wish to localize has not changed; it was written into the original bargain. What has changed is who holds design authority over the system. That question matters less for one 508-km line than for the national network India now says it wants to build.
The Memorandum of Cooperation (MoC) signed on December 12, 2015 committed India to develop the corridor "with the use of Japanese high speed rail technologies (i.e. the Shinkansen system) and experiences." Japan offered yen loans that could cover up to about 81 percent of the project cost, at 0.1 percent a year over 50 years with 15 years' grace. The same document promised a training institute, instruction for about 4,000 railway officials, and "'Make in India' of high speed rail systems including manufacture of rolling stock," to be promoted in a phased manner. That month the Cabinet Committee on Economic Affairs (CCEA) approved a cost of ₹97,636 crore (one crore is 10 million), about US$14.6 billion at the Reserve Bank of India's reference rate of December 11, 2015. The money followed in tranches. The fifth, a ¥400 billion loan signed on December 21, 2023, kept the 0.1 percent rate and tied procurement. The Japan International Cooperation Agency (JICA) called it the largest single project loan agreement in its history. It brought JICA's cumulative commitment since 2017 to ¥1.05 trillion, about ₹64,000 crore or US$6.6 billion at reference rates published by Financial Benchmarks India Private Limited (FBIL) for October 6, 2026.
The bargain frayed first where integration is tightest: the train. In July 2023, the National High Speed Rail Corporation Limited (NHSRCL) invited bids for 24 trainsets based on Japan's E5 series, and under the loan's terms only Japanese manufacturers were eligible. Indian officials then sought changes for 50°C heat and dust, and lower prices. Isao Tsujimura, a Delhi-based Japanese railway engineer, has written that with only two makers able to build the E5, procurement became effectively single-source, and the quoted prices diverged from Indian expectations. Then the train itself began to disappear. According to the Railway Ministry's Cabinet note as reported in September 2026, Japan said in December 2024 that E5 production was being phased out. East Japan Railway Company (JR East) began designing the successor E10 in March 2025. It plans commercial service in Japan in its 2030 fiscal year, which runs from April 2030 to March 2031. The August 2025 summit joint statement recorded India's appreciation of "Japan's offer to introduce, in the early 2030s, the E10 series of the Shinkansen that runs on the Japanese signalling system." Both sides also agreed to start work on early installation of signaling "including the Japanese Shinkansen system."
That offer left India with a sequencing problem that diplomacy alone could not solve. The Ministry of External Affairs (MEA) has said publicly that Japan's proposed E10 trains would be available only in the early 2030s because the train "is still under development." Indian officials put the realistic date at no earlier than 2034. Meanwhile, the civil works raced ahead. The ministry told the parliamentary Standing Committee on Railways that all 1,389.5 hectares of land had been acquired, after delays in Maharashtra that held the project back until 2021. The official target for first service is now August 2027¹. On that estimate, waiting for the E10 would have left the completed Surat–Vapi infrastructure without trains for more than six years. The ministry's decision to start operations with the B28 on Surat–Vapi¹ was a sequencing decision before it was a statement about sovereignty.
Chart 1's Caption: The civil works will be ready years before a Japanese E10 is expected to reach India, which is why the first section will open with the BEML–ICF-developed B28 trainsets. Sources: 2015 agreement, BEML and Siemens filings, JR East, joint statements, Lok Sabha answer of July 29, 2026, PIB factsheet, reported official statements.
The delay deserves a calibrated reading. Masafumi Shukuri, chairman of the International High-Speed Rail Association (IHRA), said in October 2026 that not even present-day Japan could build a 500-km high-speed line within 10 years, and he pointed to land acquisition and the COVID-19 pandemic. He called India's own rolling stock "absolutely fine" provided operations are stable, safe and reliable, but warned that training people on time and running enough trials matter alongside procurement. The warning has force. The B28 contract envisaged delivery by the end of 2026. Testing is now reported for May to July 2027; the government's target remains August 2027, although recent reporting has put commercial operation as late as December 2027. For the first BEML–ICF-developed train intended to operate initially at 249 km/h, that leaves little margin for testing, certification and operational readiness.
Engineering explains why the train decision carried the signaling decision with it. Under the Shinkansen's digital automatic train control (DS-ATC), on-board equipment searches braking patterns stored in a database and issues braking commands, which moves intelligence onto the train and reduces trackside equipment. The design is efficient because the train and the control system are developed together: the braking model is tightly integrated with the characteristics and performance envelope of the train. ETCS approaches the problem through a standardized, multi-vendor architecture. The European Union Agency for Railways (ERA) describes the wider system as a single standard intended to ensure interoperability and to cut the cost of buying and maintaining signaling. DS-ATC optimizes one operator's tightly integrated fleet; ETCS is designed to support a multi-vendor market around a common standard. Tsujimura's account pinpoints the friction: India wanted consistency with international standards and third-party certification, and a system validated by decades of Japanese operation was never built to supply that kind of evidence. He reads the January 2025 tender as making ETCS the line's permanent system, and he argues that two signalling systems cannot practically share the same line without undermining the integrated operating architecture.
That is the core of the Japanese concern. On this view, a Shinkansen is less a train with parts than a safety record produced by one integrated system, and splitting the system gives up what made the record possible. Tsujimura called the outcome an own goal by the Japanese government, faulting Tokyo for not contesting the switch. Hideki Makihara, a former justice minister, went further and blamed the Indian side entirely. The MEA rejected that as an "individual opinion and at considerable variance with facts." It added that the signaling had been ordered in line with international specifications and that "No Japanese offer was received in this context." The disagreement has not developed into an overt bilateral dispute. During Prime Minister Sanae Takaichi's visit, the July 2026 joint statement said Japan "fully understands India's target to commence commercial operations on priority sections in 2027," and both sides "acknowledged the goal of introducing the E10." Between the two statements, the reference to Japanese signaling disappeared.
Whether the E10 will ever run on this corridor is unresolved. Officials have said that trains offered by Japan's Hitachi are compatible with ETCS and that India is open to a switchover to Japanese protocols once suppliers make proposals. The ministry told the Lok Sabha in August 2026 that the technical specifications and schedule of the E10 proposal "are to be worked out." The latest procurement points the other way. On September 18, 2026, BEML disclosed an order worth more than ₹5,400 crore from NHSRCL to supply and maintain high-speed rolling stock for the corridor. Reporting has linked the award to a larger follow-on fleet, but the BEML disclosure itself does not specify the number of trainsets or their final technical configuration. That evidence tests the stopgap framing directly. The Indian train is moving from a stopgap prototype toward the corridor's principal domestically supplied rolling-stock option, while the E10 is now something to be "examined" once it becomes available.
Technology transfer is where the gap between the 2015 promise and the 2026 record is easiest to measure. The ministry's account to the Standing Committee lists real gains: the full-span girder-launching machinery, most of the materials for Japanese slab track, the track machinery and the seismic stoppers once planned as imports are now made in India, and about 1,000 engineers and skilled workers have been trained in Japanese track methods. The same document records the limits. Details of dynamic analysis "not made available by the principal designer" had to be developed independently with the Indian Institute of Technology (IIT) Bombay. Dynamic-loading parameters for Japanese-designed long-span steel truss girders were "not shared," and IIT Kanpur instrumentation is being used to validate them, because Indian Railways' own standard designs are rated for only 160 km/h. Only 74 people had completed training in Japan by early 2026, against more than 3,000 needed for full operations. None of this implies bad faith; proprietary design knowledge is the commercial core of any supplier and seldom travels with a loan. The record therefore suggests that India acquired the ability to build to Japanese specifications, rather than the full reasoning behind those specifications, and that reasoning is what a network designer needs.
On cost, the Railway Board chairman said in January 2026 that the project's estimated cost had risen about 83 percent to ₹1.98 lakh crore (one lakh crore is ₹1 trillion), about US$20.5 billion at the FBIL rate for October 6, 2026. That 83 percent is measured against the ₹1.08 lakh crore figure long cited for the project; against the ₹97,636 crore sanctioned in 2015, the increase is about 103 percent. The Cabinet reportedly approved more than ₹2 lakh crore in September 2026. No official release confirming that figure could be found as of October 7, and the ministry told Parliament in July 2026 that the final cost can be ascertained only after all works are complete. The ministry's itemized list to the Standing Committee is more revealing. Taxes and cess account for ₹29,330 crore and inflation since 2015 for ₹19,084 crore. Together, they account for about 55 percent of the itemized cost-escalation components listed by the ministry. Indigenous rolling stock and ETCS signaling together account for ₹16,500 crore, under a fifth. On the ministry's own accounting, localization is a cost driver, but not the main one.
The headline figure also blends cost concepts that behave very differently once design authority shifts.
| Cost concept | What it measures | MAHSR evidence (as of October 2026) | Effect of the shift toward Indian development and open standards |
|---|---|---|---|
| Headline capital cost | Sanctioned project estimate | ₹97,636 crore (December 2015); ₹1.98 lakh crore stated (January 2026); more than ₹2 lakh crore reportedly approved (September 2026) | Indigenous rolling stock and ETCS: ₹16,500 crore of listed cost escalation, behind taxes and inflation |
| Financing cost | Interest, tenor and currency of borrowed money | JICA loans at 0.1 percent over 50 years, tied; cumulative ¥1.05 trillion, roughly a third of the revised cost; increase to come from the budget (reported) | A greater share of project financing shifts toward rupee funding, reducing yen-denominated repayment exposure while potentially carrying a higher domestic financing cost. |
| Procurement cost | Price paid for systems and trains | E5: Japanese makers only; ETCS: ₹4,140.06 crore winning bid, rival bid about ₹12,700 crore (reported); B28: ₹27.86 crore per car | Competition widens, but first-of-kind prices and risks fall on Indian buyers |
| Lifecycle and development cost | Design, tooling, maintenance and integration over decades | B28 contract of ₹866.87 crore includes design and tooling for future projects; ETCS contract includes 15 years of maintenance | Up-front costs pay off only if spread across a fleet and a network |
The JICA loan was concessional on any measure: on October 5, 2026, the 10-year Government of India benchmark yielded more than seven percentage points above its 0.1 percent rate. The concession came with tied procurement and currency exposure, and the rupee cost of a yen has risen about 8 percent since the December 2023 tranche. The cumulative commitment now covers roughly a third of the revised estimate, against a ceiling of about 81 percent of the original. Which packages the yen loans will finance after the switch has not been publicly specified. By subtracting the disclosed per-car figure from the total B28 contract value, roughly ₹421 crore appears attributable to the contract's design, development and tooling components, which the company says will be used "for all future High-Speed projects in India." Such costs become economically more defensible when spread across a larger fleet and network.
Scale is the real subject. A Press Information Bureau (PIB) factsheet lists seven corridors announced in the Union Budget for the 2026-27 fiscal year, spanning nearly 4,000 km with an estimated investment of ₹16 lakh crore, which is ₹16 trillion or about US$165.9 billion at the FBIL rate for October 6, 2026. They are announced, not sanctioned: in March 2026, detailed project reports (DPRs) for five awaited approval and three were in preparation, and any sanction depends on the DPRs, techno-economic feasibility and financing. None had been sanctioned by early October 2026. Yet the design choices are already being made. NHSRCL is preparing a standardized set of designs and technical specifications for future corridors, train control systems included. The ministry has reportedly told the Cabinet that it will develop a 350 km/h B35 train by 2030 and its own Bharat Train Control System (BTCS). Precedent suggests that ownership comes from repetition across a network, not from a single line. After Alstom's transfer to Korea, 34 of the first 46 Korea Train Express (KTX) trains were built locally. The World Bank's account of China's early high-speed rail development similarly describes a progression from imported or licensed technology toward local adaptation and standardization. On that reading, MAHSR is less the network's first segment than its demonstrator.
Two coherent industrial philosophies meet here, and neither deserves caricature. The Japanese model binds operator, manufacturer and signaling supplier in a long, incremental relationship, proves safety through operating history and treats the system as indivisible. The model India is increasingly choosing splits the system at standardized interfaces, buys each layer competitively, proves safety through third-party certification and uses domestic volume to build an industrial base. The first can deliver reliability early, but at the price of greater dependence. The second offers flexibility and control at the price of carrying integration risk at home. A country planning one line might prefer the first; one planning thousands of kilometers has reasons to prefer the second.
That leaves a paradox. By the logic of technology transfer, the partnership's success would be measured by how little India eventually needs Japan. The paradox need not be adversarial. The July 2026 statement records Prime Minister Narendra Modi inviting Japanese companies to take part in the future corridors, and both sides spoke of a 7,000-km national vision and private-sector-led cooperation. The relationship appears to be moving from the sale of a closed system toward roles for Japanese firms as suppliers, consultants, investors and, possibly, providers of E10 trains, within an architecture that India is coming to define.
The strategic risk runs both ways. The E5 episode showed how a buyer that depends on one integrated foreign system can be stranded when the supplier's own product cycle moves on. Open standards reduce that risk; they do not eliminate dependence, since the ETCS consortium itself includes Germany's Siemens Mobility. What open standards change is India's ability to choose and replace suppliers. The opposite risk is just as real. Indian institutions now bear responsibility for the interfaces between an Indian train, European signalling and Japanese-derived civil and track technology, and with them the safety case for the whole. A national BTCS would raise the stakes further, because a domestic standard is only as strong as the institutions that certify it. A serious early failure would weaken the case for the entire 4,000-km plan.
The markers to watch are an official statement of the revised cost, B28 trial performance in mid-2027 and whether the August 2027 target survives, the design and size of the serial fleet, whether an E10 offer arrives with specifications that fit an ETCS line, and which signaling and rolling-stock standards the first approved DPRs adopt. MAHSR is at once a railway project, a technology-transfer experiment, an industrial-policy experiment and a test of whether a partnership can survive a change in who sets the rules. The engineering on Surat–Vapi is nearly done; the architectural choice is not. What kind of high-speed railway does India want to own when the first train finally leaves the station?
Footnote:
¹ Government communications in July 2026 have referred to both Surat–Vapi and Surat–Bilimora as the initial operating section; the later PIB factsheet identifies Surat–Vapi.
Apple News, Google News, Feedly, Flipboard, and WhatsApp Channel
