U.S. Export Control Unpredictability Is Testing the Limits of U.S.

U.S. Secretary of State Marco Rubio signed several ambitious agreements for increased defense cooperation with India during a recent visit to New Delhi. In response to increased Chinese submarine patrols in the region, for example, Rubio and Indian External Affairs Minister Subrahmanyam Jaishankar signed a comprehensive Underwater Domain Awareness roadmap and expanded current initiatives to monitor real-time activity in the Indo-Pacific with their two partners in the “Quad,” Japan and Australia. In a joint press conference, Rubio suggested the United States and India have “a tremendous strategic alliance” and noted the intention to move from simply a defense producer-buyer relationship to one of co-development of advanced military technology, something India has been seeking for decades. While the diplomatic rhetoric may increasingly project a partnership of equals, the regulatory environment tells a different story. Washington’s shift from a rules-based export licensing regime to ad hoc horse-trading in a burdened Commerce Department pushes India’s defense tech ecosystem toward architectural choices that will make these agreements more difficult to implement. Current Cooperation Promises The Biden administration and both Trump administrations have courted India as a critical partner for a wide range of digital and defense cooperation efforts. Despite crushing 50 percent tariffs that have strained the relationship over the past year, Rubio asserted in the press conference with Jaishankar that India was one of the “most important strategic partners” in the world. In 2023, the landmark U.S.-India Initiative on Critical and Emerging Technology (iCET) solidified the beginning of multi-billion-dollar joint projects on AI and cyber defense, including India’s deal for acquisition and domestic assembly of drones from American defense manufacturer General Atomics. Similarly, the TRUST Initiative and INDUS-X defense projects prioritize increased bilateral cooperation on critical and emerging technologies in several domains, including cooperation on sensitive artificial intelligence. These agreements and others are multi-administration commitments to solidify the relationship with India as an ally in everything but name, with faster and deeper access to the U.S. defense ecosystem. Under President Barack Obama, the United States designated India in 2016 with the bespoke moniker of “Major Defense Partner,” elevating its trade authorization two years later to receive “license-free access to a wide range of military and dual-use technologies regulated by the Department of Commerce.” But the partner designation is a technology transfer grey area, as defense partnerships are more flexible and transactional than agreements with formal allies. Note the hedging language in the White House announcement: “The United States will continue to work toward facilitating technology sharing with India to a level commensurate with that of its closest allies and partners.” (emphasis added) Relationships with traditional allies like those in the AUKUS agreement — Australia and the U.K. — come with a higher level of trust, which means that the export controls licensing process generally works quicker and with higher access than it does for lower-level partners. India’s partnership is something in between, which limits access and slows the processing of export licenses. One of the highlights of Rubio’s recent visit to New Delhi focused on increased maritime surveillance cooperation within the Quad, which establishes a mechanism for joint efforts on real-time submarine tracking in the Indo-Pacific. These defense technologies require high levels of real-time intelligence sharing, radar technology, and intelligence, surveillance, and reconnaissance (ISR) collaboration. Regulatory Mixed Signals Under the Biden administration’s AI Diffusion Rule, there was a tiered approach to access to frontier semiconductors and compute required for any sustained high-level AI research. Despite the strong bilateral relationship, India was placed on a restricted access tier, signaling the limits of U.S. trust with sensitive technologies due to national security concerns of American technology falling into the hands of adversaries. Under the Trump administration, the Diffusion Rule was discarded without a replacement strategy. Instead, an ad-hoc review of each license took its place. The Commerce Department’s Bureau of Industry and Security has been hemorrhaging staff responsible for reviewing these licenses, losing 20 percent of their employees in the past year. The licensing regime is also largely controlled personally by the Under Secretary of Commerce for Industry and Security Jeffrey Kessler, with no published criteria. This has compounded bottlenecks and caused significant delays, eliciting complaints from semiconductor industry associations that have seen approvals that used to take weeks now being processed for months. At the same time, while Indian firms are starving for compute access, Trump flew to a meeting with Chinese President Xi Jinping earlier this month with Nvidia CEO Jensen Huang to cut a special deal to increase Chinese access to American semiconductor chips. In Rubio’s visit to India, no such offer was made. Despite being hailed as a critical partner, Washington’s inconsistencies are causing Indian tech innovators to consider ways to reduce their dependencies on U.S.-controlled chips. Indian Firms Pivot The combination of unpredictability in the U.S. export control regime for both semiconductor hardware and compute access with domestic pressures to create a home-grown industry reflecting the values of the Indian government works to create strong incentives for a defensive hedging posture in AI development. Under Prime Minister Narendra Modi, now in office more than 12 years, the domestic messaging on AI has been that India should be a producer of AI technology, not just a consumer market for Western companies. India doesn’t currently have the capacity to build its own cutting-edge Large Language Models (LLMs) like OpenAI’s ChatGPT, which operate using more than a trillion adjustable variables, nor is there the desire to pivot to Chinese models or hardware. In fact, Indian data protection law explicitly prohibits data being stored on Chinese servers. There is clearly an appetite for “sovereign AI” that would be less vulnerable to geopolitical interference in the chip supply chain, as illustrated by the numerous indigenous AI strategy panels at the most recent AI Summit in New Delhi and the announcement of the IndiaAI Mission, which aims to expand compute access and domestic semiconductor production in the country. Where does this leave Indian deep-tech defense firms? Instead of waiting for more access to hardware in an unpredictable and expensive licensing queue, they are building software pipelines that can be trained and run on cheaper, legacy-node hardware rather than U.S.-controlled chips. Currently, users of Nvidia hardware are forced to also use their operating software. To avoid being locked into dependency on American hardware, Indian firms are building more domain-specific Indian models that are built to be “chip-agnostic,” meaning that they can be run on domestically produced chips or older processors. This pivot is not only a domestic political move, but also a technical necessity. Practically, however, this means that the software-heavy systems the two countries are investing in are being deliberately decoupled from U.S. hardware. Interoperability Challenges If Indian firms continue investing in chip-agnostic software as a hedge against U.S. export unpredictability, this presents several challenges for the United States in executing the ambitious defense cooperation with New Delhi. The potentially growing divergence between U.S. and Indian advanced computing hardware and software has the potential to cripple the interoperability required for complex joint ISR and maritime drone operations. Maritime drone systems, in order to operate well as a swarm, rely on sensor information that changes by the millisecond and can’t afford to be buffering in translation with other software systems. For joint development of automated cyber defense systems, the U.S. zero-trust model will reject plugging into a cyber defense system that wasn’t built on American software and hardware. The INDUS-X initiative has already begun facilitating co-production of maritime autonomous systems between Indian and U.S. defense firms to enable more effective submarine tracking. However, if the software foundations of these systems continue to diverge, we may end up with maritime drone swarms unable to operate smoothly with each other in the water during a real-time Indo-Pacific contingency. These challenges persist at every level of the new cooperation agreements: the Underwater Domain Awareness roadmap, Quad maritime surveillance efforts, cyber defense initiatives, and co-development of advanced military technology. In order to ensure that U.S. defense cooperation with India stays on track, it is critical that the two countries ensure that their tech stacks are also moving forward together. Paths Forward Rubio’s visit underscored the genuine desire to keep India engaged in supporting American interests in the Indo-Pacific, but the lack of clarity and predictability is creating more of a wedge than a bridge. Quicker and more transparent rules for export controls could slow the Indian pivot away from the heavy reliance on U.S. technology. Recently introduced in Congress, the Bureau of Industry and Security (BIS) Licensing Efficiency Act of 2026, aims at increasing speed and transparency in license administration at the Commerce Department, which could potentially decrease some of the hardware access issues for the Indian market. At the very least, it would signal a consistent standard that partners could rely on rather than forcing them to read tea leaves while China cuts deals for more access. More broadly, it may be time to address the regulatory grey area of India as a “Major Defense Partner” even beyond the trade partnership. This designation attempts to hold India’s importance and nonalignment at once, but has never been fully defined in terms of what the partners can expect of one another. As the scope of the U.S.-India defense cooperation agreements continue to expand into one of co-development of autonomous systems and high-sensitivity ISR cooperation, this regulatory ambiguity is becoming increasingly untenable. The United States can’t expect India to build its defense future on American technology while treating semiconductor and compute access like a trade bargaining chip.