#75 Conducting Cosmic Commerce: NSIL's Starry Rise
NewSpace India Limited - Taking stock before forging ahead, Mistral Tap dances around the EU AI Act
Today, Ashwin Prasad comprehensively examines reforms leading to private-sector engagement in India’s space sector.
Rijesh Panicker follows it up with a brief rumination on what Mistral AI’s newly announced partnership with Microsoft means for its exemptions under the EU AI Act.
Also,
We are hiring! If you are passionate about working on emerging areas of contention at the intersection of technology and international relations, check out the Staff Research Analyst position with Takshashila’s High-Tech Geopolitics programme here. For internship applications, reach out to satya@takshashila.org.in.
Antariksh Matters: NewSpace India Limited - Taking stock before forging ahead
— Ashwin Prasad
During the 20th Century, national space agencies laid the foundation for space power around the globe. A burgeoning wave of new applications in the 21st century has pushed space-faring nations to engage private companies in the sector. India also embraced this trend through its space sector reforms that began with the inception of New Space India Limited (NSIL) in 2019 and the Indian National Space Promotion and Authorisation Centre (IN-SPACe) in 2020.
Creation
The government created NewSpace India Limited (NSIL) as a Central Public Sector Enterprise under the Department of Space (DoS). NSIL was well placed to exploit the commercial opportunities in the space sector by taking advantage of the Indian Space Research Organisation’s (ISRO) technological capacities through the private sector. This would also foster the growth of auxiliary industries and infrastructure in India, resulting in massive positive externalities.
Yet, NSIL was not the first governmental endeavour in the realm of commercial space ventures. Incorporated in 1992 as a commercial and marketing arm of ISRO, Antrix has many similarities with NSIL: both were incorporated as wholly government-owned companies; both are under DoS; and, both seek to tap into the market applications of space. In fact, two of the initial directors of NSIL, Suma Devaki Ram and Radakrishanan Durairaj were moved to NSIL from Antrix.
All this begs the question: why didn’t the government simply expand Antrix’s mandate instead of creating another company? One possible explanation comes from people within the DoS who perceive that the government felt a need to start afresh after Antrix’s reputational damage and legal liabilities associated with the Devas deal controversy. The fact that most of Antrix’s business segments, including transponder leasing, launch services and mission support services, have been quietly transferred to NSIL gives further credence to this reasoning.
Function
NSIL has been active since its creation with a wide variety of projects. To assess its performance, a detailed examination of its functions is necessary. Initially, in 2019, NSIL’s mandate was limited to producing the Polar Satellite Launch Vehicles (PSLV) through industry, as well as overseeing technology transfer and space-based services. The 2020 reforms that created IN-SPACe also enhanced NSIL’s mandate. NSIL was now tasked with capital-intensive functions to take over the operational part of ISRO’s missions gradually. In addition to its original functions, NSIL would now build, launch, own and operate satellites as per market demand.
PSLV production through industry is considered one of the primary mandates of NSIL. It has found an industry partner in a consortium formed by Hindustan Aeronautics Limited (HAL) and Larsen & Toubro for the end-to-end production of five PSLVs. The process took three years, from the issuing of the Expression of Interest for this project in 2019 to the signing of the contract in 2022. The first of the 5 PSLVs will be delivered to NSIL by 2024. The fifth is slated to take until 2027.
Around 30 technology transfer agreements have been signed with the industry as of 2023. Notably, the IMS-1 satellite bus technology developed by the U R Rao Satellite Centre has been transferred to the industry.
As of last year, NSIL facilitated the launch of 62 international and two domestic commercial satellites aboard ISRO’s PSLVs, including four dedicated launches. The customer base included the USA, Lithuania, Luxembourg, Singapore and Brazil. NSIL also facilitated the dedicated launches of 72 OneWeb satellites on board two LVM3 rockets - one of the largest commercial orders for ISRO and the first use of LVM3 rockets for commercial use. This is set to continue thanks to the MoU signed between NSIL and Arianespace, kicking off a partnership to meet the global heavy satellite launch needs using LVM3 and Ariane-6.
In 2022, NSIL’s GSAT-24 was launched on an Ariane-V rocket. Fully funded by NSIL, ISRO built this Ku-band communication satellite for NSIL. NSIL wholly owns the rocket and is leasing the satellite capacity to Tata Play for its DTH application. Another demand-driven communication satellite - GSAT-20, is scheduled for launch aboard SpaceX’s Falcon-9 in 2024.
NSIL also provides satcom services by providing transponders across various bands from INSAT/GSAT and foreign satellites to Indian users. In 2022, the Government of India approved the transfer of ownership of 10 in-orbit GSAT satellites to NSIL. In addition to all this, NSIL is involved in many other projects and contracts surrounding ground segment and mission support.
Its revenue from operations has grown year-on-year, and so have the profits, as indicated in the chart below.
Therefore, NSIL has consistently tapped into the commercial opportunities in the space sector globally. While its performance so far has been commendable, potential drawbacks are visible on the horizon.
Impediments
Firstly, NSIL has to reconcile its conflicting objectives as both a promoter and a service provider. The technology transfers and growth of the private industry will result in a scenario where private entities become NSIL’s competitors. NSIL already faces such conflicts of interest with companies like Agnikul and Skyroot developing launch capabilities that will put them in direct competition with NSIL’s Small Satellite Launch Vehicle (SSLV) program.
Secondly, NSIL needs to expand its vision if it genuinely wants to spur growth and innovation in the Indian space industry. Its current model of transferring technology and outsourcing critical manufacturing processes to the industry does aid in capacity building and supply chain resilience, among other benefits. However, it also risks limiting the industry to the role of vendors for NSIL. Beyond manufacturing outsourcing, NSIL can provide services directly from the private industry by signing long-term purchase agreements. This will empower the private industry to undertake fully independent projects. With appropriate technology transfer, the purchase agreements can support a culture of independent innovation and help the industry develop end-to-end capabilities without reinventing the wheel.
India is set to host the Quad Leaders' Summit in 2024. Subscribe to Takshashila's Quad Bulletin, a fortnightly newsletter that tracks the Quad's activities through the Indo-Pacific.
Your weekly dose of All Things China, with an upcoming particular focus on Chinese discourses on defence, foreign policy, tech, and India, awaits you in the Eye on China newsletter!
The Takshashila Geospatial Bulletin is a monthly dispatch of Geospatial insights for India’s strategic affairs. Subscribe now!
Cyberpolitik: Mistral Tap dances around the EU AI Act
— Rijesh Panicker
French generative AI start-up Mistral AI announced its latest AI model on Tuesday. Called Mistral Large, this iteration of their large language model is reportedly competitive with the latest ChatGPT offerings from OpenAI. In an interesting turn of events, Mistral has decided that this AI model will not follow an open-source approach and instead offers tiered API access to consumers, following a model similar to OpenAI and others. Given Mistral’s espousal of open source models as a way to compete, this has caught observers by surprise and raises questions about the sustainability and longevity of their open source strategy.
An additional source of surprise was the announcement by Microsoft that they are investing $16 million in Mistral, to be converted to equity as part of the next round of funding. Mistral gets access to the Azure supercomputing resources for training purposes, while Microsoft improves the diversity of its Azure AI offering with models from Mistral. In addition, Mistral and Microsoft have announced that they will be collaborating on specific models for enterprise customers. This is not an exclusive partnership, with Mistral having previously announced that their models would also be available on the Google Cloud platform.
As expected, the European Commission has announced that they will evaluate Microsoft’s investment in Mistral. This is unsurprising, given that the EU is also investigating the Microsoft and OpenAI deal. During discussions for the EU AI Act, Mistral was at the forefront of lobbying last year. In particular, the French government had argued that tight regulation of foundational AI models would deter the growth of homegrown AI firms. As a result of this pushback, exemptions were carved out for open-source AI models in the final EU, which acts as a way to allow companies like Mistral to flourish.
Following on from an investment of $400 million by Andreessen Horowitz in their last round of funding, this new investment from Microsoft raises concerns about the independence of Mistral and the dominance of a few global (mostly US-based firms) in the space. There are also concerns that having lobbied for the exemption as a necessary step towards maintaining independence, Mistral's current deal is in bad faith.
Our View
Mistral’s deal with Microsoft is a rational choice for a startup that wants access to large amounts of compute and the ability to scale up to new consumers. Ironically, this deal allows Mistral to compete with the likes of OpenAI and Anthropic globally, providing a path to becoming the European AI champion the EU wants to develop.
There are two key lessons to take away from this event. First, trying to regulate a fast-moving technology like AI runs the risk of obsolescence, the EU AI Act had no way to regulate generative AI and LLM’s when it was first drafted, and the entire section on regulating so-called foundational models was a bolt-on. Second, carving out exemptions based on the nature of AI models (open source v. closed source) may be a bad idea because it prevents firms from making the most sustainable choices for their businesses.
What We're Reading (or Listening to)
[Opinion] Implications of Chinese fishing boats in Taiwan’s ‘prohibited’ waters, by Anushka Saxena
[Newsletter and Podcast [ChinaTalk] India's Chip War, ft. Jordan Schneider, Chris Miller, and Pranay Kotasthane
[Video] The Geopolitics of Semiconductors ft. Pranay Kotasthane and Shrinivas Gorur