Tech Tactics

India quietly updated its Special Economic Zones (SEZ) policy in June 2025, a subtle but important shift in New Delhi’s approach to high-tech manufacturing. The core change – for sectors like semiconductors and electronic components – is that minimum land requirement for setting up an SEZ has been slashed from 50 hectares to just 10 ha. On paper, it’s a procedural tweak. But in reality, it acknowledges an industry truth: capital-intensive, long-gestation tech sectors don’t fit neatly into legacy frameworks built for textile exports and assembly lines.

There’s more flexibility, too, in how land is counted. Even mortgaged or leased land held by government entities can be approved. The shift signals intent to make space for businesses likely to shape the next decade. SEZ-housed units in these sectors can also sell into the domestic market (after paying duty), and goods supplied for free are now counted in Net Foreign Exchange (NFE) calculations, a recognition of how supply chains actually work in modern manufacturing.

India isn’t the only country reshaping its SEZ playbook. Globally, these zones have evolved from simple export enclaves into full-fledged platforms for strategic investment. Whether Vietnam’s Hoa Lac Hi-Tech Park, Rwanda’s Kigali Innovation City, or Saudi Arabia’s NEOM Oxagon, SEZs act as semi-autonomous launchpads where policy ambition meets investor caution. In 2023 alone, greenfield foreign direct investment (FDI) in high-tech sectors topped $150 billion, much of it funneled through SEZ-style projects spanning semiconductors, Electric Vehicle (EV) batteries, climate tech, and AI. It’s a reminder that SEZs, when well-executed, can offer the one thing both states and investors want control without rigidity.

The old model of SEZs — cheap labour, generous tax holidays, and bare-bones customs clearance — was built for a different era. What matters today is Intellectual Property (IP) protections, arbitration mechanisms, reliable utilities, and legal predictability. China’s Shenzhen–Hong Kong innovation zone, for instance, runs two legal systems to attract global AI labs and chip startups. Rwanda’s Kigali Innovation City anchors its contracts in international law to reassure skittish investors. Taiwan’s Hsinchu Science Park, focuses on minimising “time-to-manufacture” with climate-controlled cleanrooms and dedicated fab transport within a tight radius.

SEZs are becoming as much geopolitical instruments as economic ones. They’re used to trial reforms, onboard sensitive industries, and even shift diplomatic dependencies—sometimes without needing to say so out loud. Vietnam, for example, has quietly attracted Taiwanese and Korean suppliers via its SEZs, softening its dependence on Chinese electronics without provoking open confrontation.

Of course, this isn’t a fairytale. SEZs can fail — and often do — when ambition outpaces execution. Nigeria’s Calabar zone promised an ICT cluster but never fixed its power grid. Russia’s Skolkovo Innovation Center collapsed under geopolitics and internal trust deficits. And valid concerns remain about SEZs becoming loopholes — bypassing regulations, enabling elite capture, or undermining labor protections. Especially as high-tech SEZs host AI and data infrastructure, the line between economic sandbox and surveillance outposts can blur quickly.

Still, India’s recalibration of its SEZ norms, especially for semiconductors and electronics, is a deeper awareness that in an age of supply chain shocks, techno-nationalism, and industrial decoupling, growth needs spaces protected and experimental. SEZs, when done right, offer just that. They’re not just tax-free land parcels anymore – they’re where a country quietly rehearses its future.

On June 25, 2025, India took another step towards the milestone of indigenous human spaceflight capability with the successful launch of Axiom Mission 4 (Ax-04), which included IAF Group Captain Shubhanshu Shukla as Mission Pilot aboard SpaceX’s Dragon spacecraft. The mission lifted off from Launch Complex 39A at NASA’s Kennedy Space Center in Florida and docked with the International Space Station (ISS) the following day. This marked the first time an Indian astronaut set foot on ISS. Before Ax-4, the only instance of an Indian national — then Squadron Leader Rakesh Sharma — being in an orbital installation was onboard the smaller unilateral Soviet space station Salyut 7.

Shukla was selected from ISRO’s Gaganyaan astronaut pool and could also be part of future Indian space flight missions. His training at NASA’s Johnson Space Center, SpaceX’s simulators, and Axiom Space facilities, covered spacecraft operations, space medicine, survival skills, and payload deployment. During his stay on the ISS, Shukla executed a suite of scientific experiments curated by ISRO in partnership with India’s premier research institutions. Three experiments that focused on microgravity impacts:

1. Testing edible microalgae behaviour under space conditions indicates a drive towards developing regenerative source of sustenance for spacefarers.
2. Testing germination of salad seeds to explore closed-loop nutrition systems. This was an advanced stage experiment on method, more than an attempt to develop a food source for spacefarers.
3. Influence of metabolic supplements on muscle regeneration in microgravity. The most consequential of the five themes, the impact of muscle atrophy from extended stay in orbit was under observation. How to retain muscle mass in zero gravity has been researched, but regeneration, especially under targeted dosing of anabolic amino acids, is a less studied area.

The mission included Earth observation during an overpass over India. To recall, Rakesh Sharma’s phone call from orbit to Indian PM Indira Gandhi resulted in the iconic “Sare Jahan se Accha Hindosta Hamara” moment which captured national pride. Group Captain Shukla’s call with PM Modi had similar symbolism.

DRDO-industry-academia Centre of Excellence (DIA-CoE) at IIT Delhi has demonstrated secure entanglement-based quantum communication over a free-space optical link. The experiment by Professor Bhaskar Kanseri’s research team is a significant advance in indigenous quantum communication technology.

The team established a quantum-secure link over a distance over 1km in IIT Delhi campus using entangled photon pairs transmitted through air rather than fibre optics. Entanglement means a shared property — such that any attempt at measurement or interference or eavesdropping is immediately detectable, and disrupts the quantum state alerting users. The experiment achieved a secure key rate of nearly 240 bits per second and maintained a quantum bit error rate below 7%. These parameters represent a reliable performance.

The technology enables unbreakable encryption for defence, finance, telecom, and national command infrastructure while avoiding costly and often infeasible fibre infrastructure, advancing a practical satellite-based quantum networks. Ministry of Electronics and Information Technology (MeitY) has also published a whitepaper detailing a roadmap for cybersecurity in the inevitable quantum era, which should serve as an early theoretical and conceptual framework upon which best practices and norms could later evolve.

Most large language models (LLMs) function as task-specific user agents, including OpenAI, and also India’s emerging platforms such as Krutrim’s Kruti. These systems summarise emails, generate code or media, parse documents, and increasingly serve as intermediaries between humans and digital systems. They have no subjective awareness, continuity of self, or any form of agency beyond probabilistic pattern recognition.

What pathway might lead from these tools to something closer to sentient AI, and what technological, economic, or policy developments could accelerate or constrain that journey?

Extending context memory and enabling persistent embodiment may be a starting point. Current models operate in discreet prompt windows and lack the ability to retain long-term knowledge of users or tasks. Developing goal-oriented agents that can memorise and recall preferences, track evolving identities, and adapt across sessions needs major architectural changes — integration with memory-augmented transformer designs and scalable vector databases. The aim would be to ground these systems in real-world experience to model continuity, causality, and consequence.

Economic incentives will drive this. As personal AI agents become common in education, legal advisory, and productivity management, users will demand consistency, autonomy, and emotional intelligence. Companies will compete to provide agents that feel familiar, context-aware, and attuned to individual needs.

This dynamic could give rise to AI ‘personas’ that modify themselves over time, reinforcing internal preferences, adapting to feedback, and adjusting their own heuristics. These systems may not achieve sentience in a biological sense, but they would approximate functional agency.

Policy frameworks become highly relevant as these systems take on more autonomous roles. Most current regulations focus on transparency, fairness, and prevention from harm. But once AI agents decide on behalf of users or institutions, governments will be confronting questions on legal personhood, liability, and representation.

So, expect AI-generated contracts, co-signed documents, or limited agency within decentralised platforms. Even partial recognition of AI-driven actions in legal systems will signal a redefinition of how intent and accountability are understood.

Sentience is not matter of scale alone but of synthesis. For a model to approximate selfhood, it has to have the ability to reflect on its own outputs, learn from mistakes, and refine its goals. Developments in neurosymbolic reasoning, active inference models, and quantum-enhanced decision-making provide the framework for such capabilities.

Certainly, sentience, if at all it appears, is likely to emerge incrementally. It would manifest first as an expansion in planning depth, emotional inference, or ethical judgment before becoming something recognisable as a continuous self.

Technological singularity becomes plausible only once such systems begin designing new generations of themselves – in model architecture, reward shaping, and resource optimisation. AI systems that improve themselves might become too fast and complex for humans to understand or control. This might make it near impossible to stay aligned with our goals — how to verify goals of an intelligence faster than any external audit can track?

This makes global governance unavoidable. States that control access to frontier models, computational infrastructure, and scalable alignment methods will shape the ethical and operational rules of AI systems worldwide. India can play a decisive role, well positioned to lead in setting democratic standards. A multi-stakeholder approach that distributes oversight across institutions, geographies, and interest groups is the best way ahead.

Not policymakers or academics but a small circle of Silicon Valley venture capitalists, technologists and bloggers are the ideologues, the influential class, that has shaped our approach to breakthroughs in AI, biotechnology, defence automation, and post-liberal governance models, especially since the 2016 US presidential election cycle.

Peter Thiel and Curtis Yarvin’s philosophies are increasingly reflected in how capital flows into high and deep-tech sectors. Peter Thiel’s Palantir, a Big Data company with lucrative US government contracts as its main revenue stream, argues that technological stagnation is not due to scientific limits but political failure.

Yarvin on his substack blog critiques liberal democracy from a more radical angle. He advocates for formalism, a philosophy that favours openly centralised power, ideally administered through technical systems rather than political consensus. For Yarvin, bureaucratic sprawl is not a safeguard but failure mode.

Together, Thiel and Yarvin form an intellectual backdrop for a new model of technological ambition. This model favours small, mission-driven teams supported by concentrated capital, building tools that often operate outside constraints of existing regulatory or democratic frameworks.

Leading tech companies ae reshaping governance around three core principles. First, that concentrated vision and operational control beat democratic process for existential challenges. Second, direct stakeholder involvement creates more accountability than distant representation. Third, technology drives structural change more effectively than politics. From AI safety labs, like Anthropic and Conjecture, to defense contractors like Anduril, and blockchains’ decentralised organisation, this philosophy increasingly defines the intellectual terrain and influences how young entrepreneurs and investors approach the building of systems that bypass traditional institutional frameworks.

The challenge for the global south is how to reconcile the execution speed and ambition of deep-tech with needs of democratic accountability. Can public institutions be retooled to match the agility of sovereign-capital startups without compromising legitimacy? Or will new policy architectures emerge that integrate open-source development, public infrastructure, and state-aligned innovation pathways?

The thinkers behind these ideas influence and shape the building of tomorrow’s foundational technologies. As these tools begin to challenge or bypass legacy institutions, understanding the philosophical software driving their creators is no longer optional. Big Tech and the associated people with capital become major political donors able to sway regulation and legislation in their favour or against their opposition.

1. The “Dark Elf” Leading Tech’s Extreme Right – The Nation
2. Gaganyan Test Vehicle Detailed Specifications – ISRO
3. The special economic zones and innovation: Evidence from China – China Economic Quarterly International
4. Financing Irish high-tech SMEs: The analysis of capital structure – International Review of Financial Analysis
5. Large language models: assessment for singularity – AI & Society
6. Will the technological singularity come soon? Modeling the dynamics of artificial intelligence development via multi-logistic growth process – Physica A: Statistical Mechanics and its Applications
7. The AI Control Problem in a wider intellectual context – Philosophy Bear on Substack

Thank you for reading Tech Tactics. Stay tuned for our next edition.