Newspaper Summary 180226

 

India’s sovereign push gathers pace as local firms roll out AI models

USER TO BUILDER. BharatGen, Gnani.ai and Sarvam launch language, voice and real-world AI tools

Sindhu Hariharan New Delhi

India’s push for sovereign AI took concrete shape on Tuesday as BharatGen, Gnani.ai and Sarvam unveiled new models spanning language, voice and real-world interfaces, signalling a shift from being a user of global artificial intelligence tools to building domestic AI infrastructure aimed at government services, healthcare, education and agriculture at population scale. The three players demonstrated that their models are better localised for Indian use cases than global models.

KEY PLAYERS

BharatGen launched its Param2 — a 17-billion parameter model that supports 22 Indian languages. This ‘Mixture of Experts’ (MoE) foundational model is trained extensively on Indian data and is optimised for multiple languages. At the summit, BharatGen is also showcasing sector-focused demos developed with government and industry partners.

Similarly, Gnani.ai launched a voice-to-voice model capable of contextual understanding while preserving emotional tone. Gnani.ai joins a select group of companies worldwide to build this architectural capability. Unlike traditional voice-to-text systems, Inya VoiceOS consumes and generates speech tokens directly, and thus encodes and decodes phonetics, semantics and intent behind the speech. It also handles overlapping speech and mid-sentence corrections.

Besides releasing a slew of AI capabilities across voice, OCR and other products over the last 10 days, Indian AI model maker Sarvam on Tuesday upped it a notch with its own AI-enabled hardware. Sarvam launched AI-powered smart glasses Sarvam Kaze. “Sarvam Kaze moves intelligence from the screen to the real world. You wear it. It listens, understands, responds and captures what you see. And you can build custom experiences for it with the Sarvam platform,” Sarvam Co-founder Pratyush Kumar said in a post on X, teasing a launch as early as May.

MORE CAPABILITIES

Rishi Bal, Executive VP and Head of BharatGen, told businessline that their previous model was about 3 billion parameters, and the one launched now is 17 billion, achieving a 6x increase. “The roadmap will continue with us growing the size and the capabilities of the models as well,” he said.

Further, sovereign AI makers have also shown willingness to grow the entire ecosystem of Indian AI companies by putting their playbook out there. “We will release all of our models in open source. We will release new benchmarks, release new datasets and also recipes for developers to finetune their model,” Bal said.

---

LEVERAGING AI

• BharatGen launches Param2, a 17-billion parameter ‘Mixture of Experts’ model supporting 22 Indian languages.
• Gnani.ai introduces a voice-to-voice AI model.
• Inya VoiceOS works by directly consuming and generating speech tokens, capturing phonetics, semantics and intent.
• Sarvam launches AI-powered smart glasses, which is named Sarvam Kaze.

India, France amend double taxation pact; Modi, Macron launch ‘Year of Innovation’

Aneesh Phadnis Mumbai

India and France will forge a special global strategic partnership and boost trade and investments, as announced by Prime Minister Narendra Modi and French President Emmanuel Macron on Tuesday. During their bilateral discussion in Mumbai, the two countries cemented their long-standing relations by signing multiple agreements in the fields of defence, health, critical minerals, and taxation.

TURNING POINT

Prime Minister Modi described 2026 as a turning point in India’s relationship with Europe, specifically noting the signing of the Free Trade Agreement with the European Union last month. On Tuesday, both nations amended the protocol on their double taxation avoidance agreement, a move Modi stated will boost bilateral trade and investments.

The leaders also virtually inaugurated the final assembly plant for H125 helicopters, a joint venture between Airbus and Tata Advanced Aerospace Systems located in Karnataka. Following this, they launched the ‘India-France Year of Innovation,’ which will focus on AI, digital technologies, and start-ups.

France remains one of India’s closest allies, with a strategic partnership spanning nearly three decades. “Relations between India and France are special. France is one of India’s earliest strategic partners. Together with President Macron, we have brought unprecedented energy to these relations,” Modi said while announcing the upgraded partnership.

SHARED AMBITIONS

President Macron remarked that Indo-French relations are built on "openness, trust and ambitions," adding that both nations believe in a free Indo-Pacific and oppose a hegemonic order. He highlighted ongoing cooperation in defence and space, noting that France supports technology transfer to India.

Piyush Srivastava, Additional Secretary in the External Affairs Ministry, confirmed that discussions covered civil nuclear cooperation, counter-terrorism, and defence, with a newly strengthened focus on people-to-people ties.

BEL-SAFRAN JV

Key security agreements included a pact on the reciprocal deployment of defence personnel at each other's units. Additionally, a joint venture between Bharat Electronics Ltd (BEL) and Safran was announced to produce Hammer missiles. Other collaborative initiatives launched include:

• The Indo-French Centre for AI in Health.
• A Centre of Digital Science and Technology.
• A National Centre of Excellence for Skilling in Aeronautics.

Resource adequacy planning critical for hyper scalers, say experts

Our Bureau New Delhi

As India readies to scale up its AI data centre ambitions, with capacity expected to reach 6-8 gigawatts (GW) by 2030, government officials and industry leaders stressed on comprehensive resource adequacy planning, meticulous modelling exercises coupled with long-term firm power availability, and regulatory certainty.

ENERGY DEMAND

The shift is driven by the fact that AI-enabled servers consume significantly more electricity; for instance, creating a single image with generative AI uses energy equivalent to charging a smartphone. According to the International Energy Agency (IEA), power consumption by data centres has already grown 12 per cent annually over the last five years.

Understanding future requirements will require various scenario-based models to offer perspectives on timelines relevant for decision-making in the energy sector. Participating in a panel at the India AI Impact Summit, Grid India CMD Samir Chandra Saxena noted that at the current rate of growth, the penetration by 2030 will likely be equivalent to a state’s total demand, roughly 8-10 GW.

GRID OBLIGATIONS

From a grid operator’s perspective, Saxena emphasised that with such large chunks of demand, data centres must be obligated to meet resource adequacy requirements themselves, rather than relying solely on distribution utilities. The CEO of the distribution utility BSES also stressed the need for stable, long-duration electricity-generation sources.

POWER SOURCES

Because scaling up nuclear power is still some time away for India, hyper scalers need to leverage multiple sources, including captive generation alongside grid-connected power.

Experts warned that the location of these centres must be pre-decided and planned. Without such coordination, the rapid expansion could lead to chaos, suboptimal organisation, and an unnecessary tariff burden on end consumers due to the cost of adding extensive new wiring and infrastructure.

Army used AI tool to anticipate Chinese aggression in Arunachal: Lt Gen Rana

Dalip Singh New Delhi

India successfully used small artificial intelligence (AI) software to anticipate and repulse an unprecedented Chinese attempt along the Line of Actual Control (LAC) in Yangtse, Arunachal Pradesh, Lt General Dinesh Singh Rana, Commander-in-Chief, Strategic Forces Command, said on Tuesday. Lt Gen Rana emphasised the broader role of AI in reshaping war games, strategic doctrines and the national security ecosystem during his address at a session at the India AI Impact Summit 2026.

AI FORESIGHT

He recalled that during his earlier stint as General Officer Commanding of 4 Corps (Gajraj Corps) in Arunachal Pradesh, hinting at the December 2022 physical face-off between the Indian Army and Chinese PLA, they used small AI software with low capability to analyse an emerging situation, triggered by an unprecedented attempt made by China. That helped the Army to predict the timing of Chinese soldiers’ approach in Yangtse — a disputed territory held by India — and successfully deploy troops in the region close to the LAC to thwart Chinese ingress and prevent casualties, he told the audience, highlighting the low-hanging fruit of AI-powered prediction tools.

TECH SHOWCASE

The Army also showcased indigenous dual-use AI technologies at the AI Summit, which, it stated, can deliver significant value in education, disaster response, cybersecurity, transportation safety and digital infrastructure protection.

For instance, “AI Examiner” is an AI-enabled automated evaluation system that can be integrated into any learning management system for creating assessments, analysing submissions, generating scores and providing structured feedback to trainees, the Army said.

It has also come up with SAM-UN (Situational Awareness Module for UN Operations), an indigenous web-based platform that integrates geospatial intelligence, structured reporting, and AI analytics to provide unified real-time situational awareness for mission planning, monitoring, coordination and rapid response. This is ideally suited for disaster management centres, emergency coordination hubs, and smart city control rooms.

Similarly, “EKAM” is an AI-as-a-Service tool, which is a secure, modular, indigenous, air-gapped cloud platform providing unified access to AI chat, document and presentation generation, multilingual translation, summariser, RAG and specialised bots for multimedia generation. It ensures access to compute resources while maintaining full data sovereignty.

WEATHER INTELLIGENCE

Another solution is PRAKSHEPAN, an AI-driven military climatology and disaster prediction system. PRAKSHEPAN is India’s first hybrid military climatology decision-support system capable of predicting landslides, floods, and avalanches three to seven days in advance using multi-agency scientific datasets, terrain intelligence, and AI/ML-based modelling.

Developed in collaboration with national scientific agencies and the Ministry of Earth Sciences, it provides highly accurate and location-specific hazard forecasts for sensitive operational regions, the Army informed. Among other tools developed are: Facial Recognition System, Driver Fatigue Detection, Vehicle Tracking System, and Deepfake Video Detection System.

Eternal announces strategic tie-up with OpenAI

Meenakshi Verma Ambwani New Delhi

Eternal Ltd and OpenAI on Tuesday announced a strategic collaboration to advance artificial intelligence (AI) capabilities across Eternal’s diverse business ecosystem. The partnership will encompass platforms such as Zomato, Blinkit, District, and Hyperpure, as well as its AI-native venture, Nugget. The initiative will focus on deploying AI across Eternal’s applications, partner platforms, and internal systems.

PLATFORM INTEGRATION

As part of this collaboration, Eternal is considering the integration of OpenAI’s latest coding models, including GPT-5.3-Codex, within Stitch, its proprietary developer orchestration and automation platform.

The company will also leverage OpenAI’s Enterprise API platform to:

• Explore new ways for customers and partners to interact with Eternal’s platforms.
• Deploy advanced AI tools within partner applications.
• Experiment with next-generation search and discovery interfaces.

“From high-leverage areas like software development to customer-facing innovation, we are happy that this collaboration with OpenAI will open up even more surface area for us to learn and innovate,” said Albinder Dhindsa, Group CEO of Eternal.

AI DEPLOYMENT

Eternal is strengthening its AI infrastructure to build AI as a foundational layer across its entire commerce ecosystem. The collaboration intends to deploy OpenAI models for specific use cases, including AI-assisted workflows for merchants and delivery partners and contextual AI assistants embedded within partner portals.

For Nugget, OpenAI’s models will be used to accelerate product development and iteration. Oliver Jay, Managing Director, International at OpenAI, stated, “Eternal operates at a meaningful scale across consumer and partner platforms. We are excited to support their teams in applying AI advancements to real-world systems, from AI-native ventures to partner-facing initiatives”.

UPSKILLING INITIATIVE

Furthermore, the two entities will explore a structured Partner Upskilling Programme designed to drive AI adoption among Eternal’s vast network of restaurant and delivery partners. This programme will focus on deploying advanced tools to support operational efficiency, compliance, and business decision-making.

British International Investment’s climate commitments surpass $1 billion in India

Our Bureau New Delhi

British International Investment (BII) announced on Tuesday that it has exceeded its commitment to invest $1 billion in climate finance in India during the 2022–2026 strategy period. Coinciding with the start of Mumbai Climate Week, the UK’s development finance institution (DFI) confirmed that its cumulative climate investments in the country have reached $1.1 billion.

NEW EV INVESTMENT

As part of this milestone, BII announced a ₹43 crore investment in Turno, an India-based electric vehicle (EV) battery infrastructure company. The funding is designed to support the launch of ElectricGo, which is Turno’s new business unit focused on e-buses. This investment builds on an initial commitment made in 2024 to expand support for India's evolving EV sector.

STRATEGIC TRANSITION

“India is central to BII’s climate investment strategy. Reaching $1.1 billion in climate investments reflects both the scale of India’s opportunity and BII’s long-term commitment to supporting its transition,” said Shilpa Kumar, Managing Director and Head of India at BII. She noted that partnerships in areas such as clean mobility, smart metering, and agri-technology illustrate how climate finance can drive innovation, inclusive growth, and economic resilience.

India is working toward a goal of reaching net-zero emissions by 2070, with transitions in clean energy, electric mobility, and sustainable agriculture serving as key pillars.

PORTFOLIO MILESTONES

BII highlighted several recent expansions within its green portfolio:

• GreenCell Mobility: Planning the deployment of 570 e-buses in Delhi to expand zero-emission public transportation.
• EnerGrid: Following a $110 million commitment from BII, the firm commissioned India’s first standalone utility-scale Battery Energy Storage System (BESS) with 360 MWh of capacity.
• Fasal: An agri-tech firm that has launched precision-automation tools to support farmers across India and South-East Asia.

According to the institution, over 40% of its annual commitments across Asia and Africa are dedicated to climate-related projects. In 2023 alone, BII’s renewable energy initiatives helped avoid 1.5 million tonnes of carbon emissions.

How quick comm became an FMCG growth engine

BY VAESHNAVI KASTURIL

Established FMCG companies have historically depended on general trade for sale. Now, quick commerce—often seen as niche and favoured by premium and digital-first brands—is slowly becoming central to their growth plans. What has changed? Mint explains.

How did q-comm aid legacy FMCG sales?

Quick commerce accounts for about 3%, or ₹1,800 crore, of Hindustan Unilever’s ₹60,000-crore-plus annual revenue. The firm has created a focused team for quick commerce since the channel demands a distinct set of capabilities. For Marico, quick commerce is a key growth driver in its India business. Emami’s share of organised and new-age channels in its domestic mix has risen from 11% in FY20 to 29% in FY25. Quick commerce contributed about 20% of its total e-commerce business in the December quarter. Tata Consumer Products has also seen a sharp rise in the quick commerce business.

How has quick comm evolved as a channel?

While it is a small part of the overall grocery market, quick commerce is growing rapidly. A Redseer report says the sector may grow at 37-39% annually between 2025 and 2030, increasing from about ₹1 trillion in 2025 to ₹5.8 trillion. It already accounts for 47% of all online grocery sales, and that share could rise to 67% by 2030. User adoption has surged, with monthly transacting users more than doubling from 23 million in 2024 to 51 million in 2025. While quick commerce may represent only around 7% of India’s total grocery market by 2030, it is already a dominant growth engine within online grocery.

What do legacy FMCG firms sell on q-comm?

Quick commerce sales are led by grocery purchases, primarily staples, packaged foods and daily-use products. However, impulse and indulgence-driven categories are growing too. Several FMCG majors have strengthened their quick commerce presence by acquiring digital-native direct-to-consumer (D2C) brands, especially in segments such as health foods, nutrition, grooming and premium personal care, which already have strong online traction and are well-suited to fast-delivery platforms.

How does quick commerce change FMCG distribution?

Q-comm is evolving from a convenience solution into a key grocery channel, projected to capture two-thirds of online grocery sales by 2030. This forces FMCG firms to overhaul supply chains, prioritise fast-moving stock-keeping units and focus on dark-store optimisation. However, high-density requirements limit the model’s scalability in small cities.

Is q-comm here to stay for FMCG firms?

Quick commerce is unlikely to replace traditional retail formats anytime soon; however, it plays a strategic role by catering to higher-spending urban consumers and enabling the sale of high-margin products. For FMCG and D2C brands, the channel offers strong visibility and recall, but it requires careful calibration of assortment and pricing across staples, discretionary items and impulse-led purchases. The model is also capital-intensive, giving larger FMCG companies a clear advantage over smaller players.

Centre considers social media curbs for under-16 users

Sejal Sharma New Delhi

The Union government is considering restricting social media access for those below the age of 16 by amending the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, according to a senior IT ministry official.

REGULATORY APPROACH

“Certain accounts should be accessible but there are certain accounts which shouldn’t. We must think judiciously if kids should have access or not (to these),” said the official, seeking anonymity. The official noted that while the Centre is considering the Australian model, it may not opt for a complete ban as that country did.

On Tuesday, Union IT minister Ashwini Vaishnaw confirmed that the government is holding consultations with social media platforms on age-based restrictions. “This is something that’s accepted by many countries—age-based regulation must be there. This was part of our DPDP (Digital Personal Data Protection) Act. Right now we are in conversation regarding deepfakes and age-based restrictions with various social media platforms,” Vaishnaw added.

CURRENT LEGAL STAND

Currently, India has no law barring children below a certain age from accessing social media. However, under the Digital Personal Data Protection Act, platforms must obtain verifiable parental consent before processing the personal data of anyone under 18. This requirement indirectly limits use by minors because platforms typically rely on parental permission to legally handle a child’s data and target ads.

GLOBAL PRECEDENTS

The government’s move follows a global trend of age-based restrictions:

• Australia: Became the first country to implement a social media ban for those under 16.
• France: Passed a bill for minors under the age of 15.
• Spain: Currently considering similar legislation.

DOMESTIC RECOMMENDATIONS

Within India, the Madras High Court has recommended the Union government consider an Australia-style law. Additionally, the Economic Survey in January recommended age-based restrictions, warning that excessive screen use among young people imposes measurable economic and social costs. The survey advised that companies should enforce age verification and implement age-appropriate defaults.

States such as Goa and Andhra Pradesh are also reportedly considering Australia-style age bans for children under 16, though implementation methods remain unclear.

AI DEMAND FOR MEMORY IS FUELLING A CHIP CRISIS

A shortage of memory chips is beginning to hammer companies—and the crunch is only going to get worse.

Bloomberg

A growing procession of tech industry leaders, including Elon Musk and Tim Cook, are warning about a global crisis in the making: A shortage of memory chips is beginning to hammer profits, derail corporate plans, and inflate price tags on everything from laptops and smartphones to automobiles and data centers—and the crunch is only going to get worse.

Since the start of 2026, Tesla Inc., Apple Inc., and a dozen other major corporations have signaled that the shortage of DRAM (dynamic random access memory)—the fundamental building block of almost all technology—will constrain production. Cook warned it will compress iPhone margins. Micron Technology Inc. called the bottleneck “unprecedented,” while Musk declared that Tesla might have to build its own memory fabrication plant to avoid hitting a "chip wall."

The AI Buildout

The fundamental reason for the squeeze is the buildout of AI data centers. Companies like Alphabet Inc. and OpenAI are gobbling up an increasing share of memory chip production by buying millions of Nvidia Corp. AI accelerators that come with huge allotments of memory. This has left consumer electronics producers fighting over a dwindling supply of chips from manufacturers like Samsung Electronics Co. and Micron.

The resulting price spikes are beginning to look like hyperinflation. The cost of one type of DRAM soared 75% from December to January, and retailers are reportedly changing their prices daily. Analysts warn that demand between now and the end of the decade will "overwhelm all other sources of demand."

Financial Strain and Product Delays

Alphabet and Amazon.com Inc. have announced construction blitzes for this year reaching $185 billion and $200 billion respectively—unprecedented capital expenditure levels. While this brings lavish profits to firms like Samsung, Micron, and SK Hynix Inc., the rest of the electronics sector is paying a painful price. Lenovo Group Ltd. CEO Yang Yuanqing explained that this structural imbalance is not a short-term fluctuation and will last at least through the rest of the year.

The disruption is already upending long-term plans:

• Sony Group Corp. is considering pushing back the debut of its next PlayStation console to 2028 or 2029.
• Nintendo Co. is contemplating raising the price of its new Switch 2 in 2026.
• Chinese smartphone makers including Xiaomi, Oppo, and Transsion are trimming shipment targets for 2026.

The Shift to HBM

Chipmakers are prioritizing High-Bandwidth Memory (HBM) over standard DRAM because the margins are better and the demand from AI hyperscalers is insatiable. Nvidia’s latest Blackwell chip comes with 192 gigabytes of RAM—six times what a modern PC needs. An integrated rack-scale system like the NVL72 uses enough memory for a thousand high-end smartphones.

Demand for HBM is expected to increase 70% year-over-year in 2026. This leaves the rest of the world bereft of the memory needed for photos, cars, and movies.

A New "Super-Cycle"

Industry executives believe they are experiencing an AI "super-cycle" that may eradicate the traditional decades-long cycle of boom and bust. Micron Executive VP Manish Bhatia called this the most significant disconnect between supply and demand in 25 years.

For consumers, the impact is immediate:

• Skyrocketing costs: DRAM could soon account for 30% of a low-end smartphone's bill of materials, tripling from early 2025 levels.
• Higher price tags: Manufacturers like Samsung and Dell have warned consumers to brace for higher prices this year.
• Panic buying: Signs of panic buying have already emerged within the auto sector.

Emergency corpus or health insurance: what works best

Policy fine print—caps, sub-limits, deductibles—often reduces payouts more than you realize

Deepti Bhaskaran Mumbai

One medical emergency can wipe out years of disciplined investing. Despite this obvious risk, health insurance remains under-penetrated across Indian households. At the Mint Money Festival, Mahavir Chopra, founder of Beshak.org, and Shilpa Arora, co-founder of Insurance Samadhan, addressed whether families can realistically skip insurance and rely on a large emergency fund instead.

EXPECTATION MISMATCH

The root cause of many health insurance complaints is an expectation mismatch. Chopra explained that the person setting the expectation (the sales agent) and the person delivering it (the claims handler) have different incentives. While the sales agent makes promises to sell the policy, the claims team focuses on protecting the insurer's profits.

This disconnect extends to product design. For instance, a policy might promise an "unlimited sum insured" to hook customers, but the claims team may later reject a deluxe room choice because it wasn't communicated clearly at the point of sale.

FINE PRINT TRAPS

Poor understanding of the "fine print" is another major issue. Factors that drastically affect final payouts include:

• Room rent caps
• Sub-limits on specific treatments
• Caps on modern therapies
• Deductibles and co-payment clauses, especially in senior citizen policies

While the customer information sheet (CIS) aims to simplify terms, its language can still be dense and difficult for those without an insurance background. Furthermore, third-party administrators (TPAs) at hospital desks often act as "postmen," merely passing files back and forth rather than helping resolve documentation errors.

EMERGENCY FUND VS INSURANCE

Can an emergency fund replace insurance? Both experts were clear: No.

Chopra noted that a ₹10 lakh cover is not equal to ₹10 lakh in savings because the cover replenishes every year even if fully used, whereas savings do not. However, building a corpus to handle small, recurring hospital bills can reduce dependence on insurance for minor expenses, allowing one to pair savings with a super top-up plan for major hospitalizations.

For those in metro cities, Arora suggested a minimum base cover of ₹10 lakh, supplemented by a super top-up cover of ₹25 lakh to ₹50 lakh. Financial protection against prolonged illnesses like cancer is increasingly critical.

CHOOSING WISELY

When selecting a plan, Arora recommended three key checks: room rent caps, coverage for modern treatments, and the insurer's hospital network. Chopra suggested looking beyond headline settlement ratios to examine the speed of claim settlement and the percentage of claims settled within 30 days. He noted that it is often easier to eliminate weak insurers than to identify a single "best" one.