Pan-India economic intelligenceDaily Edition — 2026-06-29
AI2

Ai2India

Daily Edition

One daily issue tracking AI adoption, markets, hiring, layoffs, real estate, credit and gig-work signals across India.

PublishedJune 29Daily issue
Nifty 50 (Jun 26 close)24,056+34 pts; Jun 27 Muharram holiday; Mon Jun 29 first session; resistance 24,200; support 23,800–23,750; Tuesday Jun 30: triple F&O expiry (Nifty weekly + monthly + Bank Nifty monthly)
Sensex (Jun 26 close)77,100.47+109 pts; profit booking + FII outflows; pre-expiry positioning Monday ahead of Tuesday triple expiry Jun 30
USD / INR~94.65Carry-forward from Jun 24 close; Jun 27 holiday; crude relief and macro context intact
Brent Crude~$73/bblCarry-forward; US–Iran diplomacy holding; India import-bill relief; low $70s level sustained

Lead Analysis — AI-First

OpenAI launches GPT-5.6 (Sol, Terra, Luna) — the first frontier model family designed with explicit US government coordination and phased access controls; Terra at approximately half the cost of GPT-5.5 is a pricing inflection for Indian enterprise; the US Commerce Department simultaneously grants Anthropic permission to release Mythos 5 to 100-plus Fortune 500 trusted partners, with Fable 5 restoration described as imminent; and Google moves to cap Meta’s Gemini API access, confirming that AI compute supply is a binding constraint even at hyperscaler scale.

The weekend of June 28–29, 2026 delivers the densest simultaneous AI policy and product news since the June 12 BIS order suspended Anthropic’s models globally. Three developments — the GPT-5.6 launch, the Mythos 5 partial restoration and the Google-Meta capacity cap — together define a new phase in frontier AI governance and access economics that Indian enterprise AI planners must absorb before Monday markets open. The signal is not that AI is arriving; it is that AI is being rationed, priced and regulated at scale, and the terms of that rationing are being written by the US government in real time. India sits between two forces: the most capable AI models are now subject to US access controls and trusted-partner restrictions, while compute infrastructure constraints mean that even unrestricted access to second-tier models is not guaranteed. The IITAAS-Singapore training programme is a quiet but concrete example of what “AI inclusion at the base” looks like in practice — and it contrasts sharply with the governance complexity at the frontier.

The GPT-5.6 launch, confirmed by Reuters and reported by Economic Times on June 27–28, is OpenAI’s most significant model release since GPT-5.5. Unlike previous launches, it introduces a family of three models rather than a single flagship. Sol is the flagship: it offers two new access modes beyond standard chat. Max mode allows the model to spend additional compute time reasoning through problems before generating a response — relevant for complex coding, biological analysis and cybersecurity vulnerability identification. Ultra mode coordinates multiple AI agents working in parallel to solve problems that exceed a single model’s capacity — the first commercially available multi-agent architecture from OpenAI that will reach enterprise users through the standard API. OpenAI states Sol outperforms prior models across coding, biology and cybersecurity tasks. The safety architecture is materially different from prior releases: Sol is designed to reject harmful cyber requests, while monitoring systems can review, delay or block risky responses before delivery. This is not a post-hoc safety layer but a design-level constraint — OpenAI explicitly notes that the safeguards may cause the model to refuse more requests or respond more slowly than earlier versions. Terra is the model with the most immediate India relevance from a cost perspective. OpenAI describes Terra as offering capabilities close to GPT-5.5 while costing approximately half as much. For Indian enterprises that have been evaluating GPT-5.5 for production deployment but held back on cost grounds, Terra represents a direct inflection point: the cost-capability frontier just moved significantly. Luna is the smallest and fastest in the family — designed for high-throughput, real-time applications that need to process large numbers of requests efficiently. Luna is the model most relevant for Indian IT services firms building AI-augmented delivery pipelines at scale, where per-token cost and latency are the binding constraints on economics. The rollout is the most explicitly government-coordinated in OpenAI’s history. The company limited the initial preview to a small number of organisations whose participation has been shared with US government officials. OpenAI says it does not expect this process to become the standard for future launches but agreed to the temporary arrangement while broader policies around frontier AI are developed. This is the same framework that the CASI pre-release review executive order (June 2, 2026) established for all frontier model labs — and GPT-5.6 is the first model explicitly deployed under it. For Indian enterprise and developer access: GPT-5.6 is currently in a limited preview restricted to a small group of trusted organisations selected in coordination with the US government. Broader public rollout through ChatGPT and the developer API has been delayed indefinitely at the request of the US government — OpenAI confirmed this in its June 26 blog post and separately in its Help Center guidance for the limited preview. OpenAI stated it does not expect this arrangement to become standard but agreed to the temporary coordination while broader frontier AI policies are developed. The Indian Express reported on June 29 that “almost no one can use them yet.” Separately, OpenAI announced on June 26 the appointment of Prabhjeet Singh — former President of Uber India and South Asia — as Managing Director for India, one of the company’s top-priority markets. Singh will join in September 2026 and report to Kiran Mani, Managing Director for APAC. This is the first dedicated India country leadership position at OpenAI, underscoring India’s significance as a developer and enterprise market. No India-specific access restriction on GPT-5.6 has been announced, but Indian API users should note that the limited-preview phase means GPT-5.6 is not yet available for general developer or enterprise access. When the public rollout opens, Indian API users should also expect the safety filters — which reject harmful cyber requests — to produce higher refusal rates on security-adjacent prompts than GPT-5.5 produced. Teams using GPT-5.5 for penetration testing, vulnerability assessment or red-teaming should evaluate whether GPT-5.6 Sol’s safety architecture is compatible with their use case before migration. Terra is the planned migration target for most Indian enterprise production deployments once the broader rollout launches.

The Anthropic Mythos 5 and Fable 5 developments represent the most significant partial resolution of the June 12 BIS export control order to date. Commerce Secretary Howard Lutnick wrote to Anthropic on June 27, stating: “Since the issuance of my June 12 letter, Anthropic has worked with the U.S. government to address risks associated with the Covered Models. These efforts have yielded significant progress.” The letter grants Anthropic permission to release Mythos 5 to over 100 organisations — reported to include many Fortune 500 companies — without requiring an export licence for those firms or their foreign national employees. This is a materially significant carve-out: it means that Fortune 500 companies with India-based development centres, GCCs and subsidiaries can access Mythos 5 through their US parent’s trusted-partner authorisation, even if the individual engineers accessing the model hold Indian citizenship. The Fable 5 situation is different. Fable 5 is the public-facing model — the version that Indian developers, startups, SMEs and individual users accessed directly through the API and through Anthropic’s chat platform before June 12. The Lutnick letter does not mention Fable 5. However, Axios reported on June 28 — citing a source close to the situation — that the Trump administration is close to allowing Anthropic to restore Fable 5 access as well, potentially “as soon as this week.” A second Axios source said conversations between the parties continued over the weekend, and Anthropic expects to restore Fable access soon. As of Monday June 29 morning (Day 17 of the June 12 BIS order), Fable 5 remains offline — confirmed by real-time API checks and TechTimes reporting from June 28. The isfable5back.com tracker, which polls the Anthropic API every minute, shows no restoration as of 7:00 AM IST June 29. The key technical distinction matters for India planning: Fable 5 and Mythos 5 use the same underlying AI model, but Fable 5 is designed for general public use while some safeguards are lifted for Mythos to serve research and specialised applications. Mythos’s trusted-partner framework is unlikely to directly help the Indian startup ecosystem or the 90%-plus of Indian API users who are not Fortune 500 subsidiaries. Fable 5 restoration — if it happens this week — will be the development that restores direct India access. The planning assumption changes materially: instead of “treat Fable 5 as unavailable through at least Q3 2026” (the June 25 edition’s guidance), Indian enterprise teams should now plan for potential Fable 5 restoration within the next seven to fourteen days, while maintaining current alternative model arrangements until official confirmation from Anthropic.

Google’s decision to cap Meta’s access to its Gemini models — reported by the Financial Times on June 27 and highlighted by The Verge — is a structural signal about the AI industry that Indian enterprise architects should carry into their infrastructure planning. Meta is among Google Cloud’s largest AI customers, using Gemini APIs to power AI features across WhatsApp, Instagram and Facebook. Despite Google spending tens of billions of dollars on chips, data centres and power, it cannot provide the compute capacity that Meta wants. Google has notified Meta and a number of other large clients that it will cap their access to its models. The FT frames this as a “rare glimpse into the infrastructure pressures and bottlenecks building across the AI industry.” For Indian enterprise: the Google-Meta capacity cap demonstrates that AI compute availability is not a solved infrastructure problem — even at the hyperscaler level where Google operates. If Google cannot guarantee capacity to Meta, the implicit assumption that India-based cloud customers have guaranteed access to AI APIs on demand is incorrect. This has direct implications for Indian enterprise architects designing AI-dependent production systems: single-provider AI API dependencies carry availability risk that has not been visible in practice until now. The bottleneck also has a structural cause: demand for AI inference is growing faster than chipmakers and data-centre builders can supply capacity. NVIDIA, AMD and custom chip programmes (Google TPU, AWS Trainium, Microsoft Maia) are all running at near-full capacity. For Indian enterprises evaluating multi-provider AI architectures: the Google-Meta cap is a data point for the business case. India-based enterprises that distribute inference workloads across two or more providers — rather than committing to a single frontier model vendor — now have a concrete, named infrastructure risk to cite in their resilience planning documentation.

The IITAAS-Singapore programme — confirmed by Economic Times on June 28 — is the week’s most important India-originated AI inclusion story. The Indian Institutes of Technology Alumni Association in Singapore signed a Memorandum of Understanding with the Migrant Workers’ Centre on June 28, supported by the Indian High Commission and High Commissioner Dr Shilpak Ambule. The programme will deliver digital literacy and AI training to approximately 1,000 migrant workers — including a large proportion of Indians working in construction, marine and manufacturing sectors — over two years. Training sessions will be held twice a month at the MWC Recreation Club at Soon Lee in the Jurong industrial zone, beginning August 2026. The curriculum covers foundational digital literacy, practical workplace AI applications and emerging technologies including AI, designed to be accessible and immediately useful to the workers’ daily lives. The March 2026 pilot — a full-day AI literacy workshop for over 100 migrant workers who chose to spend their weekly day off learning about AI — provides the proof-of-concept. For Indian enterprise and policy planners, the IITAAS-Singapore programme is notable for what it demonstrates about demand: migrant workers with weekly days off are choosing to spend them learning about AI. This is not a supply-led training initiative imposed on a reluctant workforce; it is a demand-driven response to workers’ own understanding that AI knowledge will affect their employment prospects. For India’s MeitY and NASSCOM: the IITAAS model — IIT alumni network delivering structured AI curriculum to non-technical workers through an existing welfare infrastructure (MWC) — is a template for India’s domestic AI skilling challenge, particularly for the approximately 100-plus million workers in construction, logistics, manufacturing and domestic services sectors who will be affected by AI but are not served by existing corporate AI training programmes.

June 29 signal board: GPT-5.6 public rollout delayed by US gov; Prabhjeet Singh named OpenAI India MD; Mythos 5 to 100+ Fortune 500; Fable 5 Day 17 still offline; Google caps Meta Gemini; India–US AI cooperation USISPF; Nifty 24,056 Jun 26; USD/INR ~94.65; triple expiry Tuesday Jun 30
Today’s economic signal board. Full analysis in the Daily Edition.

AI Developments Today

Monday, June 29: four developments that pass the “Would this change what an Indian enterprise AI planner does this week?” filter. The GPT-5.6 launch and Anthropic partial restoration are the most significant platform events of the week; the Google-Meta capacity cap is the most important infrastructure signal.

DevelopmentSource + DateIndia RelevanceWhat this means for Indian enterpriseStatus
OpenAI launches GPT-5.6 family — Sol (Max + Ultra modes), Terra (~50% GPT-5.5 price), Luna (fastest) — first frontier model release under explicit US government coordination framework

OpenAI has introduced GPT-5.6 as a family of three models rather than a single flagship. Sol is the top-tier model: new Max mode allows extended reasoning time before response generation; Ultra mode coordinates multiple AI agents in parallel for complex problems. Sol performs better than prior models in coding, biology and cybersecurity. Safety architecture is design-level: Sol rejects harmful cyber requests and monitoring systems can review, delay or block risky responses. Terra targets enterprise cost efficiency: OpenAI claims near-GPT-5.5 capabilities at approximately half the price. Luna is the fastest and lightest, designed for high-throughput, real-time and large-volume API workloads. Rollout: initial preview to a small group of organisations whose participation was shared with US government officials; then broader rollout through ChatGPT and the developer API. OpenAI states it does not expect the government-coordination process to become standard but agreed to the temporary arrangement while broader frontier AI policies are developed. Safety note: Sol may refuse more requests or respond more slowly than GPT-5.5 due to enhanced safety filters. No India-specific access restriction has been announced for GPT-5.6.		 Reuters (Jun 27, 2026); Economic Times (Jun 27–28); OpenAI announcement Terra’s approximately 50% cost reduction vs GPT-5.5 is the most direct India relevance signal. Indian enterprises that evaluated GPT-5.5 and deferred on cost grounds should immediately re-evaluate Terra as a production target. Ultra mode’s multi-agent coordination is directly relevant to Indian GCCs and IT services firms building agentic AI delivery pipelines — this is the first OpenAI multi-agent mode available through the standard API. For Indian BFSI enterprises: Sol’s enhanced safety architecture for cybersecurity — designed to support vulnerability identification without enabling cyberattacks — aligns with Indian regulators’ AI-use requirements for financial services. For Indian IT security firms: assess whether Sol’s higher refusal rate on security prompts is compatible with existing penetration testing and red-teaming workflows before migrating from GPT-5.5 Cyber or other current tools. Luna is directly relevant for Indian IT services firms running high-volume AI-augmented delivery pipelines where per-token cost and latency are the binding economic variables. Immediate action for Indian enterprise AI teams: evaluate Terra as the primary production migration target from GPT-5.5. At approximately half the price with near-equivalent capability, Terra changes the cost-benefit calculation for mid-market enterprise deployment. Do not rush Sol adoption without testing safety filter compatibility with your specific prompting patterns — the refusal rate increase may require prompt engineering adjustment. For teams building agentic AI: Ultra mode is the first commercially available multi-agent mode from OpenAI; prioritise evaluation for complex multi-step workflows. Update pricing models in all AI business cases immediately with Terra’s estimated cost. Verified — Reuters; ET; Jun 27–28
US Commerce Department grants Anthropic permission to release Mythos 5 to 100+ trusted Fortune 500 partners; Fable 5 restoration described as imminent by Axios — partial reversal of June 12 BIS export control order (day 16)

Commerce Secretary Howard Lutnick wrote to Anthropic on June 27: “Since the issuance of my June 12 letter, Anthropic has worked with the U.S. government to address risks associated with the Covered Models. These efforts have yielded significant progress.” The letter grants Anthropic permission to release Mythos 5 to over 100 organisations, including many Fortune 500 companies, without requiring an export licence for those firms or their foreign national employees or for Anthropic’s own foreign national employees. The Commerce Department characterised this as demonstrating that “America remains the global leader in AI while safeguarding our security.” Key technical distinction: Fable 5 and Mythos 5 use the same underlying AI model but serve different purposes. Fable 5 is the public-facing general-use model that Indian developers, startups and API users accessed before June 12. Mythos 5 is the research and specialist version with some public safeguards lifted. The Lutnick letter does not mention Fable 5. However, Axios reported on June 28 — citing two sources — that the Trump administration is close to allowing Anthropic to restore Fable 5 access “as soon as this week,” with conversations continuing over the weekend. Anthropic-US government relations context: the suspension was triggered by fears that Mythos 5 and Fable 5 could be deployed by military intelligence users in China, Russia or other countries of concern. The partial restoration after 16 days is faster than the June 25 edition’s Q3 2026 planning horizon suggested.		 Reuters (Jun 27, 2026); Commerce Department letter (Lutnick to Anthropic); Economic Times (Jun 27); Axios (Jun 28); The Verge (Jun 27) Mythos 5 trusted-partner access: directly available to Fortune 500 companies and their foreign national employees — meaning Indian engineers at GCCs of Fortune 500 companies can access Mythos 5 through their employer’s authorisation without a separate export licence. This is significant for Indian-origin engineers at Accenture, IBM, Microsoft, Google, AWS GCCs and equivalent firms. For Indian startups, independent developers and API users who are not Fortune 500 subsidiaries: Mythos 5 trusted-partner access does not directly help. Fable 5 restoration — if confirmed this week — is the development that restores direct India access for this group. For Indian enterprise procurement teams: update Anthropic continuity planning from “Q3 2026 planning horizon” to “possible within 7–14 days.” Do not cancel alternative model arrangements until official Fable 5 restoration is announced by Anthropic — the Axios sourcing is reliable but the timeline is not officially confirmed. Immediately update your Anthropic platform continuity planning. Mythos 5 is now available to Fortune 500 trusted partners — if your company is in that category, contact your Anthropic account manager today about the authorisation process. For Fable 5: maintain current alternative model arrangements (GPT-5.5/Terra, Sarvam AI, Gemini 3.1 Pro) but begin technical readiness review for rapid Fable 5 re-activation when official confirmation arrives. Do not rely on the Axios timeline for SLA commitments. The biometric ID pathway announced for the July 8 date remains relevant if Fable 5 general access restoration happens in parallel. Verified — Reuters; Commerce Dept letter; Axios; Jun 27–28
Google moves to cap Meta’s Gemini API access due to compute capacity constraints — reveals AI infrastructure as a binding constraint even at hyperscaler scale

The Financial Times reported on June 27 that Google has notified Meta and a number of other large clients that it cannot provide the AI compute capacity they want and will cap their access to Gemini models. Meta is among Google Cloud’s largest AI customers: it relies on Gemini APIs to power AI features across WhatsApp, Instagram and Facebook — which collectively reach hundreds of millions of users, including India’s largest AI-user base. The FT frames the disclosure as a “rare glimpse into the infrastructure pressures and bottlenecks building across the AI industry.” Despite Google spending tens of billions of dollars on chips, data centres and power, it cannot meet demand. This is not a bilateral negotiation issue between Google and Meta; it reflects a structural supply-demand imbalance across the AI compute market. The bottleneck affects the entire inference stack: NVIDIA, AMD, Google TPU, AWS Trainium and Microsoft Maia capacity are all near full utilisation as AI inference workloads grow faster than chip production and data-centre construction can keep pace.		 Financial Times (Jun 27, 2026) via The Verge (Jun 27) For Indian enterprise architects with single-provider AI API dependencies: the Google-Meta capacity cap is a concrete, named example of the availability risk that has been theoretical until now. If Google cannot guarantee Meta’s API access, India-based cloud customers should not assume guaranteed AI API availability for mission-critical production workloads. Multi-provider AI architectures — distributing inference across two or more providers — are now a risk management requirement, not an engineering preference. For Indian WhatsApp and Instagram users: Meta’s AI features (powered by Muse Spark via Gemini API) may degrade in quality or availability if Google’s capacity caps reduce Meta’s inference budget. This is a concrete consumer-facing risk for India’s largest AI user base. For Indian IT services firms building cloud-native AI platforms for clients: include provider capacity risk as an explicit architectural assumption in 2026 AI platform designs. Single-vendor dependencies on AI APIs are a service continuity risk. Audit current single-provider AI API dependencies across production workloads. Identify the top-3 AI API calls by volume and business criticality. For each, design a failover path to an alternative provider. Document provider capacity risk as an explicit assumption in AI platform architecture reviews. For Indian enterprises on Google Cloud with significant Gemini API usage: discuss capacity commitments and SLAs with Google Cloud account managers this week. The Meta cap is a signal that large-volume customers face rationing risk; your own volume commitments may need formal agreement. Verified — FT via The Verge; Jun 27
IITAAS signs MOU to deliver AI and digital literacy training to 1,000 Indian migrant workers in Singapore — supported by Indian High Commission; programme starts August 2026

The Indian Institutes of Technology Alumni Association — Singapore (IITAAS) signed a Memorandum of Understanding with the Migrant Workers’ Centre (MWC) on June 28, supported by Indian High Commissioner Dr Shilpak Ambule. The MWC is a 16-year-old bipartite welfare body between the National Trades Union Congress and the Singapore National Employers Federation. The programme will deliver structured AI literacy and digital skills training to approximately 1,000 migrant workers — primarily Indians working in construction, marine and manufacturing sectors — over two years. Training sessions twice a month at MWC’s Recreation Club at Soon Lee, Jurong industrial zone; first session August 2026. Curriculum: foundational digital literacy, practical workplace AI applications, emerging technologies including AI. Proof-of-concept: March 2026 full-day AI workshop for 100+ migrant workers who chose to spend their weekly day off learning about AI. This builds on IITAAS’s earlier financial literacy workshops in migrant worker dormitories, which the Indian High Commission helped facilitate. MWC Director Michael Lim: “Technology must empower, not exclude.” IITAAS President Dhruv Jain: “We aim to provide accessible and practical training that supports their safety, communication and overall well-being.”		 Economic Times (Jun 28, 2026) The IITAAS model is directly replicable in India. The structure — IIT alumni network + existing welfare infrastructure (IITAAS uses MWC in Singapore; equivalent: Construction Workers’ Welfare Board, unorganised workers’ sanghas, or e-Shram-registered platform cooperatives in India) — provides a proven template for delivering AI literacy to non-technical workers without corporate training infrastructure. For Indian enterprises with CSR mandates under Companies Act 2013 (Schedule VII): AI literacy training for construction workers, gig workers or manufacturing workers using the IITAAS model qualifies as a scientifically sound education intervention. For MeitY and NASSCOM: the March 2026 pilot data — migrant workers choosing to spend their one weekly day off on AI training — is the strongest demand-side evidence of worker-level AI adoption aspiration that has emerged from the Indian diaspora context. This demand exists in India too and is not being met by current programmes. India’s MeitY AI skilling programmes and NASSCOM’s FutureSkills Prime target employed tech workers and students. The IITAAS-Singapore programme addresses a different population: non-technical manual workers who will be AI-adjacent (not AI-native). This population is orders of magnitude larger and materially underserved. The IITAAS model should be proposed to MeitY as a template for a domestic AI-adjacent worker training initiative through existing welfare infrastructure channels. Verified India — ET; Jun 28

India AI Ecosystem

Monday, June 29: three new India AI ecosystem signals — the ICAI + Sarvam AI MoU (June 26), Digital India’s 11th anniversary disclosures (July 1, 2026), and NASSCOM’s production-maturity data (June 26) — add significant domestic context to the IITAAS-Singapore programme covered in the lead. GPT-5.6 Terra pricing is the most actionable frontier model signal for India’s Tier 4 AI platforms.

Platform / OrganisationDevelopmentIndia AI SignificanceStatus
OpenAI India — Prabhjeet Singh named Managing Director; GPT-5.6 public rollout delayed
Managing Director appointment
June 26, 2026
Join date: September 2026		 OpenAI announced on June 26, 2026 the appointment of Prabhjeet Singh as Managing Director for India, one of the company’s top-priority markets globally. Singh is the former President of Uber India and South Asia, widely regarded as one of India’s most experienced technology business executives from the global platform economy. He will join OpenAI in September 2026 and report to Kiran Mani, OpenAI’s Managing Director for Asia Pacific. This is the first dedicated India country leadership role at OpenAI, established as the company deepens its commitment to India’s developer community, enterprise AI market and government digital transformation agenda. The appointment was made simultaneously with the GPT-5.6 launch announcement. However, the GPT-5.6 public rollout has been delayed indefinitely at the request of the US government — OpenAI confirmed this in its June 26 blog post. GPT-5.6 Sol, Terra and Luna are currently in a limited preview restricted to a small group of organisations selected in coordination with US government officials. Broader public rollout through ChatGPT and the developer API has no confirmed timeline as of June 29 morning. The Prabhjeet Singh appointment is the most significant OpenAI-India signal since the company opened its India office. Singh’s background — building Uber’s India and South Asia business from early stages to a major market — maps directly to the challenge OpenAI faces in India: converting a large, price-sensitive, technically sophisticated developer and enterprise market from experimentation to committed production deployment. For Indian enterprises evaluating OpenAI partnerships: the arrival of India-specific leadership means commercial partnership discussions, enterprise pricing and local support escalation will no longer route through APAC generically. For Indian startups building on OpenAI APIs: Singh’s Uber India background suggests OpenAI may pursue startup ecosystem partnerships and developer community investments similar to Uber’s driver-partner and city-specific growth playbook. For Indian AI policy: OpenAI establishing India-country leadership is a signal that it views India as a priority jurisdiction for regulatory engagement alongside the US and EU. The simultaneous GPT-5.6 public rollout delay (indefinite, per US government request) is a complicating factor: Singh will be leading India operations through a period where OpenAI’s most advanced models are in a US-government-coordinated access regime. Verified India — The Hindu; Bloomberg; Outlook Business; Jun 26, 2026
IITAAS + MWC (Singapore)
AI literacy programme
1,000 Indian migrant workers
MOU signed June 28		 Indian Institutes of Technology Alumni Association — Singapore signed MOU with the Migrant Workers’ Centre on June 28, 2026, supported by Indian High Commissioner Dr Shilpak Ambule. Programme will deliver AI and digital literacy training to approximately 1,000 migrant workers (mainly Indians in construction, marine, manufacturing) over two years. Training twice a month at MWC Recreation Club, Soon Lee, Jurong. Starts August 2026. Curriculum: foundational digital literacy, practical workplace AI applications and emerging technologies. March 2026 pilot established proof-of-concept: 100+ workers chose to attend on their only weekly day off. The partnership follows earlier IITAAS financial literacy workshops at migrant worker dormitories, also supported by the Indian High Commission. This is the first confirmed, structured AI literacy programme for Indian blue-collar workers outside India, run by IIT alumni through official welfare infrastructure. It demonstrates that the demand for AI skills exists across the full spectrum of the Indian workforce — not only among tech workers and students. For MeitY, NASSCOM and state-level skill development bodies: the IITAAS model (professional alumni network + welfare organisation infrastructure) is a cost-effective, demand-validated template for domestic AI skilling at the base of the pyramid. The March pilot’s voluntary attendance rate (workers spending their day off learning about AI) is a proof-of-demand that does not require further pilot work before national replication. Verified India — ET; Jun 28, 2026
ICAI + Sarvam AI — MoU signed at AI Innovation Summit 2026
New Delhi, June 26, 2026
Union Minister inaugurated		 The Institute of Chartered Accountants of India (ICAI) signed a Memorandum of Understanding with Sarvam AI at the ICAI AI Innovation Summit 2026 in New Delhi on June 26, 2026. Union Minister Gajendra Singh Shekhawat inaugurated the event. The MoU covers AI training, capacity building, research, workshops and the development of AI-centric learning resources for chartered accountants. ICAI simultaneously launched AICA Level 3 — the most advanced tier of its AI for Chartered Accountants certification programme — and a new AI practice guide for CA members. ICAI President CA Prasanna Kumar D disclosed that more than 50,000 ICAI members have already been trained in AI across 150+ chapters. The medium-term aim is a custom accounting large language model optimised for Indian audit, tax and governance workflows. ICAI has approximately four million members, making this the largest professional body in India to formally adopt a domestic AI platform for institutional capacity building. ICAI’s choice of Sarvam AI — India’s sovereign AI platform — over international frontier models for its professional development infrastructure is a strategic endorsement with broad implications. The BFSI, audit and legal adjacent sectors that depend on CA-grade analysis are now being trained on Indian-language, Indian-regulatory-context AI tools. For Indian enterprises deploying AI in accounting, compliance and audit workflows: ICAI-aligned AI tools will benefit from a growing trained professional base. For Sarvam AI: the ICAI MoU provides a credentialed, large-scale deployment channel (4M members) for its accounting AI capabilities that no international provider can replicate through the regulatory and professional-body trust pathway. The custom accounting LLM, when released, will be the first domain-specific Indian AI model with a national professional body’s formal endorsement. Verified India — TOI; Tribune; Careers360; Newkerala; Jun 26, 2026
Digital India at 11 — ₹1.64 lakh crore semiconductor investments; 45,000 GPUs deployed for AI research
July 1, 2026 anniversary
MeitY / GoI disclosures		 As Digital India approaches its 11th anniversary on July 1, 2026, the Government of India disclosed headline infrastructure numbers: ₹1.64 lakh crore (approximately USD 19.5 billion) in approved investments across 12 semiconductor manufacturing projects under the India Semiconductor Mission; over 45,000 GPUs deployed for AI research nationally. The 12 semiconductor projects include one semiconductor fabrication facility, two compound semiconductor plants, and additional ATMP and OSAT units. The government also cited 45,000 GPU deployment as the base layer for the IndiaAI Mission’s national AI compute backbone. Separately, NASSCOM disclosed at the NASSCOM US CEO Forum (June 26) that approximately 25% of Indian technology services companies have moved AI experiments into production; the industry is generating an estimated USD 10–12 billion in AI services revenue; more than 2 million professionals are skilled in AI; and NASSCOM projects a USD 400 billion AI opportunity for India’s IT services sector. The Digital India infrastructure numbers — ₹1.64 lakh crore semiconductor pipeline and 45,000 AI GPUs — provide the domestic infrastructure baseline against which India’s frontier AI ambitions must be measured. For context: the 45,000 GPU figure for national AI research is approximately the same scale as a single large hyperscaler training cluster. This is a foundation, not a frontier. The NASSCOM 25% production-deployment figure is the most important adoption metric to emerge this quarter. That one in four Indian IT services firms has moved AI out of pilot and into production — generating $10–12B in revenue — means India’s AI transformation is no longer speculative. It is a production-scale economic event. For Indian enterprise AI planners: the 75% of firms still in experiment mode are now competing against a sector cohort that has already operationalised AI. The production gap between early movers and laggards is widening. For investors and policy planners: the $400 billion NASSCOM projection, if achievable in the current AI export control environment, requires that Indian firms maintain access to frontier AI APIs — which makes the Fable 5 restoration timeline a direct NASSCOM-target variable. Verified India — ET Government; TechObserver; Tribune; NASSCOM; Jun 26–28, 2026
Sarvam AI
10M API calls/day baseline
Carry-forward		 Sarvam AI’s 10 million API calls per day metric — reported June 22–23 — carries forward as the baseline production-scale measure for India’s sovereign AI infrastructure. Sarvam’s thevam30B edge model and its 105B parameter large model remain available without export-control restrictions. The Fable 5 partial restoration (Mythos 5 trusted partners; Fable 5 imminent) does not diminish Sarvam’s strategic value — the June 12 BIS order demonstrated that even unrestricted-access models can be restricted with 24-hour notice. Sovereign model availability requires no US approval pathway. GPT-5.6 Terra’s pricing at approximately half GPT-5.5 creates a new competitive comparison point for Sarvam’s pricing teams: to retain enterprises that might now switch back to OpenAI on cost grounds, Sarvam’s combined price-sovereignty-reliability proposition requires explicit re-articulation. Sarvam at 10M API calls/day is no longer a pilot or aspiration — it is a production-scale sovereign AI infrastructure benchmark. The question for the post-Fable-5-partial-restoration environment: is Sarvam’s proposition purely about export-control resilience, or does it have a standalone cost-capability argument that holds when GPT-5.6 Terra is available at half GPT-5.5 price? Indian enterprises that adopted Sarvam as a Fable 5 substitute should now run a deliberate comparison exercise: Sarvam vs GPT-5.6 Terra on their specific workloads, with sovereign resilience weighted as a separate factor in the decision matrix. Verified India — BusinessLine; Moneycontrol; Jun 22–23 (carry-forward)
India startup ecosystem
Seed rounds doubling
AI/deeptech focus
CRED mega round		 Indian VC data from the week of June 22–28 shows seed and early-stage funding rounds approximately doubling in average cheque size as VCs front-load capital for more mature AI, deeptech and healthtech products. The number of deals has dropped — indicating a more concentrated investment approach — while average round size has risen. CRED’s mega round lifted total startup funding to $1.09 billion for the week (up approximately 290% year-on-year). Qualcomm’s $4 billion acquisition of AI startup Modular — reported this week — is the largest edge-AI infrastructure acquisition of 2026, signalling that edge inference (running AI locally on chips rather than in the cloud) is a priority investment category. The Hyderabad Economic Forum 2026 (500+ leaders at SAS ITOWER) highlighted Telangana’s commitment to AI initiatives and educational reform as part of the city’s growth vision. The seed-round-doubling and VC-concentration trend directly affects how Indian AI startups should approach fundraising in H2 2026: smaller numbers of better-funded rounds means VCs want demonstration of product-market fit before leading rounds, not after. The Qualcomm-Modular acquisition is a signal for Indian AI chip and edge inference startups — the international M&A market for AI infrastructure assets is active at $4 billion scale. Indian founders in edge AI, on-device inference or AI compiler tooling should note this as a strategic comparables data point. Verified India — ET; Jun 22–28
India–US AI & semiconductor cooperation roundtable — USISPF; MeitY Secretary S. Krishnan; Ambassador Vinay Kwatra
Washington, June 25, 2026		 India and the United States deepened cooperation in artificial intelligence, semiconductors, quantum technologies and critical minerals at a high-level roundtable on June 25 in Washington hosted by the U.S.–India Strategic Partnership Forum. India’s Ambassador to the United States Vinay Kwatra said: “The opportunity before the United States and India extends from chips to neural networks. India’s mission-based approach across semiconductors, AI, and quantum technologies, combined with America’s innovation ecosystem, creates enormous potential for collaboration.” MeitY Secretary S. Krishnan said India is “rapidly emerging as a trusted partner in global technology manufacturing” and noted that semiconductor fabrication is becoming a reality with the next phase of India’s Semiconductor Mission building on current momentum. Ministry of External Affairs Additional Secretary (Americas) K. Nagaraj Naidu said India–US ties have evolved into a “comprehensive strategic partnership fit for the 21st century.” The roundtable directly followed the June 12 BIS export control order affecting Anthropic — the India-US cooperation framing is partly a diplomatic response to the narrative that US export controls are restricting India’s AI access. The India–US AI cooperation roundtable is the policy context within which the GPT-5.6 US-government-coordinated rollout and the Anthropic Mythos 5 partial restoration should be read. MeitY Secretary Krishnan’s “trusted partner” positioning and Ambassador Kwatra’s explicit inclusion of India in the “chips to neural networks” collaboration frame signals that India is seeking bilateral arrangements that reduce the export-control friction for Indian enterprises and GCCs accessing frontier AI models. For Indian enterprises: the India–US cooperation language at senior diplomat and MeitY level suggests that Fable 5 restoration and future frontier model access will be influenced by bilateral tech diplomacy, not only by commercial or legal frameworks. India’s leverage in this negotiation is its position as a priority AI market (OpenAI naming India MD; NASSCOM $400B opportunity) and its semiconductor and defence tech cooperation with the US. Verified India — Indica News; Jun 25–29, 2026 (USISPF roundtable)
Carry-forward: C-DOT + IIT Hyderabad 6G/AI centre; Maharashtra AI Policy 2026; IndiaAI Mission portfolio C-DOT + IIT Hyderabad joint research centre (6G, AI, quantum, cybersecurity) — announced June 25 — carries forward as Hyderabad’s sovereign AI research anchor. Maharashtra AI Policy 2026 (₹10,000 crore investment target, ₹500 crore venture fund, 12 incubators) — announced June 23 — remains the most detailed state-level AI ecosystem framework in India. IndiaAI Mission’s portfolio rationalisation question (“How many models can India sustainably build and maintain?”) remains the headline policy question for the coming quarter. GPT-5.6 Terra’s pricing reduction may reduce the cost argument for building domestic models in some categories — but strengthens it in others where sovereignty, language support and regulatory compliance require Indian-origin model ownership. State-level AI policy momentum (Maharashtra ₹10,000 cr; Telangana Hyderabad Economic Forum) is the most active layer of India’s AI governance architecture in 2026. Central government IndiaAI Mission direction will be the next major policy signal to watch. The GPT-5.6 Terra pricing reduction is a new variable in the IndiaAI Mission’s make-vs-buy calculus that should be acknowledged in the next Mission steering committee discussion. Verified India — Multiple; Jun 23–28 (carry-forward)

AI Adoption Impact

June 29: two direct AI adoption impact signals — GPT-5.6 Terra’s pricing inflection (lowers barrier to enterprise deployment) and the Google-Meta compute cap (raises supply-side risk for AI-dependent production systems). Carry-forward signals from prior editions provide the structural baseline.

AI Impact DimensionEvidenceTrajectory
GPT-5.6 Terra pricing: ~50% GPT-5.5 cost — direct barrier reduction for Indian enterprise AI adoption Reuters / ET (Jun 27–28). OpenAI’s Terra model: near-GPT-5.5 capabilities at approximately half the price. This is the first meaningful frontier-tier price reduction since GPT-4 Turbo pricing changes in late 2023. For Indian enterprises that evaluated GPT-5.5 for production deployment and deferred on cost grounds: the Terra price point changes the ROI calculation for AI-augmented service delivery, AI-assisted coding and AI-enabled customer service at scale. The Luna model adds a third tier for high-volume, real-time applications where per-token cost is the binding variable. The net effect: OpenAI has simultaneously extended the capability frontier (Sol with Ultra mode) and dramatically lowered the cost-of-entry for production deployment (Terra). This is a deliberate market-segmentation strategy that addresses both frontier users and cost-sensitive mid-market adopters simultaneously. ↑ Accelerating; Terra pricing removes a key stated barrier to mid-market Indian enterprise AI adoption; expect faster deployment decisions in H2 2026 for enterprises with approved AI programmes but deferred budgets
AI compute supply: Google-Meta cap demonstrates hyperscaler-level AI API availability is not guaranteed FT via The Verge (Jun 27). Google caps Meta’s Gemini API access due to compute constraints. Despite tens of billions in investment, demand exceeds supply. NVIDIA, AMD, TPU, Trainium, Maia — all near capacity. This is the first public, named instance of a Tier 1 AI provider rationing access to a Tier 1 customer at production scale. Indian enterprise AI architects have been designing for software-layer risks (model availability, API pricing, provider switching costs); the Google-Meta cap introduces infrastructure-layer availability risk as a new category that was not previously documented at this scale. Single-provider AI API dependencies in mission-critical production systems are now demonstrably non-trivial availability risks. ↓ Risk rising; AI compute supply constraints are a structural constraint on enterprise AI adoption velocity, regardless of model capability or pricing improvements; multi-provider architecture is the appropriate response
Anthropic access: Mythos 5 to Fortune 500 GCCs; Fable 5 imminent — partial restoration changes enterprise planning horizon Reuters; Commerce Dept letter (Jun 27). Mythos 5 available to 100+ Fortune 500 companies and their foreign national employees — meaning Indian engineers at Fortune 500 GCCs can access Mythos 5 through employer authorisation. Fable 5 (public model) restoration imminent per Axios (Jun 28). The June 25 edition’s “treat Fable 5 as unavailable through Q3 2026” planning assumption should be updated to “possible restoration within 7–14 days, maintain alternatives until official confirmation.” The Mythos-Fable distinction is critical for India planning: Mythos is a research tool for specific applications; Fable is the general-purpose API that the Indian developer ecosystem was using. Only Fable 5 restoration fully restores India’s pre-June-12 access position. ↑ Positive; Anthropic access restrictions partially lifting faster than expected; Indian enterprise teams should update continuity plans and prepare for rapid Fable 5 re-activation while maintaining current alternatives
NASSCOM AI production maturity: 25% of India IT firms in production; $10–12B AI revenue; 2M+ skilled professionals NASSCOM US CEO Forum (June 26, 2026) disclosures — confirmed by The Hindu, Business Standard, BusinessLine, News18. Nearly 25% of Indian technology services companies have moved AI experiments into production deployment. Industry generating an estimated USD 10–12 billion in AI services revenue. More than 2 million professionals skilled in AI; 100,000+ in advanced AI roles. NASSCOM projects a USD 400 billion opportunity for India’s IT services sector driven by AI. The next growth phase is expected from enterprise AI transformation, AI foundations, application modernisation, AI operations, trust and governance, and vertical AI solutions. NASSCOM notes the opportunity comes from both global enterprises and Indian enterprises / government digital platforms requiring trusted, secure and population-scale AI deployment. ↑ Accelerating; 25% production maturity means India IT AI is no longer in experiment phase — it is a $10–12B production-scale revenue event; firms still in pilot are now the laggards, not the leaders; production gap between early movers and majority is the defining AI adoption dynamic for H2 2026
Carry-forward: AI hiring premium; Nifty IT structural derating; physical-AI training economy; Oracle AI-linked workforce disclosure Carry-forward from Jun 24–25 editions. AI/ML/data roles: 3.5L openings in 90 days; software dev postings −12.3% in 3 months. Nifty IT weight at 7.6% (20-year low). Physical-AI training economy: ₹100–250/hr; Neocambrian AI Noida 100+ sites. Oracle FY2026 10-K: AI explicitly cited for 21K global job cuts, India est. 11–12K. These structural signals are unchanged by this weekend’s AI product and policy events and form the economic baseline for the June 29 edition. ↓/↑ Structural divergence continues; AI premium roles accelerating; traditional IT contracting; no change from Jun 25 baseline

Five Things That Changed

Monday, June 29: three AI platform/policy developments from the weekend, one India AI ecosystem signal, and one macro/market context item. All five carry direct enterprise action implications for the week ahead.

SignalData PointReader ImpactStatus
OpenAI GPT-5.6: Sol (Max + Ultra modes), Terra (~50% GPT-5.5 price), Luna — limited preview only; public rollout delayed indefinitely; India MD Prabhjeet Singh (ex-Uber India) announced Reuters / ET Jun 27–28; The Hindu / Bloomberg Jun 26. Three-model family: Sol (flagship, extended reasoning, multi-agent), Terra (near-GPT-5.5 capability at ~50% cost), Luna (fastest, high-throughput). Public rollout delayed indefinitely per US government request (Indian Express Jun 29: “almost no one can use them yet”). Limited preview to trusted-partner organisations only; broader ChatGPT / API rollout has no confirmed timeline. Separately: OpenAI named Prabhjeet Singh (former Uber India & South Asia President) as its first dedicated Managing Director for India on June 26; joins September 2026. Sol safety: rejects harmful cyber requests; may refuse more prompts than GPT-5.5. No India-specific access restriction announced. Update AI budget models immediately with Terra’s estimated pricing. Schedule Terra evaluation alongside current GPT-5.5 production workflows. For agentic AI projects: prioritise Ultra mode evaluation in Sol. For high-volume IT services delivery: Luna is the cost-efficiency target for next-generation AI-augmented pipelines. Verified — Reuters; ET; Jun 27–28
Anthropic Mythos 5 to 100+ Fortune 500 trusted partners; Fable 5 Day 17 — still offline; restoration “imminent” per Axios Commerce Sec. Lutnick letter Jun 27; Reuters; Axios Jun 28; real-time API tracker Jun 29. Mythos 5: 100+ organisations, many Fortune 500, export licence waived for trusted partners and their foreign national employees. Fable 5: Axios (Jun 28) — “as soon as this week”; conversations continued over weekend. Real-time API tracker (isfable5back.com) confirms Fable 5 offline at 7:00 AM IST Jun 29. Lutnick letter does not mention Fable 5; restoration not yet officially confirmed. Day 17 of June 12 BIS order. Fortune 500 GCC teams: contact Anthropic account manager now about Mythos 5 access authorisation. All other enterprises: maintain current alternatives (GPT-5.6 Terra, Sarvam AI, Gemini 3.1 Pro) but prepare rapid Fable 5 re-activation procedures for when official confirmation comes. Do not rely on Axios timeline for SLA or client commitments. Verified — Reuters; Commerce Dept; Axios; Jun 27–28
Google caps Meta’s Gemini API access — AI compute supply is a binding constraint at hyperscaler scale FT via The Verge Jun 27. Google cannot provide Meta the compute capacity it wants for AI features across WhatsApp/Instagram/Facebook. Named as “infrastructure pressures and bottlenecks building across the AI industry.” Despite tens of billions in chips and data centres, demand exceeds supply. Affects NVIDIA, AMD, Google TPU, AWS Trainium, Maia capacity across industry. Audit single-provider AI API dependencies for mission-critical workloads this week. Design failover paths to alternative providers for your top-3 highest-volume or business-critical AI API calls. Include provider capacity risk as an explicit assumption in all AI platform architecture documentation going forward. Verified — FT via The Verge; Jun 27
IITAAS to train 1,000 Indian migrant workers in Singapore — AI literacy for construction and manufacturing sector; starts August 2026 ET Jun 28. IITAAS + MWC MOU signed Jun 28; Indian High Commission backed; High Commissioner Dr Shilpak Ambule present. 1,000 workers over 2 years. Twice-monthly sessions at Soon Lee, Jurong. Starts August 2026. Curriculum: digital literacy + workplace AI applications + emerging tech. March 2026 pilot: 100+ workers attended on their one weekly day off voluntarily. For Indian enterprises with CSR mandates (Schedule VII, Companies Act 2013): the IITAAS model is a proven template for AI literacy CSR investment targeting non-technical workers. For MeitY and NASSCOM: this is the most concrete demand-side evidence of AI learning aspiration at the bottom of the workforce pyramid — replicable domestically through existing welfare infrastructure channels. Verified India — ET; Jun 28
Markets: Nifty 24,056 (Jun 26 close); Jun 27 Muharram holiday; USD/INR ~94.65 carry-forward; Tuesday Jun 30 brings triple F&O expiry — Nifty weekly + monthly + Bank Nifty monthly simultaneously ET Markets; Moneycontrol; Univest Jun 28. Jun 26 (Thu): Nifty 24,056 (+34.35), Sensex 77,100.47 (+109). Jun 27 (Fri): all India markets closed for Muharram. Jun 29 (Mon): first session of the week. Key technical levels: Nifty resistance 24,200; support 23,800–23,750 (Moneycontrol trade setup). Crucial pre-expiry positioning session: Tuesday June 30 brings simultaneous triple expiry — Nifty 50 weekly, Nifty 50 monthly and Bank Nifty monthly all expire on the same day. USD/INR ~94.65 and Brent ~$73 are carry-forward estimates from June 24 close; no fresh India forex or crude data was generated on June 27. Q1 business updates, FII flows, monsoon progress and US bond yields are the week’s macro watchpoints per ET. Verified — ET Markets; Jun 28

Data Variables Ledger

Verified numbers as of Monday, June 29, 2026. Market data as of June 26 close (June 27 Muharram holiday). Carry-forward where same-day data not yet available.

VariableValueSourceDate
Nifty 5024,056 (+34.35)ET Markets; NSEJun 26 close (last trading day)
Sensex (BSE)77,100.47 (+109)ET Markets; BSEJun 26 close
USD / INR~94.65 (carry-forward)Jun 24 close; Jun 27 holidayCarry-forward
Brent Crude~$73/bbl (carry-forward)Carry-forward from Jun 24–25Carry-forward
Nifty IT (YTD)−29% YTD (carry-forward)Bloomberg; Business Standard; Jun 24Jun 24 (last trading day with IT data)
Nifty IT weight in Nifty 507.6% (record low, 20-year)Bloomberg; Business Standard; Jun 24Jun 24
GPT-5.6 SolFlagship; Max mode (extended reasoning); Ultra mode (multi-agent); coding/bio/cyber focus; safety-enhanced; 3× higher self-reasoning control rate vs prior models (AI Weekly safety card); public rollout delayed indefinitely by US governmentReuters; ET; Indian Express; AI Weekly; Jun 26–29Jun 26–29
GPT-5.6 Terra~50% GPT-5.5 price; near-GPT-5.5 capability; primary enterprise migration targetReuters; ET; Jun 27–28Jun 27–28 (launched)
GPT-5.6 LunaFastest; high-throughput; real-time / large-volume API workloadsReuters; ET; Jun 27–28Jun 27–28 (launched)
Anthropic Mythos 5 (access)100+ Fortune 500 trusted partners; export licence waived; foreign national employees includedReuters; Commerce Dept letter (Lutnick); Jun 27Jun 27
Anthropic Fable 5 (status)Still suspended for general public (Day 17 of Jun 12 BIS order); API tracker offline at 7:00 AM IST Jun 29; restoration “imminent” per Axios Jun 28Axios Jun 28; isfable5back.com; TechTimes Jun 28Jun 29 morning
Google Gemini / Meta APICapped; compute constraints; Meta notified of capacity limitsFT via The Verge; Jun 27Jun 27
IITAAS Singapore programme1,000 Indian migrant workers; 2 years; starts Aug 2026; MOU with MWCET; Jun 28Jun 28
Sarvam AI API calls/day~10 million (carry-forward)BusinessLine; Moneycontrol; Jun 22–23Carry-forward
Oracle India job-cut estimate~11,000–12,000 (estimate; not company-confirmed)Babushahi; ET; Jun 22–25Carry-forward
India AI openings (90-day)~3.5 lakh (350,000)Quess Corp / Moneycontrol; Jun 23Carry-forward
Software dev postings (3-month change)−12.3%Indeed India / India Today; Jun 23Carry-forward
Repo rate (RBI)5.50% (Jun 5 MPC hold; August MPC is next live window)RBIJun 5
Gemini 3.5 Pro (Google)Delayed to July; Antigravity/LMArena tuning (carry-forward)Business Insider; Jun 24Carry-forward
ICAI + Sarvam AI MoUMoU signed Jun 26 at AI Innovation Summit; AICA Level 3 launched; 50,000+ members trained; custom accounting LLM planned; Union Minister inauguratedTOI; Tribune; Careers360; Jun 26Jun 26, 2026
Digital India GPU count45,000 GPUs deployed for AI research nationally (IndiaAI Mission)ET Government; TechObserver; Jun 27–28Jun 27–28 (11th anniversary disclosures)
India Semiconductor Mission investments₹1.64 lakh crore approved across 12 projects (1 fab, 2 compound semi, ATMP/OSAT units)ET Government; TechObserver; Tribune; Jun 27–28Jun 27–28
NASSCOM AI services revenueUSD 10–12 billion; 25% of India IT firms in production; 2M+ skilled; USD 400B opportunity projectedThe Hindu; Business Standard; BusinessLine; Jun 26Jun 26 (NASSCOM US CEO Forum)
OpenAI India MDPrabhjeet Singh (former Uber India & South Asia President); joins September 2026; reports to Kiran Mani (APAC MD); first dedicated India country leadership role at OpenAIThe Hindu; Bloomberg; Outlook Business; Jun 26, 2026Jun 26, 2026 (announced)
India–US AI cooperation (USISPF roundtable)High-level roundtable Jun 25, Washington; MeitY Secretary S. Krishnan: India is “trusted partner in global technology manufacturing”; Ambassador Vinay Kwatra: “chips to neural networks” collaboration opportunity; MEA Additional Secretary K. Nagaraj Naidu presentIndica News; Jun 25–29, 2026Jun 25 (roundtable)
Jun 30 triple F&O expiryTuesday June 30 simultaneous: Nifty 50 weekly + monthly + Bank Nifty monthly expiry; Nifty resistance 24,200; support 23,800–23,750Moneycontrol; Univest; Jun 28, 2026Jun 28 pre-expiry analysis

Verified Layoff Radar

India-confirmed items only. AI-driven restructuring flagged separately from general cost cuts. No new verified India-specific layoff events in the June 27–29 period. All items carry-forward from the June 25 edition. The most recent layoff-radar sweep file in the publishing queue covers June 1 (no new publishable verified items at that date); watchlist items remain on watchlist.

CompanyAnnouncedIndia ImpactRestructuring DriverStatus
OracleFY2026 (filing confirmed Jun 22)~11–12K estimated (Bengaluru, Hyderabad, Pune); global: 21,000 confirmed in 10-K; India not officially broken outAI-driven (company’s own regulatory filing explicitly cites AI deployment as cause)Verified global; India est. — Bloomberg; Oracle 10-K; ET; Jun 22
LinkedInMay 2026300–350 India employees across engineering, product, marketing and GBO teamsAI-linked restructuring (broader Microsoft AI transition)Verified India — ET; May 2026
OpendoorJun 2026~250 India employees; India operation closingAI-linked restructuringVerified India — ET; TOI; Jun 2026
TCSFY2026Headcount 584,519; down 23,460 YoY; no new layoff programme announcedAI-led delivery efficiency + demand softnessOfficial workforce change — ET; FY2026
FreshworksMay 2026~500 global; India and US affected; India count not separatedGeneral cost reduction; AI transitionVerified global — earnings call; May 2026

Watchlist items (HCLTech Xerox BPM ~170–200 Noida; Cognizant Project Leap 12,000–15,000 global; Accenture guidance risk; Nokia India ~3,000+ potential) remain on the watchlist and are not published as verified items without a company-backed India count or filing disclosure. No new watchlist items promoted in the June 27–29 period.

Hiring Demand Watch

AI/ML/data roles versus general IT: the structural divergence is unchanged from the June 25 baseline. GPT-5.6 Terra’s pricing inflection may accelerate AI hiring demand in the mid-market in H2 2026 as enterprise deployment barriers fall. IITAAS-Singapore programme adds a new data point for AI skills demand at the base of the workforce pyramid.

Role CategoryDemand SignalSourceDirection
GenAI engineering, prompt engineering, AI deployment3.5 lakh openings in 90 days (all AI-native); 10+ unfilled roles per qualified candidate (Randstad Digital)Quess Corp / Moneycontrol; Randstad Digital / ET; Jun 23 (carry-forward)↑ Accelerating
Multi-agent AI architecture (GPT-5.6 Ultra mode)New demand category emerging with Ultra mode launch; enterprise agentic AI pipeline roles will be required to integrate and manage multi-agent coordinationReuters/ET GPT-5.6 launch; Jun 27–28↑ New category
AI-augmented security (Sol/GPT-5.5-Cyber overlap)Sol’s enhanced cybersecurity capabilities + safety architecture create demand for specialists who can navigate the higher refusal-rate environment; distinct from GPT-5.5-Cyber Trusted Access pathwayOpenAI GPT-5.6; Jun 27–28↑ Growing
OpenAI India country operations (commercial, BD, policy) — September 2026 onwardsPrabhjeet Singh’s arrival as India MD (September 2026) will create demand for local OpenAI staff in commercial, enterprise sales, developer relations and policy roles. Uber India’s scaling precedent under Singh included significant local hiring in operations, engineering and public policy. For Indian enterprise AI planners: OpenAI’s India country presence adds a local escalation channel for enterprise API issues, pricing negotiations and safety compliance discussions.The Hindu; Bloomberg; Jun 26, 2026↑ Emerging (September 2026 onwards)
Physical-AI training data (robot training — carry-forward)₹100–250/hr; Neocambrian AI 100+ Noida sites; 100M hrs needed; under 1 year old as sector (carry-forward)CNBC Inside India; Guardian; Jun 24–25↑ Emerging (low wage)
Software development (traditional)−12.3% in 3 months; steepest single-category IT decline in current cycle (carry-forward)Indeed India / India Today; Jun 23↓ Contracting
General IT (broad)Overall tech postings −0.7% in May 2026; active openings at 28-month low (~93K) (carry-forward)Indeed India; Xpheno; Jun 2026↓ Soft

Real Estate Pulse

GCC and AI company office moves only. No new verified Grade A GCC or AI company leasing event reported June 27–29. Carry-forward items from prior editions remain in effect.

Company / SignalLocationDetailStatus
Target India GCC leaseEmbassy Manyata, Bengaluru₹1,250 crore; 830,000 sq ft; 10-year term; 15% tri-annual escalation (carry-forward)Verified India — Jun 18
India data-centre pipelinePan-India8.33 GW pipeline (Knight Frank India); Haryana AI data centre (Panchkula) adding tier-II supply (carry-forward)Verified India — Jun 21
C-DOT + IIT Hyderabad centreHyderabad6G/AI/quantum/cybersecurity research centre; physical infrastructure anchor for Hyderabad’s sovereign AI research geography (carry-forward)Verified India — Jun 25

Market Signals

Four ticker cards only. June 26 close data (last trading day — June 27 Muharram holiday). USD/INR and crude are carry-forward estimates.

Nifty 5024,056+34 pts (Jun 26 close); Sensex 77,100 (+109); mixed volatile expiry session; range-bound trade expected ahead
USD / INR~94.65Carry-forward (Jun 24); Jun 27 holiday; macro backdrop of crude relief and stable DXY unchanged
Brent Crude~$73/bblCarry-forward; US–Iran diplomacy intact; India import-bill relief sustained; low $70s level holding
Repo Rate (RBI)5.50%Jun 5 MPC hold; August MPC is next live window; Warsh Fed hike bias caps RBI cut room near-term

Forecast Tracker Updates

June 29 evidence notes added to active predictions where today’s data is directly relevant. No new forecasts initiated today — GPT-5.6 Terra pricing is noted as a watch variable for the existing AI adoption prediction.

PredictionUpdate Note (Jun 29)Confidence
Frontier AI governance will require nationality verification for advanced model access by end-2026 (initiated 2026-06-23) GPT-5.6’s rollout under explicit US government coordination — “access restricted to a small number of organisations whose participation has been shared with government officials” — is the first frontier model launch to deploy under the CASI pre-release review framework (June 2 executive order). The public rollout has since been delayed indefinitely at the US government’s request — going further than the trusted-partner coordination originally described. Mythos 5 trusted-partner list (100+ Fortune 500; export licence waived) is a named-company verification regime. Both implementations add evidence that nationality and partner-identity verification is becoming infrastructure for frontier AI access, not an exceptional measure. New June 29 evidence: OpenAI’s appointment of a dedicated India MD (Prabhjeet Singh) while simultaneously restricting GPT-5.6 access demonstrates the tension between commercial market priority and US access-control frameworks — OpenAI values India as a top-priority market even as US government coordination constrains the speed and breadth of model deployment there. Confidence raised to 79%. 79% (↑ from 74%)
AI talent concentration at OpenAI/Anthropic will widen frontier capability gap vs all other labs (initiated 2026-06-24) GPT-5.6 Sol’s Max mode (extended reasoning) and Ultra mode (multi-agent) represent the most significant capability extension in OpenAI’s history since GPT-5. The simultaneous launch of Sol (frontier), Terra (cost-optimised) and Luna (throughput-optimised) demonstrates a product-line depth that only organisations with both frontier research talent and production engineering talent can execute simultaneously. Confidence raised to 63%. 63% (↑ from 60%)
AI, cloud, data and premium roles will outperform generic hiring (initiated 2026-05-30) GPT-5.6 Ultra mode’s multi-agent coordination creates a new enterprise engineering specialisation (multi-agent AI architect, agentic pipeline engineer) that does not exist in the current India hiring market at scale. Terra’s cost reduction may accelerate AI platform deployment decisions in H2 2026, adding demand for AI integration and deployment engineers. IITAAS-Singapore programme adds demand-side evidence that AI skills aspiration extends to the base of the workforce pyramid. New this edition: NASSCOM US CEO Forum (June 26) reports 2 million professionals skilled in AI with 100,000+ in advanced AI roles — the strongest supply-side confirmation yet that specialised AI talent is growing rapidly in India. The 25% production-deployment figure means companies are now scaling proven AI use cases, which increases demand for AI engineers, integration architects and domain-AI specialists. Confidence raised to 85%. 85% (↑ from 83%)
If crude stays below $90 and rupee holds near 95, market stress should unwind faster than hiring stress (initiated 2026-06-12) Markets closed June 27 (Muharram). Last close: Nifty 24,056 (Jun 26). Crude and rupee carry-forward are within thesis parameters (~$73, ~94.65). Weekend AI product news (GPT-5.6, Anthropic partial restoration) is net positive for tech sector sentiment in Monday opening. Hiring stress unchanged — no new data over the long weekend. Market-hiring divergence thesis intact. Confidence unchanged at 78%. 78% (unchanged)

Source Notes