|
Getting your Trinity Audio player ready...
|
SMART INVESTING INDIA | INVESTMENT RESEARCH & ANALYSIS | NOVEMBER 2026
The Grand Ambition vs. The Harsh Reality of AI Infrastructure 🏗️🧠
Reliance Industries has never been a company to think small.
Under the leadership of Mukesh Ambani, Jio disrupted the Indian telecom space with ruthless efficiency and staggering capital expenditure.
Now, the conglomerate is setting its sights on the next frontier: Artificial Intelligence.
With announcements suggesting a monumental ₹10 lakh crore investment in AI infrastructure by 2033, the market is buzzing with excitement.
However, as practitioners of the Lethargic Investing framework, we must separate technological hype from sustainable shareholder value creation.
We do not invest in press releases, nor do we buy into futuristic visions unless they are backed by durable competitive advantages and a clear path to profitability.
This deep dive evaluates whether Reliance’s AI ambitions can translate into a wide-moat, highly profitable enterprise, or if it risks becoming a massive sinkhole of shareholder capital.
Reliance’s AI strategy: Building an integrated AI ecosystem 🌐🔗
Reliance’s approach is broader than launching an AI chatbot or developing an indigenous large language model.
It is attempting to build an integrated ecosystem spanning computing infrastructure, connectivity, cloud services, enterprise applications and consumer distribution.
The strategy rests on four interconnected capabilities: AI infrastructure (GPU clusters and large data centres), digital distribution (Jio’s mobile, broadband and enterprise customers), AI applications (consumer, business and industry-specific AI), and energy integration (renewable power and existing energy infrastructure).
The strategic logic is important: Reliance does not necessarily need to win the global foundation-model race.
It could become a commercially successful AI infrastructure and application provider by combining technology developed elsewhere with its own infrastructure and distribution.
Reliance Intelligence and the Jamnagar AI backbone 🏭💻
Reliance has established Reliance Intelligence as the vehicle for its deeper technology ambitions.
At its 2026 AGM, management outlined a substantial AI infrastructure programme at Jamnagar, targeting the commissioning of the first 120 MW of AI infrastructure by the end of 2026.
An initial fleet of NVIDIA GB300 GPUs is being operationalised, which management described as equivalent to more than 75,000 H100 GPUs on an AI-inference basis.
Once the first 120 MW becomes fully operational, capacity could scale beyond 200,000 H100-equivalent GPUs.
However, these are management targets and capacity comparisons, not evidence that the entire infrastructure is already operational or commercially utilised.
A data centre with 120 MW of electrical capacity is not the same thing as 120 MW of productive computing capacity.
Actual commercial output depends on GPU availability, networking, cooling, utilisation, model efficiency and customer demand.
Strategic partnerships: Google and Meta 🤝🌐
Reliance is not attempting to develop every component of the AI stack independently.
The October 2025 partnership provides eligible Jio users with Google AI Pro access, including Gemini capabilities, for 18 months at no additional cost.
The Reliance–Meta enterprise AI initiative aims to build Llama-based solutions for Indian businesses, combining Meta’s models with Reliance’s infrastructure and Indian enterprise relationships.
Reliance can potentially avoid bearing the entire cost and risk of developing frontier AI models while concentrating on infrastructure, localisation, hosting, deployment and customer relationships.
However, partnerships also introduce dependence on external model developers and hardware suppliers.
The Jamnagar–Meta data centre agreement 🤝🏢
One of the most interesting developments is the June 2026 agreement with Meta, where Reliance will build a 168 MW AI-enabled data centre in Jamnagar, which Meta will lease.
Meta is the anchor customer and will cover the facility’s energy and water costs.
This is strategically significant because it potentially converts a large infrastructure investment into a contracted commercial relationship.
However, investors still need disclosure of the lease economics, construction expenditure, commissioning schedule, contract duration, expected returns and financing structure.
Energy: Reliance’s potentially important structural advantage ⚡☀️
AI infrastructure consumes substantial amounts of electricity, and energy availability can become a competitive constraint.
Reliance’s clean-energy investments could potentially lower the cost and improve the reliability of electricity supplied to its AI facilities.
But renewable energy is not automatically cheap, uninterrupted electricity, as solar and wind generation are intermittent.
The relevant measure is the fully loaded cost of reliable electricity delivered to the data centre, not simply the announced cost of solar generation.
Where can Jio actually make money? 💰📈
Reliance has several potential AI revenue streams, including AI infrastructure (GPU and compute leasing), enterprise AI (implementation and recurring contracts), AI cloud (usage-based computing and subscriptions), and consumer AI (premium subscriptions and bundled services).
Enterprise AI is potentially the most attractive opportunity, as Indian businesses increasingly need practical AI solutions for customer service, document processing, and manufacturing optimisation.
Reliance could offer an integrated enterprise solution combining cloud hosting, model access, data security, deployment and ongoing support.
However, this is a competitive market, and Jio will need to demonstrate measurable productivity improvements and attractive customer economics.
Inference economics: The real test of the AI business ⚙️📉
The cost of running AI models is recurring, and every inference request consumes computing resources.
A data centre’s profitability depends on how many GPUs are operational, the percentage of available computing hours sold, the revenue each GPU generates, and the electricity each workload consumes.
AI hardware has a potentially shorter economic life than traditional industrial infrastructure, creating a depreciation problem.
A company could report healthy EBITDA while its free cash flow remains weak because substantial capital expenditure is required to maintain computing competitiveness.
Competitive positioning: Can Jio compete with global technology companies? 🌍🥊
Reliance is entering a market with formidable competitors like Amazon Web Services, Microsoft Azure, and Google Cloud.
Reliance also faces competition from domestic infrastructure providers like the Adani Group, Tata Communications, and Yotta Data Services.
Where Jio has a genuine structural advantage is its distribution, connectivity, energy integration, and access to substantial internal cash flows.
However, these are enabling advantages rather than guaranteed competitive moats.
Capital allocation: How much is Reliance putting at risk? 💸📊
Capital invested in AI infrastructure competes with other potential uses of capital, such as existing business expansion, debt reduction, and shareholder returns.
The separate listing of Jio Platforms in October 2026 gives investors a more direct way to assess the digital and technology business.
Ideally, Reliance should disclose separate metrics for AI-related capital expenditure, installed GPU capacity, utilisation rate, and AI revenue.
The hidden risk: AI infrastructure can become commoditised ⚠️📉
There is an important distinction between owning infrastructure and owning a competitive advantage.
If competitors can offer lower computing prices or use the same open-source models, Reliance faces a risk of price competition.
The dependency on NVIDIA for advanced GPUs and the potential for hyperscalers to aggressively reduce AI computing prices are significant risks.
Execution and regulatory risks 🛑⚖️
Investors should consider risks like technological obsolescence, energy availability, regulatory uncertainty, data privacy, and model reliability.
The Jio Platforms prospectus explicitly identifies risks associated with AI/ML model errors, data quality, algorithmic bias, and unintended outcomes.
Will AI become a profit centre or remain a supporting capability? 🎯🏢
Reliance has two distinct opportunities: AI as a standalone business or AI as an enabler of existing businesses.
Using AI to lower operating costs in existing businesses like Jio Telecom or Reliance Retail may offer a more predictable initial return.
However, the standalone AI infrastructure business must eventually demonstrate that it can earn attractive returns without cross-subsidisation.
Conclusion: Profitability or Pipedream? 🏁🔮
Reliance’s AI bet is a high-stakes gamble that leverages its core strengths in infrastructure, energy, and distribution.
Jio undoubtedly possesses the scale and capital to build a formidable sovereign AI backbone for India.
However, building infrastructure is easy when you have endless capital; generating superior, cash-backed returns from it is exceedingly difficult.
Until the company demonstrates that its AI operations can achieve high utilization, command pricing power, and deliver an ROIC above its cost of capital, the Lethargic Investor must remain cautiously on the sidelines.
Watch the cash flows, ignore the headlines, and demand evidence of durable profitability.
Discover more from Smart Investing India
Subscribe to get the latest posts sent to your email.