AI Capex and India: Follow the Power, the Orders and the Cash

AI infrastructure is a physical spending story as well as a software story. It requires buildings, electricity, equipment, connectivity and people who can implement and operate the system. But a large spending headline does not establish how much work reaches a particular business, when it is delivered or whether the resulting cash covers its costs.

For Indian sector research, the useful question is not “Who wins from AI?” It is: where is the documented connection between demand, execution and cash? This article traces that connection without selecting securities or forecasting returns.

What the public outlooks actually contribute

In When the Tech Growth Story Meets the Bond Market, 31 August 2026, Sanctum Wealth connects technology investment with funding conditions and commercial outcomes. The selected argument is that spending and financing need to be examined together, rather than assuming the growth story resolves the cost-of-capital question.

Julius Baer’s Mid-Year 2026 outlook places AI among several investment-intensive demands on capital. It is a dated macro view, not evidence of contracts for Indian suppliers. Its exact publication day was not visible in the accessed page.

Reviewed 10 October 2026. These short, attributed views are context. The research framework below is Altys’s own educational synthesis. It does not reproduce either publisher’s investment recommendations or claim they endorse Altys. The Global Intelligence desk links the wider source register.

Begin with the physical requirement

A proposed facility needs more than a building announcement. Power availability, connection timelines, cooling arrangements, equipment delivery and network connectivity can constrain its operation. Research these requirements separately rather than treating announced capacity as commissioned capacity.

There is also a difference between an owner’s capital budget and work that an external supplier receives. Some spending may be internal, imported, delayed or spread across multiple vendors. An Indian business needs a specific, evidenced role before the global headline can be connected to its operating activity.

Useful evidence includes a disclosed contract, project milestone, commissioning update or customer delivery schedule. A broad statement about market potential is less specific. Where the link is unavailable, leave it unavailable.

An order does not arrive as cash

Consider an invented equipment supplier with no opening order backlog. During an illustrative period, it receives firm orders worth ₹300 crore, with no cancellations. It delivers and recognises ₹80 crore of revenue from those orders. Of that recognised revenue, it collects ₹60 crore by period end.

Assume there are no opening receivables, customer advances, taxes, other billings or other collection differences. These simplifications are for teaching, not a description of normal contract accounting.

Invented one-period example. Measures represent different stages, not amounts to add. No opening backlog or receivables, cancellations, advances, taxes or other collection differences. Not company data or a forecast.

Under those assumptions, ₹220 crore of the orders remains undelivered. Another ₹20 crore of recognised revenue remains uncollected. The first is future work; the second is a receivable. They are different claims and are not equivalent to cash available today.

Adding ₹300 crore, ₹80 crore and ₹60 crore would count overlapping activity multiple times. Nor can the ₹60 crore be described as profit: suppliers still have materials, labour, overheads, financing and other obligations to pay.

In real disclosures, collections can relate to earlier periods and customers can pay advances before delivery. Contract assets, retentions and billing terms may complicate the bridge. Reconcile the actual statement and contract definitions instead of importing this simplified example unchanged.

Execution can consume financing before it generates receipts

An equipment business may need to buy materials and pay workers before reaching a billing milestone. The order therefore creates work but may also create a funding requirement.

Ask who bears delays, cost changes and performance obligations. Is the commercial arrangement fixed-price, subject to escalation or otherwise structured? Can the supplier pass through a higher input cost, and on what schedule? These terms can matter more than the size of the announced project.

The cost pass-through guide explains why higher prices need not improve margins. The sector-thesis assessment method shows how to record the assumption and the evidence that could challenge it.

Demand growth and profitability need separate evidence

More activity can require additional capacity, staff or debt. A business may recognise more revenue while its cash conversion deteriorates or its financing expense rises. Neither outcome is established merely by the existence of a growing end market.

The reverse is also possible: an initially modest project may become operationally useful without producing a dramatic spending headline. Research the economics of the disclosed work rather than using the most visible announcement as a proxy for profitability.

For service providers, distinguish new implementation work from changes to existing contracts. Productivity commitments, customer pricing and scope can affect how additional AI work translates into revenue. Avoid assuming a single uniform effect across all IT services.

Build an evidence chain, not an AI association list

Record the original demand claim and its date. Then record the specific contract or project connection, delivery milestones, recognised activity and collections. Keep financing requirements and execution risks beside that chain, not in a distant footnote.

Use consistent definitions. New orders are a period flow; a closing order book is a balance at a date. Revenue is recognised over a period. Cash receipts have their own timing. If a source does not provide a clean reconciliation, say what is missing rather than calculate a false one.

Our Sector Intelligence desk extends these questions across industries. Altys’s source-linked research and monitoring can help revisit the original evidence as later disclosures arrive. Review available Excel exports to verify inputs and calculations. Request access to explore that process.

The strongest research question is not whether a company can be described as “AI-linked.” It is whether the documented work can be delivered, funded and collected on the terms the business actually faces.

General business education only. All worked figures are hypothetical. No securities recommendations, price targets, expected returns or portfolio allocations. Altys Labs is not a SEBI-registered Research Analyst or Investment Adviser.