Prisma Global Ltd: what investors need to know about this unlisted AI story

Prisma Global Limited, operating under the brand Prisma AI, is a Mumbai-headquartered computer vision company. It has been circulating in India's unlisted-share market as a pre-IPO "AI platform" opportunity, currently priced around ₹7,000-7,500 per share. The technology story is real and the client relationships are genuine. But the financial structure underneath the AI branding tells a more complicated story.

What does Prisma Global Ltd do?

Prisma Global was incorporated in December 2013, though the brand traces back further, to 1981, when a former IBM team founded Prisma Global GmbH in Germany to focus on optical character recognition. Founder Dr. Shreeram Iyer eventually built on an earlier partnership arrangement to acquire the German parent, and the business has since evolved into a visual AI and computer vision company.

The company offers following suite of product portfolio:

• Gryphos (Core Platform)

It is the flagship visual AI engine that powers all Prisma AI products with machine learning and deep convolutional neural network (CNN). It provides real time analysis of videos, static images, object recognition, OCR, and behavioral pattern recognition. It is deployed via private cloud – integrating directly with existing CCTV systems.

• Dedicated Product Suites

Prisma also package and deliver specialized turnkey products designed for specific needs:

• GryFAS – Biometric & Facial Payment Authentication
• Veri5 – Digital Identity & Access Verification
• GrySCAN – X-Ray & Baggage Threat Inspection
• GrySEC – AI Perimeter Security & Surveillance

• API Engine and Vertical Solution

Prisma also offers 230+ pre-trained API engined that can be deployed directly into client applications, some of them includes:

• Smart Mobitlity & highways
• Aviation & Transit Hubs
• Retail and Inventory
• Specialized Analytics – Document Reconstruction, Postural Analysis, Safety Monitoring.

Analysis

Its core product, Gryphos, is a proprietary computer vision platform. It processes video and image data into real-time, actionable outputs: facial recognition and authentication, vehicle number-plate reading, emotion and behavioral analysis, footfall tracking, and automated baggage tracking. The company markets this as "Pragmatic AI", ready-to-use API libraries meant to plug into existing camera and surveillance infrastructure rather than requiring a ground-up rebuild. It serves high-security, high-compliance sectors: aviation, banking and financial services, infrastructure, and manufacturing. The company reports more than 35,000 AI-powered cameras deployed globally, and it has a notable partnership with the Adani Group deploying AI solutions across multiple international airports.

The company operates six global offices, in India, Germany, Singapore, Dubai, Italy, and London, plus additional partner offices. It describes itself as the first Indian company in its sector to receive ISO 42001:2023 certification for AI management systems. It is currently pursuing what's been described as a "reverse flip", merging its global operations into the Indian entity rather than listing an offshore holding company. The stated goal is to create a single India-listed platform ahead of a future IPO.

How does Prisma Global Ltd make revenue?

Prisma Global Ltd generates revenue through a hybrid business model combining recurring software subscriptions (SaaS) model, volume-based transaction fees, enterprise system integrations and long-term O&M contracts.

Revenue Streams

• SaaS & Feed Licensing

Here, Prisma earns a recurring revenue via 230+ tier-based API licensing and per camera or per feed subscription.

• Usage & Transaction-based Revenue

Through its products like GryFAS, the company charges per facial verification or transaction approval across banks and fintech companies and via specialized suites like GrySCAN, monetization is structured around the volume of data processed.

• Turnkey Integration & Custom Deployment Fees

Prisma delivers large-scale infrastructure, government and smart city projects and charges upfront revenue for defining deployment scope, fine tuning visual AI models. In addition, system licensing payouts generated by embedding Prisma AI software into multi-million-dollar tenders.

• Operation & Management Services

This is an ongoing service fees model where Prisma charges annual maintenance fees and tiered support contracts providing 24/7 aid.

Analysis

Revenue comes from licensing and deploying the Gryphos platform: API access, hardware-integrated deployments, and ongoing service contracts across its aviation, BFSI, and infrastructure client base. On paper, the growth has been substantial. Reported figures show revenue climbing from roughly ₹150 crore in FY21 to around ₹484 crore in FY24, a reported compound annual growth rate near 47%.

Here's where the picture gets more complicated. FY24-to-FY25 filings show revenue continuing to grow, but profitability is moving the opposite direction. Profit is declining alongside rising expenses and rising debt, a pattern independent commentary has flagged without fully explaining the mechanism. The company generates hundreds of crores in "AI platform" revenue, yet reported net margins have sat in the low single digits (commonly cited around 3-4%). That's a strikingly thin margin for a business marketed as a scalable, proprietary software platform rather than a hardware-and-services integrator.

The balance sheet adds real weight to that concern. MCA filings show Prisma Global's paid-up capital at roughly ₹5.03 crore against authorized capital of ₹6.2 crore, a small equity base. Meanwhile, the company carries open secured charges of approximately ₹112-117 crore, more than 20 times its paid-up capital. Investors would typically expect a capital-light, API-driven software platform to be equity-funded and asset-light. A balance sheet this debt-heavy looks more like a systems integrator financing hardware purchases, installation, and project-based deployments than a pure-play software business.

There's also a structural reason margins might look this thin regardless of how the group is actually performing. Prisma Global's parent, Prisma AI Corporation Pte Ltd (Singapore), holds an 82.23% stake in the Indian entity. If the Singapore parent holds the Gryphos IP itself, then the India entity's profit would be structurally suppressed by design, independent of underlying group performance (due to the brand, or any associated licensing and royalty economics, a common structure for cross-border technology groups). That's a materially different explanation for a 3-4% net margin than "rising costs." A review of the Indian entity's financials alone can't fully surface this, since the profit isn't missing; the group may simply recognize it elsewhere.

Who are the competitors of Prisma Global Ltd?

Prisma Global competes in the computer vision and visual AI space, most directly against companies applying AI to video and image analytics for security, surveillance, and operational intelligence:

• Global computer vision and video-AI platforms: companies building camera-agnostic AI layers for facial recognition, behavioral analytics, and surveillance intelligence at global scale. These companies are typically well-capitalized and equity-funded rather than debt-financed, a structural contrast worth noting against Prisma's own balance sheet.
• Domestic Indian AI-in-surveillance and analytics vendors: a fragmented but growing field of Indian companies applying computer vision to security, retail analytics, and public-safety use cases. They generally compete on cost and local integration relationships in a similar way to Prisma's own client wins.
• Traditional security systems integrators: companies that install, maintain, and service CCTV and surveillance hardware for airports, banks, and infrastructure clients without necessarily branding themselves as AI platforms. This is arguably the most relevant comparison for evaluating Prisma's actual balance sheet and margin profile. The debt-funded, project-heavy financial structure looks closer to this category than to a scaled software platform.

Peers of Prisma

• Swan Solutions and Services Private limited
• Solutions Infini Technologies (India) private Limited
• Bloom Electronics private Ltd
• Trianz Digital Consulting Private Limited

The honest positioning question is which of these two identities fits better. Is Prisma Global a proprietary AI platform business with recurring, high-margin software economics? Or is it a systems integrator, buying, installing, and maintaining camera hardware on a project basis, that has adopted AI-platform branding on top of a services-and-hardware core? The balance sheet leans toward the second reading more than the marketing materials suggest.

Is Prisma Global Ltd a good investment opportunity?

Green flags:

• This is a real, revenue-generating business with genuine enterprise relationships. The Adani Group airport deployments and multi-decade OCR/computer-vision lineage aren't fabricated, and the company has built real technical credibility, including ISO 42001:2023 certification.
• Revenue has grown substantially and consistently over multiple years, and the company has completed institutional-adjacent private placements at rising per-share prices, suggesting some sustained investor demand.
• The "reverse flip" restructuring, consolidating global operations into the Indian entity, is a genuine step toward IPO-readiness. If the company executes it transparently, it could resolve some of the ownership-structure questions below.

Red flags:

• Revenue and profit moved in opposite directions in the most recent filed year. Combined with the capital structure below, this margin-compression pattern looks less like temporary cost pressure and more like a structural feature of the business.
• The capital structure doesn't match the platform narrative. Paid-up capital of about ₹5 crore against roughly ₹113-117 crore of secured borrowing is a debt-funded, balance-sheet-heavy profile more typical of a systems integrator than an equity-funded software platform.
• An offshore Singapore parent owns 82.23% of the Indian entity. Retail and HNI investors aren't actually being offered the parent; they're being offered a subsidiary of the brand-holding parent. If IP and royalty economics sit with Singapore, thin India-entity margins may be structural rather than a sign of business weakness or strength either way. But it also means the entity being priced on the unlisted market isn't necessarily the value-accruing part of the group.
• Governance shows a closely-held, high-churn board. Current directors include Manoj Tande, Mayur Rondhe, Shreeram Iyer, Amitabh Roy Chowdhury, and Maryann Shreeram Iyer (a surname pairing consistent with a founder/spouse director structure), while past directors, including Girish Pandit, Vidhi Dhavade, and Volker Brendel, have since exited. The company added two new directors as recently as September 2025. That pattern of rotation looks more consistent with pre-listing board composition adjustments than a long-institutionalized governance structure.
• The founder's promoted "PhD" credential, from the European International University (EIU) - Paris, doesn't hold up to the scrutiny the marketing materials invite. EIU-Paris's own published materials confirm its degrees are "institutional degrees", not French national diplomas, and they explicitly fall outside France's national qualifications framework. This says nothing about the underlying technology, but it's a consistent thread with other prestige-signaling claims in the company's promotional materials, and it's the kind of detail that rewards a second look rather than being taken at face value.

Bottom line: strip away the AI branding, and Prisma Global reads less like a scaled software platform and more like a project-and-hardware-driven systems integration business. An offshore parent majority-owns it, it carries the balance sheet and margin profile of a low-single-digit-margin IT services contractor, and yet the unlisted market prices it in the range of much richer earnings multiple than that profile would typically command. No single data point here disqualifies the company on its own. Margin compression happens, offshore holding structures are standard for cross-border tech companies, and board churn ahead of a listing isn't unusual. But together, they suggest the entity being sold to retail and HNI investors may not be the value-accruing half of this group, and that enthusiasm around India's AI infrastructure theme is underwriting the current price more than the actual FY24-FY25 financial trendline, which is moving the wrong way. The clearest way to test this thesis going forward would be visibility into related-party transaction disclosures, specifically any royalty or management-fee payments to the Singapore parent, in the next filed balance sheet, and whether FY25's margin compression reverses as volumes plateau.