Dubai, 27 October 2026: PointAI says it is seeking $150 million to develop a 250-billion-parameter commerce foundation model and expand its Simulation AI products. The proposed funding would support enterprise retail tools, virtual try-on and its Agentic Commerce App. The announcement describes a fundraising target, rather than a completed investment round.
How PointAI Simulation AI Works
PointAI describes its approach as physics-based simulation for digital commerce. Its official research overview explains how the company models garment fit, fabric drape and product details for virtual shopping experiences. These descriptions represent the company’s claims about its technology.
The explanation usually offered is organisational: poor data, weak change management, unclear ownership. PointAI argues the real reason is technical. Generative AI is probabilistic by nature — it predicts what something should look like. That works for conversation. It fails when a model must stand in for a physical object. More compute does not fix it – it merely sharpens the average.
Simulation AI is a fundamentally different approach, built for real-world understanding. Rather than predicting what a product looks like, PointAI’s engine computes it — modelling an object’s physical properties and how it actually behaves. In apparel, for instance, that means modelling fabric weight, drape and stiffness against real body geometry, then rendering the specific garment on the specific body from a single flat-lay shot and one selfie, against more than 200,000 body types. The same principle holds for any physical product a customer needs to see before buying.
PointAI’s Claims on Computing Costs and Rendering Speed
PointAI says its models rely primarily on CPUs and use fewer GPU resources than generative alternatives. It describes the following potential benefits. The cost, speed and accuracy comparisons below are company claims, rather than independent benchmarks verified by GulfInside:
- Cost — published API pricing puts a generated try-on image at around fourteen cents; PointAI renders at roughly a hundredth of that, the difference between a pilot and a full catalogue.
- Speed — under a second per render, against thirty to sixty seconds for generative alternatives.
- Accuracy — because the system computes rather than predicts, the same input returns the same output on every run, preserving the actual product rather than reinterpreting it.
- Control — the company runs its own models and calls no foundation-model API, so enterprise clients can host a dedicated model on their own premises and confidential catalogue, pricing and customer data never leaves the building.
- Scale — less compute per render means deployment economics that hold as volume grows rather than breaking under it.
Two Products, One Simulation Core
PointAI has built two products on that same engine. PointAI Platforms serves enterprises and brands — catalogue imagery and video for every SKU without a photoshoot, interactive 3D and AR product models, virtual try-on, size and fit recommendation, real-time mix-and-match styling, and in-store solutions including an AI-powered virtual trial standee — all through one platform and a single SDK rather than a stack of separate vendors.
The Agentic Commerce App brings the same engine to shoppers. Instead of a flat product image, a customer sees the actual garment on a digital twin built from their own measurements, mixes and matches across categories, and is guided by a video-based advisor working from body type and personal style rather than browsing history.
PointAI’s technology is deployed with Amazon, Rakuten, Flipkart and ABFRL, and the company has begun expanding into the Gulf, where retail groups are moving quickly on AI-led customer experience.
How PointAI Plans to Use the Proposed $150 Million
If the fundraising is completed, PointAI says it plans to direct the capital towards four priorities:
- A 250-billion-parameter simulation AI based commerce foundation model spanning text, speech and simulation. A product expert that renders physical products accurately, a call center agent who can converse in more than 100 languages and dialects, realistic video-call experiences through human digital twins.
- Research and development – strengthening PointAI’s Simulation AI intellectual property across video and advertising generation, improving model accuracy for more realistic digital twins etc.
- Scaling the Agentic Commerce App across 50+ categories to 20 million users globally
- Infrastructure to support commercial scale-up across both products.
The foundation model is not an exercise in chasing scale for its own sake: reasoning accurately about physical products demands a depth general-purpose models do not carry. And the entire stack runs on PointAI’s own models and its own infrastructure. The company does not wrap third-party foundation models, which is what allows it to control cost, guarantee consistency of output, and keep enterprise data inside the client’s perimeter.
“Enterprises have been asked to accept output that is approximately right, at a cost that only works in a pilot, from a model they must send their data to. None of those three is acceptable at production scale,” said Nitin Vats, Founder and Chief Executive of PointAI. “We built the engine on physics so the answer is computed rather than guessed, and we run it on our own infrastructure so the economics and the data stay in our clients’ control. This raise is about scaling that technology to enterprises and users across all domains and geographies.”
For related coverage, explore GulfInside’s technology news and business news.
About PointAI
PointAI is an AI company specialising in physics-based Simulation AI for fashion and commerce, headquartered in Noida, India, with presence in Singapore and the United States. Founded by Nitin Vats, the company holds more than 70 granted patents across the United States, South Korea and India, spanning digital twin generation, virtual try-on, 3D generation and simulation. Its two products — PointAI Platforms and the Agentic Commerce App — are built on a single simulation core. Investors include Lip-Bu Tan, Chief Executive of Intel
