Analysis · Commerce & Consumer Brands

AI inference costs worry Indian startups

What they do

Style Lounge is a Noida-based beautytech startup founded in 2024

AI inference costs account for 8–12% of its AI and cloud infrastructure costs and are expected to rise to 25%

India is the world's second-largest consumer AI market, and inference could account for 90–95% of AI compute demand by 2030

AI inference costs are a growing concern for Indian startups.
Inference is the process of using trained AI models to generate responses in real-world applications.
Zoho cofounder Sridhar Vembu likened AI compute dependence to an 'oil import bill' for the AI era.
Zoho spends 'a few million a year' on AI model subscriptions.
Style Lounge, a Noida-based beautytech startup founded in 2024, sees AI inference costs as 8–12% of its AI and cloud infrastructure costs.
Style Lounge expects AI inference costs to rise to 25% of its AI and cloud infrastructure costs.
Style Lounge relies on AWS for cloud and GPU infrastructure, plus OpenAI APIs and image analysis models.
In AI-first SaaS, every skin-analysis flow, chatbot response, recommendation, automation trigger, or customer-care interaction creates a paid inference event.

India is the world's second-largest consumer AI market.
Indian users generated over one billion images using ChatGPT Images 2.0, per Sam Altman.
India could require about 7 GW of AI compute capacity by 2030, nearly 30 times current levels, according to Zinnov president Sidhant Rastogi.
Inference is expected to account for about 90–95% of total AI compute demand in India by 2030.
The global AI inference market is expected to grow from $106 billion in 2025 to over $520 billion in 2034, per Polaris.

Style Lounge is a Noida-based beautytech startup founded in 2024.
It relies on AWS for cloud and GPU infrastructure, plus OpenAI APIs and image analysis models.
AI inference and model usage account for 8–12% of Style Lounge's AI and cloud infrastructure costs.
Style Lounge expects AI inference costs to rise to 25% of AI and cloud infrastructure costs.
Bolna is a voice AI startup.
Wingify is a Delhi-based SaaS company majority owned by private equity giant Everyone Capital.
Yotta Data Services, CtrlS Datacenters, E2E Networks, and Reliance Jio are ramping up GPU clusters and AI cloud offerings.