Fusion CX is an IPO-bound customer experience and BPM company. Its AI-enabled services and data infrastructure business should rise from 10% to 25% of turnover within 18 months, it says. Its FY26 revenue was ₹1,818 crore. The Fusion CX AI push rests on a simple wager: as models get cheaper, the durable money moves to the data they are trained on.
It is a serious bet from a company with real numbers behind it. But the target has moved, and the pivot is also a pitch to public-market investors.
What the Fusion CX AI business actually is
The AI portfolio sits inside a subsidiary called Omind and covers three layers. The first is customer experience, through MindVoice, MindSpeech and an AI quality-management tool called Arya.
Two others matter more. Annotera.ai labels text, image, audio and video data and supports reinforcement learning from human feedback, the process that teaches models which answers are better. Roborax.ai captures teleoperation, human demonstration and multimodal sensor data, including LiDAR, for teams building humanoid, surgical, industrial and logistics robots.
Fusion CX also says it passes 60–70% of the savings AI delivers back to clients. That is an unusual disclosure, and it signals how the value is shared.
The numbers, and a target that moved
The company’s headline financials are strong. In FY26, revenue from operations rose 36.8% to ₹1,818 crore. Profit after tax more than doubled to ₹170 crore, and EBITDA margin improved to 18.14% from 14.88%. Its three-year CAGR is close to 28%.
The AI share is where attention belongs. The October release says AI-enabled services are now 10% of turnover and should reach 25% within 12–18 months. In July, the same company told The Hindu BusinessLine it earned “less than 3%” from AI-enabled services and aimed for “20–25%” over “three to five years”.
Those are different claims, six weeks apart. The endpoint is similar, but the starting point and the timeline are not. The company has not explained the change.
There is also a size check. Dhanuka has estimated the AI data infrastructure business could earn about ₹160 crore in FY27 and exceed ₹600 crore by FY28. Against ₹1,818 crore of total revenue, that is a meaningful but not yet dominant line — and it is a forecast, not a signed book.
Why the data layer matters
The logic behind the pivot is sound. The AI industry spent billions on chips and models. Then it found that the scarce input is human-generated data — the examples and rankings that teach models to reason.
That market has been reshaped in the past year. When Meta took a roughly 49% stake in Scale AI in 2025, the data-labeling leader lost the neutrality that made it the default vendor. OpenAI, Google and xAI pulled back. Demand flowed to rivals: Surge AI reported annual recurring revenue near $1.4 billion, and Mercor’s annualised revenue crossed $2 billion by mid-2026.
The category is also concentrated. More than 75% of its revenue is estimated to sit with just four firms. Fusion CX is positioning itself as a neutral alternative, which is precisely the quality the market now values.
The robotics angle may be the sharpest. Robotics companies raised over $10 billion in 2025, yet their models train on fewer than 5,000 hours of open-source real-world interaction data. Capturing that data requires physical work — teleoperation rigs, staged environments, sensor fusion — that cannot be crowdsourced from a laptop. Fusion CX’s 28-language, 40-centre footprint is an unusual asset for that task.
The BPM pivot behind the IPO
The context is India’s BPM industry, which is being remade by AI. At the Nasscom BPM Confluence on 1 October, president Rajesh Nambiar said AI is driving the sector’s biggest reinvention. India, he added, could become the world’s “intelligent operations hub”. Genpact’s advanced-technology revenue grew 24% last quarter. EXL has warned that 60–70% of traditional service jobs may disappear while data and AI roles grow.
Fusion CX is making the same turn, but from a smaller base and with a public offering attached. The IPO is worth ₹1,000 crore: a ₹600 crore fresh issue and a ₹400 crore offer for sale. Proceeds will repay debt, fund AI platforms and buy more companies. It has made 16 acquisitions since 2008 and is planning two more.
That is the tension worth naming. The AI pivot is genuine and the market is real. But the timing — a fresh AI narrative arriving weeks before an IPO — means investors should read the target with care.
What to watch
Three markers will show whether the AI bet is substance or story.
First, the definition. If Fusion CX discloses how it counts “AI-enabled services” and reconciles the July and October figures, the target becomes credible. If it does not, treat the number as marketing.
Second, the margins. A data-annotation business can be low-margin if it is labour-heavy. Watch whether the AI segment lifts group profitability or dilutes it.
Third, the delivery. ₹600 crore by FY28 is a forecast. Watch whether it shows up in booked revenue and named customers.
The Fusion CX AI bet chases the right market from an unusual position. The question is whether the numbers catch up with the narrative — or whether the narrative was built for the roadshow.
Editor’s Note
Sources: Fusion CX‘s release of 3 October 2026 and its investor-relations materials, including its draft red herring prospectus disclosures. Financial figures are from ThePrint, The Hindu BusinessLine, Outlook Money and Bajaj Broking’s DRHP coverage. The July 2026 AI-share and timeline figures are from The Hindu BusinessLine’s interview with Pankaj Dhanuka. The VA Platinum acquisition is from Outsource Accelerator, citybiz and Bar & Bench.
Market context on the AI data-labeling industry is from Reuters, Forbes, and independent vendor comparisons; robotics-data context is from Forbes and robotics-data research. India’s BPM context is from People Matters’ report on the Nasscom BPM Confluence 2026, Genpact’s Q2 2026 results, and Economic Times reporting. The AI revenue-share figures, the FY28 target and the acquisition plans are as stated by Fusion CX. TechRecast has not independently verified them, and the IPO remains subject to market conditions.