Somewhere in the world, a chest X-ray runs through DeepTek’s artificial-intelligence engine roughly every five seconds, and in a large field evaluation its tuberculosis tool was reported to match radiologists with more than 30 years of experience. Yet the Pune company that built it booked only about ₹35.1 crore (about $3.7 million) in operating revenue in FY25 — and that figure went down, not up.
That gap between clinical reach and commercial scale is the whole story of DeepTek. It has a US FDA clearance, a World Health Organization recommendation, a national deployment in Singapore and Japanese strategic backers most Indian healthtech startups never get near. It also has a tiny top line, a foreign parent, and revenue that has slipped for two years running. This is a deep dive into how a radiology AI company can be world-class in the lab and still small on the balance sheet.
Quick facts
| Company | DeepTek Medical Imaging Private Limited (brand: DeepTek / DeepTek.ai), Pune. CIN U74999PN2018FTC177289 |
| Founded | Brand founded 2017; Indian entity incorporated 18 June 2018 (registered as a subsidiary of a foreign company) |
| Founders | Ajit Patil and Dr Amit Kharat (co-founders); Aniruddha Pant (co-founder, data science) |
| Businesses | Cloud PACS + radiology AI (Augmento), chest X-ray AI (Augmento X-Ray), AI TB screening (Genki), teleradiology reporting |
| Latest FY revenue | ≈ ₹35.1 crore in FY25 (year to 31 March 2025), down about 8% YoY (Tracxn, from MCA filings) |
| Latest FY profit/loss | Net profit reported positive but down 52.8% YoY in FY25; EBITDA down 43.4% YoY (Tofler, from MCA filings) |
| Listed | Private (unlisted) |
| Market value / last valuation | Not publicly disclosed; Series A post-money valuation was not announced |
| Key shareholders / CEO | Dr Amit Kharat (co-founder, CEO); backers include NTT DATA, Nobori, Doctor-Net (Japan), Tata Capital Healthcare Fund II, Pentathlon Ventures, GHV |
What DeepTek does
DeepTek sells software that helps radiology departments read scans faster and screen large populations for disease. Its customers are hospitals, diagnostic-imaging chains, teleradiology providers and public-health programmes, in India and abroad. The company positions itself as a radiology AI firm with a US FDA-cleared product suite rather than a single-algorithm vendor.
- Augmento Enterprise: a cloud picture archiving and communication system (PACS) with a built-in viewer for CT, MR and PET-CT, analytics and an AI deployment layer. It received US FDA 510(k) clearance in May 2023 (clearance K222781, company-stated and listed on the FDA database).
- Augmento X-Ray: a US FDA-cleared chest X-ray AI that auto-segregates scans and generates structured reports; showcased at HIMSS 2024 (Imaging Technology News).
- Genki: an AI lung-health and TB screening tool for public health programmes, recommended by the WHO as meeting its performance standard for computer-aided detection (CAD) in TB screening (company-stated, citing the WHO recommendation).
- Teleradiology: central reporting services that route scans to radiologists, with AI triage layered on top.
Company-stated scale figures: over 500 hospitals and imaging centres served, 2,000-plus sites globally, more than 3 million lives touched, and a presence in 30-plus countries, with the company saying a scan is processed roughly every five seconds across its network (DeepTek.ai, as of September 2026). These are self-reported and should be read as such.
The founding insight
The founding bet was simple to state and hard to build: radiology has too many scans and too few radiologists, and India feels that shortage more acutely than most. If AI could triage images and a cloud platform could move them to whoever was free to read them, the same pool of doctors could cover far more patients, including in places that have never had a radiologist on site.
The three founders came at that problem from different sides. Ajit Patil holds a master’s from IIT Kharagpur in industrial engineering and operations research and had already built and exited a company: Vertex Software, a Japan-focused IT services firm with a strategic tie to Mitsui that was ultimately absorbed by NTT DATA Japan. Aniruddha Pant, an alumnus of the College of Engineering, Pune and a PhD from the University of California, Berkeley, brought the deep-learning and medical-image-processing expertise. Dr Amit Kharat, a radiologist (MBBS, DMRD, DNB) with a PhD in musculoskeletal imaging, has published more than 60 scientific papers and is named on patents; he became the clinical anchor and CEO. Patil and Pant already knew each other from Pune; Patil met Kharat while scouting teleradiology companies for his next venture. That mix — an operator with a Japan network, an AI researcher and a practising radiologist — shaped everything that followed, including where the early money came from.
The struggle years
DeepTek’s hard problem was never only the model; it was distribution and trust. Selling AI into radiology means clearing regulators, convincing clinicians who are liable for a missed finding, and integrating with hospital IT that was not built for cloud. A young Pune startup with no brand had to do all three at once, in a market where buyers are conservative and sales cycles are long.
- Early revenue was thin. As of February 2020, YourStory reported DeepTek’s annualised revenue at just over $1 million, with the platform live across about 50 hospitals and imaging centres — respectable for a startup barely two years old, but a reminder of how small the base was.
- The business leaned heavily on services. Teleradiology reporting and screening projects are labour- and operations-heavy, which caps the software-style margins investors prize and makes revenue lumpy from project to project.
- Regulatory dependence cut both ways. Public-health screening work depends on government tenders and programme funding, which can appear and disappear with the budget cycle rather than the sales pipeline.
The MCA filings show the strain in the numbers. After a strong FY23 (operating revenue up 96.9% year on year, per aggregators reading the filings), growth reversed: FY25 revenue of about ₹35.1 crore was roughly 8% lower than the prior year (Tracxn), and reported net profit fell 52.8% while EBITDA fell 43.4% (Tofler). A company can be celebrated at global health conferences and still watch its Indian top line shrink; DeepTek did both at once.
The turning point
If there is a single hinge in DeepTek’s story, it is the sequence in 2022-2023 when it stopped being an India teleradiology startup and became a regulator-cleared, globally-deployed platform. Two events did most of the work.
First, the money. In March 2022 DeepTek raised a $10 million Series A led by Tata Capital Healthcare Fund II — the fund’s first healthtech investment — with Pentathlon Ventures and GHV participating (Inc42, reporting the round announced 7 March 2022). That capital came on top of years of strategic backing from Japan. Second, the credibility. In May 2023 Augmento received US FDA 510(k) clearance, and the same platform was adopted in Singapore, where DeepTek says Augmento was deployed as the national radiology AI platform through a partnership with Synapxe, the country’s health-tech agency. Put the numbers on each side: before, a sub-$1-million-plus startup selling teleradiology to Indian centres; after, an FDA-cleared, WHO-recommended vendor with a national government customer abroad and 500-plus sites claimed. The clinical proof point that made the screening pitch real came in the field — a Genki deployment in the Greater Chennai area was reported to have screened 25,598 people using four mobile units, at 98% sensitivity and 96.9% accuracy (company-stated impact data). The reputation arrived; the revenue, as the filings show, has not yet caught up.
The money behind it
DeepTek’s cap table is unusual for an Indian startup: its earliest serious money came from Japanese strategic investors, not domestic VCs, which is also why the Indian company is registered as a subsidiary of a foreign parent (DeepTek Inc).
- Strategic equity from Japan: NTT DATA took a strategic equity stake announced in October 2018, bringing global market access and one of the larger commercial medical-imaging databases in North America (The Bridge Chronicle). Nobori Ltd and Doctor-Net Inc, both Japan, are also named as backers (Inc42).
- Series A — $10 million, announced 7 March 2022: led by Tata Capital Healthcare Fund II, with Pentathlon Ventures and GHV participating (Inc42, Crunchbase).
- Total raised: reported at roughly $10 million by Tracxn and about $10.01 million by CB Insights, while PitchBook lists about $11.5 million. Sources differ, so treat the total as approximately $10-11.5 million across rounds.
- Valuation: the Series A post-money valuation was not publicly disclosed, and no later priced round has been reported.
What each backer changed: NTT DATA gave DeepTek an international distribution rail and imaging-data heft early, when it had neither; Tata Capital’s healthcare fund brought domestic institutional validation and growth capital in 2022; Pentathlon Ventures and GHV added early-stage healthtech and operator support. The Japan connection is not incidental — co-founder Ajit Patil’s previous company was itself absorbed by NTT DATA Japan.
How it makes money
DeepTek earns from a mix of software subscriptions and services, which is the tension at the heart of its economics. The parts that scale cleanly (software) sit alongside parts that do not (reporting and screening operations).
- Platform/AI subscriptions: recurring fees for Augmento PACS and the AI modules, sold to hospitals and imaging centres. This is the high-margin, scalable layer the company wants to grow.
- Teleradiology reporting: fee-per-study or contracted reporting revenue, where DeepTek’s AI triages and radiologists sign off. Margin here is limited by the cost of radiologist time.
- Public-health screening (Genki): programme and tender-based revenue for TB and lung-health screening, often tied to government or NGO funding and to mobile-unit operations — impactful, but budget-cycle dependent and operationally heavy.
- Strategic/international deals: national and enterprise deployments such as Singapore’s Synapxe partnership, which bring reference value beyond the immediate contract.
The part people get wrong: because DeepTek’s clinical validation and marketing are so strong, it is easy to assume it is a pure, high-margin software company. The FY25 filings suggest the reality is a blended model where services and screening still weigh on margins, which is one reason a globally-cleared product still maps to a sub-₹40-crore Indian top line. Specific per-scan take rates and subscription price points are not publicly disclosed, so they are left out here rather than guessed.
The numbers
Hard, exact rupee figures for DeepTek are limited because it is a private subsidiary and most line items sit behind MCA-filing paywalls. What is verifiable, from aggregators reading those filings, is the FY25 revenue level and the direction of travel. Units are ₹ crore, for the Indian entity (year ending 31 March).
| Fiscal year | Operating revenue | Profitability (as reported) |
| FY23 | In the ₹1-100 crore band; grew 96.9% YoY (aggregators, from MCA filings) | Net profit up 110.8% YoY; EBITDA up 49.8% YoY |
| FY24 | ≈ ₹38 crore (implied by the FY25 decline; base year) | Base year for the FY25 fall |
| FY25 | ≈ ₹35.1 crore, down ~8% YoY (Tracxn) | Net profit positive but down 52.8% YoY; EBITDA down 43.4% YoY (Tofler) |
Two things stand out. First, DeepTek is reported to be profitable — rare for an AI healthtech at this stage — but that profit shrank sharply in FY25. Second, the top line has contracted for consecutive years after the FY23 surge, so the growth story is currently a stabilisation story. The FY24 figure above is an estimate derived from Tracxn’s reported ~8% FY25 decline against a ₹35.1 crore base; it is labelled as such and should not be read as a filed number.
Where the money comes from
DeepTek does not publish a formal segment or geography split, so an exact breakdown is not verifiable. What the public record does establish is that revenue is spread across several geographies and buyer types rather than concentrated in one.
- Geography: India is the home base, but the company reports customers across Japan, APAC, Singapore, Kenya and 30-plus countries in total (DeepTek.ai; Inc42 noted 150-plus customers across India, Japan and APAC at the time of the Series A).
- Buyer type: private hospitals and imaging chains (subscriptions and teleradiology) on one side; government and public-health programmes (Genki TB screening) on the other.
- Marquee deployment: Singapore’s national radiology AI platform via Synapxe — a government-scale reference customer disproportionate to its likely revenue contribution.
The surprise is the Japan weighting. Most Indian radiology-AI peers are India- or US-first; DeepTek’s origins, its earliest strategic investor (NTT DATA) and its founder’s history point to Japan as a structurally important market, which is unusual and hard for competitors to replicate quickly.
The risks
- Shrinking top line: revenue fell roughly 8% in FY25 to about ₹35.1 crore, and reported net profit dropped 52.8% (Tracxn, Tofler). If the decline continues, the gap between DeepTek’s global reputation and its financial scale widens, making the next priced fundraise harder and the last-disclosed round ($10 million, March 2022) look increasingly distant.
- Services and tender dependence: a meaningful share of activity is teleradiology reporting and government-funded screening. Reporting margins are capped by radiologist cost, and screening revenue rides government budget cycles and tenders — both of which are lumpier and lower-margin than pure software.
- Regulatory and clinical liability: the whole business rests on clearances (US FDA 510(k), CE, WHO recommendation for TB CAD). Any adverse clinical finding, a missed diagnosis attributed to the AI, or a tightening of medical-device AI rules in a key market could stall deployments quickly, because buyers are liability-sensitive by nature.
- Competitive pressure: DeepTek competes with well-funded radiology-AI rivals, notably Qure.ai, which has raised far more capital. In a market where distribution and capital compound, a smaller balance sheet is itself a risk.
The takeaway
DeepTek is a case study in a quiet truth about deep-tech: clinical excellence and commercial scale are different games, and winning the first does not hand you the second. The company has done the genuinely hard part — building AI that regulators clear and the WHO will vouch for, and getting it adopted at national scale abroad. What it has not yet done is turn that proof into compounding revenue; the FY25 filings show a small, profitable business whose top line is drifting down, not up. The transferable lesson is that for regulated, life-critical products, validation is the entry ticket, not the finish line. The founders who eventually win are the ones who treat distribution, pricing and margin structure as problems just as serious as the model itself. DeepTek has earned the right to be believed on the medicine; its next chapter is whether it can be believed on the business.
Frequently asked questions
What does DeepTek do?
DeepTek builds radiology AI and cloud imaging software. Its Augmento platform is an FDA-cleared PACS with AI for reading CT, MR and X-ray scans; Genki is an AI tuberculosis-screening tool recommended by the WHO; and it also runs AI-assisted teleradiology reporting. Customers are hospitals, imaging centres and public-health programmes in 30-plus countries.
Who founded DeepTek and when?
The brand was founded in 2017 by Ajit Patil, Dr Amit Kharat and Aniruddha Pant; the Indian entity, DeepTek Medical Imaging Private Limited, was incorporated on 18 June 2018 in Pune. Dr Amit Kharat, a radiologist, is CEO.
How much funding has DeepTek raised?
DeepTek raised a $10 million Series A led by Tata Capital Healthcare Fund II in March 2022, with Pentathlon Ventures and GHV. Total funding is reported at roughly $10-11.5 million across rounds (Tracxn, CB Insights, PitchBook differ). It also had earlier strategic equity from NTT DATA (Japan). The valuation has not been disclosed.
Is DeepTek profitable?
Based on MCA-filing data via aggregators, DeepTek’s Indian entity reported a net profit in FY25, but that profit fell about 52.8% year on year, and operating revenue slipped roughly 8% to about ₹35.1 crore. So it is profitable but shrinking, not scaling, on current figures.
Who are DeepTek’s main competitors?
Its most prominent competitor is Qure.ai, another Indian radiology-AI company that has raised substantially more capital. DeepTek differentiates on FDA/WHO clearances, its Augmento PACS platform and a strong Japan and public-health-screening presence.
Sources
Figures are as of September 2026. Currency converted at $1 ≈ ₹96.0 as of 18 September 2026 (Trading Economics).
- Inc42 — “Tata Capital Healthcare Fund marks first healthtech investment with $10 Mn infusion in DeepTek” (March 2022)
- YourStory — “Healthcare startup DeepTek records over $1M in revenue within a year of starting up” (February 2020)
- Tracxn — Deeptek Medical Imaging Private Limited, company and legal-entity profiles, FY25 revenue and headcount (2026)
- Tofler — Deeptek Medical Imaging Private Limited financials, directors, capital, FY25 profit/EBITDA changes (2026)
- The Bridge Chronicle — “City-based DeepTek enters into strategic partnership with NTT Data” (October 2018)
- Imaging Technology News (ITN) — DeepTek.ai Augmento receives FDA clearance (2023); Augmento X-Ray at HIMSS 2024 (2024)
- US FDA 510(k) database — Deeptek Medical Imaging Private Limited, clearance K222781 (2023)
- DeepTek.ai — company website: products (Augmento, Augmento X-Ray, Genki, Teleradiology), Genki/WHO recommendation, Singapore/Synapxe deployment, scale claims, impact data (2026)
- Crunchbase; CB Insights; PitchBook — funding totals and Series A details (2026)
- Trading Economics — USD/INR reference rate (September 2026)
Found an error? Write to us and we’ll correct it in the open, dated, on the piece.