Soket AI Labs is a Gurugram-based Indian AI research company building open-source multilingual foundation models and speech systems for Indian languages.

This Startup Deep Dive explains what Soket AI Labs does, how it makes money, who backs it, how it compares with rivals and what could go wrong, using publicly available information. It is educational and not investment advice. Last updated: October 2026.

Soket AI Labs in Brief

• What it is: Soket AI Labs is an Indian AI foundation models company based in Gurugram.
• Who it serves: businesses, institutions, consumers, learners and public-sector buyers.
• Develops open-source, multilingual foundation models with a focus on Indian languages.
• Project EKA is described as a 120B+ parameter sparse mixture-of-experts model for math, code, reasoning and multilingual AI, backed by the IndiaAI Mission.
• Builds speech products including DHRITH (emotion-aware speech recognition) and TensorStudio (realtime speech API for voice agents).
• Funding: Reports differ: described as bootstrapped by one source, about $2.2 million by others, $2.35 million seed by another.

Soket AI Labs Quick Facts

Figures marked “reported” come from published reports and may differ between sources; “Not disclosed” means no reliable public figure was found.

What Is Soket AI Labs?

Soket AI Labs operates in the AI foundation models space in India. Understanding a company like this starts with three questions: who is the customer, what painful problem is being solved, and why the solution is better than the alternatives. The sections below apply that lens to Soket AI Labs.

• Develops open-source, multilingual foundation models with a focus on Indian languages.
• Project EKA is described as a 120B+ parameter sparse mixture-of-experts model for math, code, reasoning and multilingual AI, backed by the IndiaAI Mission.
• Builds speech products including DHRITH (emotion-aware speech recognition) and TensorStudio (realtime speech API for voice agents).
• Other offerings include Pragna-1B (open-source multilingual LLM), the Bhasha dataset series and a developer preview of LOOP, an open-source agent harness.

The Origin Story of Soket AI Labs

Every company starts from a noticed gap and a team willing to commit to closing it. Strong origin stories are specific: a clear problem, a founding team with relevant skills, and an early product that real customers could test. Founder-market fit, meaning the team’s lived experience or technical depth in the domain, tends to predict how fast a young company learns.

• Founders: Abhishek Upperwal.
• Founded in 2019 by Abhishek Upperwal, a former data scientist with a Master's in Computational Data Science from IISc Bangalore who previously worked with India's Smart Cities Mission.
• Registered as Soket Labs Technology and Research Private Limited; offices are listed at the NASSCOM Startup Warehouse in Udyog Vihar, Gurugram (one article calls it Bengaluru-based).
• Was selected under the IndiaAI Mission to build a sovereign model.

Key Milestones and Challenges

No startup grows in a straight line. Early years usually involve slow adoption, product rewrites, hiring mistakes and funding gaps. In the AI foundation models sector the common turning points are a first paying customer, a regulatory approval, a successful product demonstration or a funding round that lets the team invest ahead of demand. How a team responds to setbacks says more about resilience than any single headline.

• 2019: company founded.
• March 2024: filing shows founder Abhishek Upperwal holding about 75 percent.
• Released Pragna-1B open-source multilingual model.
• Selected under the IndiaAI Mission for the Project EKA sovereign model.
• Gained access to 1,536 NVIDIA H100 GPUs through the IndiaAI Mission.
• 2026: reported about $2.2 million funding round (latest round listed as a grant by one tracker).

Soket AI Labs Funding and Investors

Funding history shows how much outside confidence a company has attracted and how long it can operate before needing more capital. Figures below are reported and may differ between sources; valuation figures are estimates unless the company has confirmed them.

• Reports differ: one source lists the company as bootstrapped; another reports a $2.35 million seed with Lightspeed India and Together Fund; another reports about $2.2 million total with the latest round dated May 2026 and investors not shown.

How Does Soket AI Labs Make Money? (Business Model)

A business model describes who pays, how often and how much it costs to serve them. Strong models show repeat usage, improving margins with scale and acquisition channels that do not depend only on paid spend. Weak models show one-off sales, rising costs and growth that stops when spending stops. For companies selling to businesses, institutions, consumers, learners and public-sector buyers, sales cycles and contract size matter as much as headline growth.

• B2B and government focus: targets Indian IT services firms, government departments, public-sector enterprises, defence, agriculture and financial services.
• Sells model access, speech APIs and customised AI solutions.
• Third-party estimate puts revenue near $1.1 million, not verified.

The Numbers: Soket AI Labs Financials

Financial figures show whether growth is turning into a self-sustaining business. Small and private companies often disclose little, so any number below is included only if it was found in public reporting.

• Total expenses of about Rs 61.1 lakh in the latest financial year per one tracker, up 878 percent.

Market Opportunity for Soket AI Labs

Market size is easy to exaggerate, so a useful approach separates the total market from the portion a company can realistically reach with its current product and distribution. The main forces behind demand in this sector are rising digital adoption, cloud and mobile penetration, growing enterprise technology spending, government digitisation programmes and a deep pool of engineering talent. Each of these widens the opportunity, but capital efficiency and execution decide who captures it.

• Targets the gap in AI performance for Indian languages, a sovereign-AI opportunity under the IndiaAI Mission.
• Customers include government and enterprise buyers in India.

Technology and Product at Soket AI Labs

Technology is rarely the whole moat, but it drives speed, cost and customer experience. A practical way to assess it is to separate the capability the company controls from the commodity infrastructure it rents. In this sector the important factors are product quality, data and integration depth, security, scalability and the speed of shipping improvements.

• Sparse mixture-of-experts large language model (EKA) at 120B+ parameters.
• Speech recognition that accounts for emotional context and a realtime speech API.
• Open-source agent harness LOOP for long-running AI agents.

Team and Culture at Soket AI Labs

Teams are the most important and least measurable asset of a young company. Look at how complementary the founders’ skills are, how deep the second layer of leadership is, and how the company hires and retains specialists. Culture shows up in small repeated behaviours: how quickly bad news travels upward and whether commitments to customers are kept.

• Team size estimates range from about 10 to 17 employees.
• Partners listed include IIT Gandhinagar, IIT Roorkee, IISc Bangalore, AI4Bharat, ICRISAT and Google Cloud.

Soket AI Labs Competitors: Competitive Landscape

A fair competitive analysis names direct rivals, adjacent players who could move into the space, and the customer’s default alternative. Durable advantages include proprietary data or technology, licences and approvals, brand trust, switching costs and cost leadership.

Risks and Challenges for Soket AI Labs

• Funding and investor details are inconsistent across sources.
• Small team competing against better funded rivals for compute and talent.
• Depends on government programmes such as the IndiaAI Mission for compute and credibility.
• Competition: low barriers to entry and well-funded rivals can compress pricing and slow growth.
• Funding risk: technology businesses often burn cash before reaching profit and depend on investor appetite.

Risk is not a reason to avoid a company; it is a reason to price it correctly and to monitor the right indicators over time. Regulation in this sector is shaped by data protection rules under the Digital Personal Data Protection Act, sector regulators, consumer protection law and IT rules.

Soket AI Labs Growth Outlook: What to Watch

• Success hinges on delivering Project EKA and converting it into commercial adoption.
• Government and enterprise demand for sovereign Indic AI is a tailwind.
• Key indicators: recurring revenue, customer retention, gross margin, customer acquisition cost and runway.
• Funding position: Whether Soket AI Labs can reach its next milestone before needing new capital.
• Regulation: Changes from data protection rules under the Digital Personal Data Protection Act, sector regulators, consumer protection law and IT rules.

Industry Context: Where Soket AI Labs Fits

India's technology startup ecosystem is one of the largest in the world, supported by cheap data, a large base of smartphone users, digital public infrastructure and a deep engineering talent pool. Companies now build for education, software, mobility, logistics, agriculture, defence, media and more, selling both to Indian customers and abroad.

Durable technology companies usually combine a clear customer problem with repeatable distribution and disciplined unit economics. Growth funded only by spending tends to fade, while products that customers keep using and paying for build compounding advantages.

For readers new to this field, a useful habit is to separate announcements from evidence. A plan, a partnership or a funding headline describes intent, while a delivered product, a paying customer or an audited financial statement describes results. Comparing the two over time shows how reliably a company turns promises into outcomes.

Bull Case and Bear Case for Soket AI Labs

An investment thesis is a short, testable statement of why a company could become far more valuable, together with what evidence would prove it wrong. The points below are general to this sector and are not forecasts for Soket AI Labs.

Bull case

• A large and growing base of digital users and businesses expands the addressable market.
• Recurring revenue and strong retention can compound growth.
• Selling into global markets widens the opportunity beyond India.

Bear case

• Heavy competition can compress pricing and margins.
• Dependence on external funding can force painful cost cuts.
• Regulation or platform changes can disrupt a product quickly.

Lessons Founders and Investors Can Take from Soket AI Labs

Studying a company in the AI foundation models space offers lessons that go beyond any single firm. The points below are general lessons for the sector and are not claims about Soket AI Labs.

• Solving a clear, frequent customer problem matters more than adding features.
• Retention and repeat usage are better health signals than sign-ups.
• Disciplined spending extends runway and keeps options open when funding tightens.
• Distribution, not just product quality, decides who wins in crowded markets.
• Building trust through data protection and reliability is part of the product.

How to Evaluate Soket AI Labs as a Startup

• Customer clarity: Can Soket AI Labs name its customer and the problem it solves in one sentence?
• Evidence of demand: Look for repeat customers, signed contracts or measurable usage rather than announcements alone.
• Economics: Check whether margins improve as volume grows.
• Execution: Compare announced timelines with what has actually been delivered.
• Funding runway: Estimate how long current capital lasts at the present spending rate.
• Regulatory standing: Confirm required approvals are in place.
• Customer concentration: Check whether a few customers account for most of the revenue.
• Management depth: Look for experienced leaders beyond the founders and a stable core team.

The Takeaway

Soket AI Labs is part of India’s growing AI foundation models ecosystem. The useful lesson from studying it is the discipline of asking structured questions about customers, economics, advantage and resilience. Companies that answer them well, repeatedly and under pressure are the ones that endure. Readers should treat any single report, ranking or funding headline as one input among many, and should re-check key facts against recent company announcements. Because public information on young companies is often limited, patience in verifying claims is part of good analysis. Public sources can also describe the same company in different ways, so it is worth checking dates, definitions and whether a figure refers to revenue, total income or transaction value, and whether it covers one entity or a wider group, before drawing conclusions. Small differences in rounding, currency conversion or reporting periods can also change how a figure reads. Comparing the same figure across several recent reports helps show whether it is stable or has changed over time. This article is educational and is not investment advice.

• Clear customers and repeat demand matter more than announcements.
• Regulation and capital needs shape how fast a startup in this sector can grow.
• Track a few leading indicators over time rather than one headline metric.

Soket AI Labs FAQs: Frequently Asked Questions

What is Soket AI Labs?

Soket AI Labs is a Gurugram-based Indian AI research company building open-source multilingual foundation models and speech systems for Indian languages. It is profiled in the Startup Deep Dive series, which explains business model, funding, competitors and risks.

Who founded Soket AI Labs?

Soket AI Labs was founded by Abhishek Upperwal, and the company reports a founding year of 2019.

How much funding has Soket AI Labs raised?

Reports differ: described as bootstrapped by one source, about $2.2 million by others, $2.35 million seed by another. Funding figures are reported and may differ between sources, so readers should check the latest company announcements.

How does Soket AI Labs make money?

Soket AI Labs earns revenue from its AI foundation models products and services sold to businesses, institutions, consumers, learners and public-sector buyers. The exact revenue mix is described in the business model section and is not fully disclosed publicly.

Who are the competitors of Soket AI Labs?

Reported rivals include Sarvam AI, Gnani.ai, Krutrim. See the competitor comparison above for how they differ.

What are the main risks for Soket AI Labs?

Main risks include competition: low barriers to entry and well-funded rivals can compress pricing and slow growth, funding and capital needs, competition and regulation by data protection rules under the Digital Personal Data Protection Act, sector regulators, consumer protection law and IT rules. This is educational content, not investment advice.

Related Startup Deep Dives: Edureka — Raised $2 million from Leo Capital in 2018; FY25 revenue about Rs 72.7 crore with a loss of about Rs 12.5 crore per one tracker | Lemnisk — FY25 revenue reported at Rs 33.2 crore, up 78% from Rs 18.6 crore in FY24. Browse the full Startup Deep Dive series.

Source

Source: Publicly available data, published reports, industry research, and other credible sources.