How Razorpay Became India’s Largest Payments Company
In a Nutshell
Razorpay, India's largest payments company, grew from a YC-backed startup in 2015 by solving painful digital payment integration for Indian startups, pivoting from education to co-working spaces, enduring regulatory delays and near-death bank shutdowns through relentless customer trust via human support, and betting early on UPI ahead of competitors. Capital-efficient growth (40x pre-pandemic with near-zero burn), deep customer conviction, and founder ownership of product vision fueled compounding success, while recent AI reinvention enables faster adaptation in a commoditizing market. Key lesson: Build moats in regulated spaces, prioritize daily value over growth hype, and stay in "founder mode" for 10-year problems you're obsessed with.
These notes were generated by AI and may contain inaccuracies.
B2B is a business of trust and nothing replaces the human touch point of trust. Picking up the phone and having a human on the other side saying "yes I am with you I am managing it" gives a lot of trust.
Razorpay went through Y Combinator (YC) in winter 2015, the first Indian company YC invested in. It is now India's largest payments platform, grown at breakneck pace. Early struggles offer lessons for founders.
Harshil was a techie who loved coding from school and college. Had no idea about finance or payments before starting Razorpay. After graduating from IIT, got placed in an oil company in the Middle East, didn't enjoy it. Spent weekends and evenings coding side projects. Built a social crowdfunding platform needing payments. Standard bank channels were complicated to accept payments in India—hard to get started, poor experience not built for techies but for large companies.
Typical techie mentality: how hard could this be?
Realized it seemed like a simple problem nobody was solving. Checked Bangalore startups, Facebook groups—everyone had the same problem. Easier to accept cash than digital payments in India, reversing democratization (social media gives voice, internet channels, e-commerce selling points).
If I knew what I knew today, maybe I would not have started.
Wanted to build online payments. Chose education institutes because e-commerce seemed smaller/harder; education had large fees payments mostly physical/non-digital. Applied to YC pitching payments for education institutes.
In Jaipur, pitched small universities to accept fees digitally. Response: "If they pay digitally, they're going to pay you 1% extra, so why don't I charge 1% more fees?" Education institutes didn't care about digital collections—students pay any way required (cash, check, counter).
Pivoted to startups in co-working spaces who wanted digital payments and gave traction. Pivoted completely away from education.
Spent 3 months in YC without a single live transaction—no approval from partners. Got in-principle approval before YC; took a year after for first live transaction. Rare gestation period—most tech businesses sell from bedroom; YC recommends sell first.
Needed license, certifications, approvals. Hurdles become moat: fewer competitors because hard to build payment gateway, get approvals. Regulation fair—every company follows same guidelines, one-year process. Few have patience/energy.
Regulated businesses hard to comply with rules but create deep moat; everyone after must do same, allows differentiation on top.
Every month questioned building in payments vs. e-commerce/food delivery. Connected deeply because spoke to founders/customers who confirmed problem, nobody solving it.
If hard to sell to customers, it's a problem. If hard for other reasons (regulation), it's a moat—nobody else enters. Customers love it once built.
Customers gave energy: "Finally something solves for us." As long as enough people want it (YC: make something people want), good shape. Focus on customers, not investors/regulators.
Week before demo day, got final approval, went live, TechCrunch launch. 2 weeks after, bank pulled plug due to customer complaint—platform shut for 50 live merchants.
Hardest time: customers want to collect money; had to tell them business shut. Begged bank for time, no. Risked reputation post-YC launch.
Principle: Call every customer, explain exactly what's happening and what doing about it. Picked up all calls, heard abuses (full Hindi), let vent frustration. Took 4-5 days to get new partner, relive merchants. Some abusers still customers today.
B2B (especially finance) business of trust—human touch irreplaceable. Rule: If support crosses 2-4 exchanges, pick up phone. Human saying "I am with you, managing it" builds trust > efficiency.
Use AI everywhere except customer support—it's channel to establish trust that humans spend time solving problems.
Crisis taught: Even in worst, picking phone and saying "I'm on top of it" builds trust.
Early 2015: No UPI, no demonetization. Competitors (overseas/local) threatened: "We're in space, no background, we'll crush you, sell to us."
In YC, pre-launch acquisition offers from largest global payment companies: "Join us, build India for us." Exciting companies, but they didn't understand India's complexity.
India payments $60B in 2015 (now Razorpay alone $180B; India to $1T+). Hard to convince globals needing massive investment for scaling market.
To solve India, long-term perspective needed. No partner understood that.
Unless achieve goals faster with partner, no reason to join.
Grew 40x pre-pandemic with record capital efficiency, lean.
Series A: Raised $10-11M, burn <$200K/month. Put in bank FDs—interest > burn, became profitable. Investor unhappy: "Given money to burn/grow, why profiting?" (Invested in high-burn consumer cos.).
B2B logical: Add value, business pays. No burn needed for engagement like B2C (spend to acquire/engage, monetize later).
In B2B, business uses you today for value; stops if no value. Moats in product/value added.
Could burn more/grow faster, but no point.
It's not a DAO business where you can assign a fixed future value. The business has as much value as the value added to customers today. If you stop adding value, customers will leave tomorrow. This makes the business logical but intense, as every day you fight for your share. The customer evaluates daily if you're adding more value than you're charging. If not, they'll switch. Compounding value is harder as the organization grows.
UPI launched in April 2016 (10-year anniversary recently). Many skeptics doubted its success; the two largest banks hadn't integrated until demonetization in November 2016. Most payment gateways didn't integrate UPI, assuming low adoption without major banks.
Razorpay, small with 10,000 merchants after launching in 2015, had nothing to lose. In September-October 2016, they became the first payment gateway to go live on UPI, ahead of major banks. Post-demonetization, no other gateway scaled UPI support. Major companies like Zomato, Swiggy, BookMyShow switched to Razorpay in weeks for UPI.
Being small allowed early bets: worst case, UPI fails and they continue fighting; best case, it creates a wedge. It took 6 months for competitors to launch UPI, enabling Razorpay to enter new spaces and compound growth.
That's such a classic lesson... being able to be faster, your rate of building product, your rate of learning, your ability to adapt to markets and those advantages accrue to startups.
AI shifts product thinking and company operations. Leadership spends significant time on AI; speaker is addicted to Claude Code, OpenClaude, meeting Anthropic team, suggesting de-addiction programs.
AI lets founders return to building and shipping, bypassing people management. They reimagined Razorpay from scratch: integration, onboarding, support, platform interactions. Rejected incumbent mindset—market changes faster than responses.
Acted like a startup, not incumbent. Reinvented platform end-to-end, launched months ago with massive traction. AI evolves daily; bet on its improvement despite imperfections.
Rewired teams, accepting short-term pain for long-term survival. Hard with large capital and teams facing incumbent fallacy.
AI compresses build time; differentiation becomes speed of deciding what to build. Build execution will be easy; focus on what, how, and design. Companies responding to market changes are already dead—must anticipate and move first.
This is true founder mode right here.
From apartment in Jaipur to generational company. Founders become better versions through intensity.
Mistakes over 10 years: shifted to manager mode after hiring executives, assuming leaders handle details. Worst mistake—fine line between micromanagement and founder mode.
Founder must own core elements like product vision and direction; no leader matches founder's conviction. Nobody cares about the company as much as the founder, from day 1 to year 20 (e.g., Facebook).
Took 10 years to learn: delegate non-core, stay in essentials.
AI eases tools but not entrepreneurship. Core is deep problem connection, dedicating all time to solving it.
Risk: easier building leads to latching onto uncared-about problems. Ask: Can you spend next 10 years building this? Building/deploying is easy; company-building takes 10 years daily.
Find a problem you connect deeply with—not just cool AI projects.
Inspiring. Everyone give it up for Harshil Mathur from Razorpay.
Keep Y Combinator in your library
Save the videos and channels worth coming back to, and find them again in one place.
