Covalent: Empowering AI and DA with Expertise in Decentralized Data Indexing
TechFlow Selected TechFlow Selected
Covalent: Empowering AI and DA with Expertise in Decentralized Data Indexing
Covalent's focus is on having all this structured data.
Navigating Web3 tides with focused insightsHost: Blair Zhu, Mint Ventures
Guest: Ganesh Swami, CEO of Covalent
Original English interview link: WEB3 Founders Real Talk with Covalent Recap
Ganesh's Background and Introduction to Covalent
Blair: Hello everyone, welcome back to Web3 Founders Real Talk. Here we connect with real industry disruptors and have honest conversations. Today, we’re thrilled to have Ganesh, CEO of Covalent. Welcome!
Ganesh: Blair, thanks for having me on the show. I’m excited to engage further with the community.
Blair: Thank you so much for joining us. Could you briefly introduce yourself? How did you enter the crypto space, and how did your project get started? Also, please give us a quick overview of Covalent.
Ganesh: I’m Ganesh Swami, one of the founders of Covalent. Covalent has been around for over five years now—definitely one of the older projects in the space. My entry into crypto was quite accidental. There was just this opportunity that pulled me in. I wasn’t working in crypto before; actually, I came from the database world, building data infrastructure. Prior to that, I worked in cancer research—doing physical chemistry research and building antibodies for drug design. I was part of the founding team at one of Canada’s largest biotech companies, which is publicly traded on Nasdaq and has several drugs in clinical trials. That’s my background. I switched because pharmaceuticals take about 10 years to build a minimum viable product—it takes time. Meanwhile, my friends in IT could launch an MVP in two years, go to market, raise funding, and even do M&A. I wanted that kind of pace.
So I moved into data infrastructure. At the time, cloud data warehouses were becoming popular—like Snowflake. Many on-premise workloads were shifting to the cloud. The cloud was becoming a new piece of infrastructure. I helped many companies transition to the cloud. I spent about ten years in a co-working space. One of my mentors told me, “You should go participate in a hackathon for decentralized databases.” This was during the 2017 bull run. I thought, well, I'm in Vancouver, it rains a lot here, so I didn’t have anything better to do on a Saturday—why not check it out? I knew from the database world that no matter what backend database you use, people still want to analyze in Excel. That’s the front end for all databases—whether Oracle, SAP, or Microsoft.
At that hackathon, I built something that could directly import blockchain transactions into Excel. That was the idea. Like a search engine—a Google for blockchains, call it what you will. Back in 2017, there were only ICOs—simple ERC20 transfers. That was it. No DeFi, no NFTs, nothing complex. We ended up winning the hackathon, and then we said, “This is a cool idea—this could unlock a lot.” But where I went wrong was timing. Because the next two years were bear markets—very harsh. So don’t take my advice on market timing; my track record is terrible.
Anyway, we founded a company called Covalent. If you remember high school chemistry, the word “covalent” comes from “covalent bonding”—the way atoms bond together. We’re binding centralized systems with decentralized ones, databases with blockchains, etc. It’s a metaphor. That’s the origin story of Covalent. We started the company essentially treating blockchains as databases—you can query them, index them, do all sorts of operations from blockchains. The first few years were tough.
Then DeFi Summer came along—perfect timing, right product. People say it was an overnight success, but by then we’d already been working on it for about two and a half years. There were some bumps, but overall, that’s our startup story. If it hadn’t rained in Vancouver that Saturday, maybe Covalent wouldn’t exist. It really was that simple. The core idea here is that no matter how infrastructure changes behind the scenes, people won’t relearn or switch tools. They won’t abandon Excel. They won’t retrain existing workflows—the business process must adapt to change.
So people need that bridge proxy—an intermediary device. That’s exactly what Covalent has offered from day one. We never really pivoted or changed direction—that’s always been our model. Today, it goes by different names—some call it an indexer, others a data availability layer, etc. But fundamentally, what we do is make blockchain data more accessible in a decentralized way.
Challenges and Obstacles
Blair: That’s one of the most interesting stories I’ve heard. It all comes down to weather—I’m glad Vancouver rains so much, otherwise Covalent might never have existed. You identified a key product pain point and built this excellent project, which is doing great today. I wonder if you faced any challenges or resistance along the way? Like you mentioned, timing within the broader crypto landscape could be a critical factor. Did you face any technical or macro-level hurdles?
Ganesh: I’m a serial entrepreneur—this is my fourth startup. Any mission-driven startup carries risks, and these risks are often bundled together. There’s market risk—that’s where we failed in the first two or three years. There’s product risk, technical risk, fundraising risk, team risk. All these are intertwined. We faced fundraising risk because during bear markets, no one wants to open their wallets. Market risk because the market didn’t really exist—there were no applications. We built the product anyway. We had strong engineers, a solid team, and my other co-founder Levi has been building databases his whole life—he knows way more than I do. Of course, there’s also fundraising risk, technical risk.
On the flip side, we were lucky because EVM won—basically everything became EVM-compatible. Some teams bet on EOS, Cardano, Solana—they did well. But any project betting on non-EVM chains like XRP or Elrond—all of them failed. Technically, we were very lucky to choose EVM. These are some of the risks, but the harshest ones were fundraising and market risk. After two years of daily grind at Covalent, we operated without external capital. I became disillusioned with the industry. Not just me—many people dropped out during the bear market, like we saw in previous cycles. A mentor suggested I take a break, step away, and see if this industry truly suited me. So I went to climb Mount Everest. It was an extremely difficult journey, but I spent many hours alone—eight, nine, ten hours walking solo. The whole team was there, but I was immersed in my own thoughts, lots of time to reflect. I gained incredible insights in the Himalayas. One was: since we’ve already scraped all this blockchain data, why not leverage it, see its appeal, reach out—see if anyone else needs our product? So we came back.
I returned in October, and from November through February, I pushed through those four months—even through Christmas holidays. We found product-market fit, started generating revenue, gained consensus—that’s when the flywheel began. Again, it was a shift in perspective. We had all the data—we could see the share each protocol held. Why not just reach out to them? I’d say those were major obstacles and challenges. Then there’s another challenge: people don’t understand the value of indexers because all data is public. What’s the difference between Etherscan and Covalent? Between The Graph and Covalent? These persist, but they’re part of the journey. First, I’d say the biggest challenge was getting this thing off the ground—it took us nearly three years to see a glimmer of hope. Those days were long and hard.
Blair: Yes, but that’s also impressive because the field is still new—we still see entrepreneurs struggling with product-market fit. Sometimes I feel founders need to be selective about what they work on—not just driven by interest or personal views.
Ganesh: There’s something called product-founder-market fit.
Differentiation from Other Data Solutions
Blair: Yes, that’s exactly what I wanted to talk about. Can you briefly walk us through your product suite? I see you have a Unified API and GoldRush. Compared to manually handling data retrieval and processing via RPC, how can developers expect cost reduction and efficiency gains using your products? I’m not technically trained, but I think this is a common question—what’s the difference? Also, how does it differ from other blockchain data solutions like The Graph?
Ganesh: Great question. Maybe before diving into different types of data solutions, the key point is: a blockchain is a billboard, not a database. On a billboard, you post something, a week later you take it down and put up something new. That’s what a blockchain is. The entire purpose of a blockchain is to place data in a challenge window to see if there are any disputes. After challenging, you evict it, move on—evolve the state machine. That’s the essence of blockchain. Many misunderstand this. They don’t realize blockchain is for state propagation, not storing historical data. That’s a crucial point.
Second, every blockchain has subtle differences—some use PoS, some PoW. You see various new rollups and DA solutions. Some use Call Data, others Blob storage. From a developer’s perspective, they just want to see token balances, NFTs, cost basis, standard info. They don’t care about technical details—and rightly so.
So unification makes sense. This is novel for Covalent—we built a unified interface across all blockchains we index. We index about 200 blockchains including testnets. So if you integrate Ethereum, just change one character, and you can plug into any other chain—Polygon, Arbitrum, Phantom, Optimism, Base, Mantle, etc. For any EVM chain, just one character change. Build your UI once, and it just works. This is hugely popular with developers because they don’t want to rebuild the entire stack for every different chain.
There’s also understanding the data stack itself. You have data products like CoinGecko—more retail-focused, providing high-level stats, market cap, circulating supply, etc. This isn’t truly on-chain data—some of it is off-chain—but it’s for retail users. Then you have infrastructure-layer indexers like Covalent and The Graph, which provide structured data. Then there’s RPC—lower level, providers like Alchemy and QuickNode giving raw data. That’s the stack. Our specialty is structured data because RPC gives unstructured, messy data. That’s a key distinction. The value of indexers is taking all that raw, unstructured data from RPCs or blockchains and presenting it in usable, consumable, readable structured formats. That’s how the stack is organized.
Then there’s The Graph. I think there are two different philosophies in building indexers. The Graph uses subgraphs—you create app-specific endpoints, but each DApp has its own schema and structure. Covalent has a unified API—a single, consistent schema. It’s not tied to any specific DApp or use case. Use cases, appeal, customer base are completely different, but both solve similar problems to some extent. Interestingly, after about five years, The Graph has become more like Covalent, and Covalent has become more like The Graph—as both try to expand their scope.
Creating the Flywheel Effect
Blair: That’s fascinating. I noticed you emphasized many plans in Covalent Vision 2024. A quarter has passed—EWM seems crucial, which I understand. But how does your team decide the roadmap? It involves so many components. Can you outline current progress? Among all these initiatives, which will be your focus?
Ganesh: I think there’s method in the madness. Though it looks chaotic, it’s actually like a puzzle—everything fits into a holistic project. Let’s step back and understand the flywheel. Covalent indexes blockchains. Developers and DApps on those chains use Covalent. When they use the product, these DApps want multi-chain capabilities.
So they pull more data from Covalent, unlocking more use cases. As more developers and use cases emerge, more blockchains want to join and leverage these apps and builders. We’ve already indexed 50, 60, 70 chains—all drawn in organically. We don’t do active outreach. They come to us because we have all these products and attention. Take Rainbow Wallet—it’s super popular. All its data comes from Covalent. Unless Covalent supports a chain, Rainbow won’t go there.
So we get requests from many DApps. For example, we announced indexing for Blast—not because the Blast team asked us, or we reached out to them, but because Rainbow requested Blast support. For Rainbow, it’s just one character change. With one tweak, suddenly Blast is supported—everything on Blast just works. They don’t need to rebuild anything. It’s that seamless. That’s the flywheel effect.
All of this is like a spinning wheel. The key is the token. Revenue from demand-side—developers—is usage-based API pricing: free at first, then pay-as-you-go. This revenue flows to node operators—actual CQT holders. The entire flywheel spins—decentralized economy starts growing. It may seem massive, but our community programs, fee buybacks and conversions, EWM, demand-side product lineup, developer grants, expanded indexing for rollups and RaaS—all are parts of this giant flywheel. It keeps accelerating. It’s all one plan—just looks like scattered pieces.
Product Development Progress
Blair: It seems messy, but as you said, everything’s connected. How’s progress going? What product developments can we expect in the future? Can you share any sneak peeks?
Ganesh: Sure. A key highlight we emphasized last year was the fee conversion mechanism—revenue from external sources, basically customer payments used to buy back CQT and distribute to operators. This mechanism launched about 45 days ago—buying $1,000 worth of CQT daily. Maybe in the show notes or elsewhere, I can share the wallet address. $1,000 every day. Sometimes CQT is 20 cents, sometimes 40 cents—it doesn’t matter. As demand-side revenue grows, this sets a price floor for CQT because it buys from anyone wanting to sell. That’s the floor. This mechanism is live—exciting update, final piece of the picture.
Another thing is staking migration back to Ethereum. So far, we’ve been using Moonbeam for settlement, etc. Overall, I think the Polkadot ecosystem hasn’t lived up to expectations. So we’re moving staking back to Ethereum. All audits are complete. Next is EWM testnet—then incentivized testnet. It’s almost ready. Then double down on AI and DA narratives—build more products, engage the community. Everything’s on track.
AI Model Use Cases
Blair: Hope everything goes smoothly—sounds like a huge workload. From your social media, I know Covalent is pushing hard into AI, offering rich historical and real-time Web3 datasets. How does that work? Can you give concrete examples of AI model use cases? Web3 and AI have interacted before, but now we’re seeing legitimate use cases. Can you list some?
Ganesh: Sure. The key to large language models (LLMs) is structured data—it’s the input for everything. It’s the data reservoir needed to train these models. Covalent specializes in providing all this structured data. You feed this structured data into LLMs, fine-tune existing base models—any model you want—then start inference. That’s the whole pipeline.
It’s very similar to taking structured data, running query nodes, then querying that structured data. It’s a database product. You’re just transitioning from big data to big models. It’s a natural extension for us. We were surprised when the market started using Covalent for these use cases—but it makes sense. Structured data means clean, formatted, normalized data. Who wouldn’t want that?
So we’re seeing many use cases. Recently we published a post on all current AI use cases being built. Maybe we can link it in the show notes.
Smart Wheels is one example—a platform for on-chain copy trading. You follow any type of wallet, they summarize multiple wallets, then use AI to determine if it’s a scam or trap. Smart Wheels is a great project doing interesting things. Another is Leica. Leica.AI uses AI for analytics.You can see many such examples.
Again, we’re not in the analytics layer or retail-facing. They use all this data for training, then do complex token analysis—for research or other purposes. Leica is a great product for that. Another cool thing I recently heard about is Entendre Finance. It offers anomaly detection and predictive analytics—very attractive for financial management. In the background, it analyzes your payroll, expenses, etc. It uses AI for fraud detection.
Another example is bitsCrunch. bitsCrunch recently went public. They did a tokenless launch. Investors include Animoca and Coinbase. They use Covalent data for fraud analysis and more.
So the foundational data behind all these projects comes from Covalent. These are just some use cases—like how we started Covalent before DeFi, NFTs, GameFi emerged. Markets evolve differently at the application layer. We’re in infrastructure—we empower all these use cases.
CQT’s Role in the Ecosystem
Blair: That’s impressive. Seeing you empower these innovations and people creating impact because of Covalent is great. You’ve mentioned CQT several times today—can you elaborate on the token’s specific role in the ecosystem? I think this might be another unique aspect compared to other products. Also, recently you launched a token buyback program converting off-chain revenue to on-chain income. Can you share more insights?
Ganesh: CQT stands for Covalent Query Token—a token for staking and governance, central to Covalent’s tokenomic ecosystem. Based on our market experience, developers and product users don’t want to pay with tokens. It creates friction. Tokens in their hands are like 2017 antiques—meaningless. So everything on the demand side—revenue from customers—is priced in USD. Fixed. No volatility, no budget uncertainty. Then that USD is used in an on-chain mechanism to buy CQT. When I say $1,000 per day, that $1,000 comes from customers. Then CQT bought at market price is distributed to decentralized operators doing the actual work. They receive CQT.
So here’s an analogy: I hire contractors in the Philippines and pay in their local currency because that’s what they spend. I could pay in USD, but they’d convert it anyway. Similarly, the entire Covalent economy runs on CQT. There are other utilities—like an upcoming liquid staking-like program. Participating in delegation staking is another utility for CQT. As a token holder, you can delegate your CQT to one of the operators. We have about 14 operators, and we’ll publish a post to recruit more. They run the actual infrastructure. You delegate your tokens to them. We have a full tokenomics program—incentives for long-term data availability, even AI use cases. You might recall the New York Times lawsuit against OpenAI for using all their articles to train models.
So if there’s bias or profit derived, royalties must flow back—meaning you need to track all mutations to the base model. Blockchain is a perfect use case for exactly this kind of thing. Back to the token—it’s a standard ERC20. Tradable on OKEx, Uniswap, Sushiswap, KuCoin, Gate. You can hold the token, participate in the economy. As a delegator, hold CQT and earn yield. Or if you’re technically capable, run infrastructure and become an operator. That’s roughly how the backend works. You can also provide LP in DeFi.
Strategic Plan for Revenue Growth
Blair: It’s a beautifully designed mechanism—especially for all stakeholders in this game, all incentivized. Given your ambitious revenue growth targets, can you outline your strategic plan? With steady growth in institutional users, do you foresee the Unified API significantly driving this growth? You now have two main pillars: GoldRush and Unified API. Which will be your killer feature?
Ganesh: We take a different approach to revenue generation. On the demand side, we have three products: Unified API, Increment, and GoldRush. We designed them as different recipes using the same ingredients—same structured data. Unified API, GoldRush (blockchain explorer), and Increment (a Dune-like dashboard)—all built on the same data. That’s our strategy: multiple use cases and personas from the same indexed data. For revenue, we’ve set ambitious goals. We’ve seen consecutive monthly growth for several months.
For unlocks, there are unique opportunities coming. First, RPC-related. Currently, all RPC providers don’t store historical archive data. So the industry is consolidating around Covalent because Covalent’s purpose—and long-term goal like EWM—is to preserve the complete historical record of blockchains. We have highly customized architecture for this. We signed an agreement with Infura—they’re now routing traffic to us. We see other similar opportunities—I can’t name them yet—but they’re starting to migrate their backends to Covalent as part of their stack. So we should see significant revenue growth from this. Beyond that, we have missing pieces. We’ve openly acknowledged gaps in our Covalent Vision. We’re transparent about shortcomings in our data stack. One major gap is data provenance. For the toughest forensic and accounting cases, tracking data lineage is a missing piece. We’re making progress to fill this gap.
The entire team is structured so that whatever they do drives future revenue growth. This is a completely different part of Covalent—driven and incentivized differently. So I’m very confident we’ll hit all our goals. It’s just about product delivery, product pipeline, marketing, sales—different functions. I think few companies in the industry have this systematic capability to build beyond the token realm.
Views on Centralized Data Indexing
Blair: Thank you for sharing all these insights and behind-the-scenes stories. Sounds like your mechanisms are complex and well-designed across the board. Excited to see more innovation from Covalent. Let’s zoom out to the bigger picture of data indexing. How do you assess the current on-chain data market, especially regarding decentralized data indexing? Since there are also centralized data indexers in the market.
Ganesh: Frankly, I think centralized data indexers will eventually disappear. In the last cycle, we saw dozens enter the market—most are gone now. We see many entering again—I don’t know their fate. I believe centralized indexers don’t align with the spirit of decentralized tech—especially if you want to feed indexed data back into smart contracts. The trust assumption for data must match the level of trust when data first enters the blockchain. Break that, and the market’s potential size is capped. That’s it.
For example, look at Celestia, Eigen DA, or Avail—the trust in Celestia must be identical to the security of its underlying L1 network. Otherwise, people could attack Celestia and commit fraud on the L1. The setup must be the same. I think centralized indexers may attract some customers, maybe earn a few million dollars—suitable for simple use cases. But for the hardest use cases—that’s what crypto is for—you need trust and security. We’ve never seen centralized indexers as competitors because we’ve been in this space long enough to see them come and go, making a lot of noise.
Last year, Paradigm invested in a project called NXYZ. They raised $40 million, but shut down a year later. This happens often. We see centralized indexers simply don’t work in this domain. People criticize decentralized indexers for lacking decentralization. But if you look under the hood, you’ll find some centralization—even in Covalent. We’ve always been transparent about our gradual decentralization efforts. If you consider who’s truly introducing something new with cryptographic security, Covalent is the only indexer where every data change submits cryptographic proofs—auditable by anyone. This has scaled for years. The first network version launched summer 2022—April, exactly two years ago. Despite the Nomad bridge hack and other events, the network never stopped.
So I think many projects die from lack of focus, not because their core function fails. We don’t obsess over what others are doing. We focus on industry needs, customer needs, and solving the hardest problems—which drive the industry forward. It’s all about cryptographic security. Two years ago when we built this, no one demanded it. Two years ago we launched it. We’ve been working on this for four to five years. But this is what the industry needs—that’s how you push the frontier. We’re pioneers in this journey, and it’s important for people to know that.
Trends to Watch
Blair: I deeply admire your mindset. I agree—centralized players sometimes make a lot of noise, but it backfires eventually. I know you don’t want me to ask about investment timing, but are there notable trends you’re watching you’d like to share? Regarding your business metrics—data indexing volume from Layer1 or other indicators—there’s much speculation about this cycle.
Ganesh: I’d say Covalent’s revenue is real. Real customers paying to use the protocol and data. That proves data quality and service quality. No other indexer has revenue at this scale. We have clients like Fidelity, EY. That tells you how trustworthy the data is. Second, in terms of impact, a few weeks ago we calculated that over 250 million wallets use or benefit from Covalent data. This includes all wallets, custodians. If you look at Jump’s custody products, they all love using Covalent. Projects like Ambient Finance on Scroll, AirSwap, SushiSwap—all use Covalent data to enrich structured data. This industry has over 250 million wallets. It’s real. That’s the number of unique wallets using Covalent data. Proof submitted on-chain—anyone can download and reconstruct the entire Ethereum state from genesis. It’s real—you don’t even need to talk to us. Just download the proofs and rebuild the entire stack. I’d say these are real things on Covalent.
In terms of trends, at ETH Denver, I joined several panels. I’m convinced this cycle is the DA cycle—Data Availability. I’d also say there’s strong momentum in AI—a macro trend, very exciting. LSD, LRT seem to be hot topics. I’m not a finance person, so I don’t understand the complexities and risks of liquid restaking tokens, but they’re clearly getting a lot of attention. Maybe with EigenLayer and restaking, this year will bring big shifts. But we stay focused on where we excel—data, data availability, and AI.
Blair: Thank you so much for sharing your expertise today. The Web3 industry is still in its infancy. Changes and adjustments happen constantly. We’ve seen a flood of fresh capital and innovation entering this space, bringing all kinds of experiments. Let’s wait and see how things unfold.
Two Final Thoughts
Ganesh: I’d like to leave listeners with two thoughts. First, Covalent has about 60,000 developers. Infura probably has around 500,000. So Covalent has about 10% of Infura’s developer count. GitHub has 30 million developers—Covalent is 0.1% of that. We’re still so early—everything has just begun. Second, in bull markets, people wonder: should I invest in meme coins? Should I invest in LRT, LST? I’d say the biggest investment you can make is investing in yourself—your knowledge, research, and beliefs. If you believe in yourself, you should back yourself even harder. In my limited experience, those who did this created great outcomes for themselves. So I want to pass this message to our community and listeners.
Blair: So true. Thank you so much for sharing—all very insightful.
Ganesh: Thank you for your excellent work. I think we need more sincere builders and people with strong convictions. If listeners haven’t read it yet, please check out the English and Chinese versions of that research report—comprehensive, detailed, deeply explored. Thank you so much for everything you do for this industry.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
Mint Ventures
