Bryce Ferguson

The through-line

Ferguson's consistent argument is that crypto infrastructure keeps failing because it is built one level too high. Every generation of key management, he says, was fused to whatever use case happened to matter at the time, and then broke when the market moved: "What we saw is infrastructure was really tightly bundled with a narrow use case. We realized with turnkey we needed to go a layer below that. We needed to build something that had very flexible API based primitives that were super scalable and could evolve with the world of crypto" [1]. The conviction came from having been on both sides of the same wall, building custody infrastructure at Coinbase and then consuming it at Trade Republic, where he "got to see what the rough edges were, what what was working and what wasn't working" [1], and where his co-founder Jack had the mirror-image experience wrangling custody providers at Polychain [1][4].

The position has not changed so much as widened. In the earlier material Turnkey is framed narrowly as "a simple API to create private Keys sign transactions and then manage permissions around those transactions," a deliberately "pretty narrow uh surface area" [2]. Later, the same primitives are described as a platform that can serve "everything from payments to AI agents to trading bots to NFT platforms" [1], with the ambition that Turnkey becomes the thing "to make it as easy as possible to go from zero to one with whatever ideas you have for new crypto applications" [4]. What stayed fixed is the refusal to package: private keys treated as infrastructure, never as a bank account [2].

On why wallet infrastructure keeps getting rebuilt

Ferguson tells the history of crypto custody as a sequence of overfits. Early exchanges had no choice but to build key management in house, because holding hundreds of millions of dollars badly was "a company ending problem," so key management came bundled with trading [1]. When institutions arrived asking whether the exchanges could hold their assets too, that produced Coinbase custody and its peers, which were excellent for someone sitting on a hundred million dollars of crypto for a year and useless for anyone moving assets daily [1]. That gap created the short-term custody wave, Fireblocks and Copper and others serving market makers moving assets between centralized exchanges [1], which in turn could not serve consumer wallet-as-a-service, which produced a layer of high-level abstractions [1]. Now the demand is machine-driven: "a lot of folks are doing really high throughput machine-based transactions. They might be doing payments or AI transactions which are always going to be much more instantaneous and much more high throughput" [1]. His conclusion from the pattern is that "every single time people were too brittle with how they approached infrastructure because they were overfitting on a specific use case" [1], and that the antidote is to sit below the abstraction so as not to be "painted into a corner with a specific use case" [1].

The older custodians, he argues, carry their origins in their design. They were built around 2017 and 2018 when crypto was buy and hold, serving enterprises that "expected a web app first experience" and expected to phone a human before moving money, so "you didn't really have apis as a first class citizen" [2]. Deep per-action integrations follow from that web-app worldview: Compound governance inside the Coinbase custody app is a rich set of "pretty buttons and flows," whereas his view is that "fully flexible signing is the most important primitive you should be able to sign any transaction that you want and that's the only thing that's going to be able to keep up with the evolution of crypto" [2]. He is explicit that Turnkey is "a tech infrastructure provider we're not a regulated custodian or a regulated money transmitter" [2], and that developer self-service is a differentiator against competitors whose website button is "contact sales" leading to "months-long sales processes bespoke contracts and very lengthy integration timelines" [2].

On flexibility as the actual product

He concedes the word is close to meaningless on its own: "flexibility is a little bit of a throwaway word at times so might be helpful to also explain kind of what that means," which for Turnkey means low-level control over supported assets, supported transaction types, the mapping of users to private keys, and the policies over them, so that people are "never be blocked by TurnKey" [4]. The market test he cites is customers saying "I wouldn't be able to build what I wanted to build on any product other than turnkey" [1]. He positions Turnkey against rivals building "higher level more packaged closed source experiences" [1], and Jack's framing of the long-run bet is that packaged developer tools are "kind of capped in terms of the bread they can have," while primitives compose into abstractions nobody planned for [4].

That flexibility also let the market redirect them. They had expected investment funds and active traders to be the core buyers, and instead "we really got pulled in a different direction with folks building at the application layer folks who were integrating wallets within their experiences," which he describes as the market "yank Us by the hair in this other direction" while the low-level primitives still held [4]. Ferguson sorts the resulting demand into two buckets: embedded wallets, where users sign into a non-custodial wallet with email, OAuth or SMS, and transaction automation, where a business uses Turnkey as "API powered private keys" [1]. Moonshot, one of the largest crypto trading apps and briefly the number one finance app, runs entirely on Turnkey wallets, letting users sign in with email and buy a memecoin with Apple Pay [1]. Axiom offers advanced order types such as TWAPs and limit orders through the policy engine, which he notes is usually hard to do with non-custodial wallets [1]. Bridge uses Turnkey for backend payments flows, automating unique-address wallet generation and transfers [1], and Polymarket uses it for smart contract deployment and management, including the market on the next pope [1]. Elsewhere he points to Alchemy building a wallet as a service on top of Turnkey, Infinex from the team behind Synthetix, Mural for B2B cross-border payments and invoicing [3], and Goldfinch's Heron Finance [4].

On security as the precondition for everything else

Security is where he claims the deepest moat and the hardest engineering. Turnkey is, in his words, "creating the biggest honeypot in the world where security is everything," which forced innovation: "we've built a way to deploy verifiable applications into trusted execution environments" [1]. The credibility argument rests on prior scale, Jack having built the Coinbase cold storage system that "stores over 10% of all crypto in existence" [1]. Jack's account of the earliest days makes the tradeoff explicit: this "is not the kind of product that you can launch and just kind of you know follow the classic Silicon Valley phrasing of like move fast and break things," so the team spent roughly ten months building without real-time market validation, on conviction from their own prior jobs [4]. Ferguson adds that starting at private keys is "a very very sticky and strategic place to start" precisely because "a lot of the dev tooling companies in the space just don't want to touch it takes a lot of expertise it takes a a lot of kind of tinfoil hat paranoia" [4].

On the policy engine and who is allowed to do what

The policy engine is the piece he returns to most often, and its design inspiration is not crypto at all: "We took a lot of inspiration actually from how Cloud platforms manage access and security so AWS has something called identity and access management IAM" [2]. He argues existing permission systems are too coarse, mostly withdrawal thresholds triggering approvals, whereas his view is that "security has to be deeply deeply customizable" down to specific contract calls [2]. The canonical example is one Jack wanted at Polychain: users with the trader tag can trade up to a set dollar amount using the swap function on Uniswap V3 and can do nothing else with the key [2]. He distinguishes this from Gnosis Safe multisig, which uses private keys within the signing scheme, while Turnkey's engine is "a layer over private Keys" that can ingest both on-chain and off-chain data such as organizational structure, and can itself supply a key that participates in a multisig [2]. He accepts the parallel to banking compliance controls, where only certain users may make certain transactions, as "exactly the right way of thinking" [2].

The agentic case sharpens the argument. "If I have all of my money managed by an AI agent, that's pretty scary. They could go off, they could hallucinate about something and send it all to North Korea and that would not be great for me" [1]. The answer is the same primitive: limiting "the permissions of a specific API key and how that can access a wallet" [1].

On performance, and why machine speed is now the bar

He treats latency and throughput as a first-class competitive dimension rather than a footnote, because "blockchains are really fast. They're really cheap. People need to do thousands and thousands of transactions at very consumer level speeds" [1]. The numbers he cites: Turnkey signing latency of 50 to 100 milliseconds against MPC and older-style providers at five to ten seconds, "100 times slower," and throughput of 2,000 transactions per second where other providers "will fall over at like 20 transactions per second" [1]. Security, flexibility and performance together are what he says customers actually buy [1]. Reliability is framed as an obligation of being an infrastructure company: "we need to focus on scalability and reliability and and invest early in that because crypto grows fast and we don't want to be caught with our pants down" [3].

On usability, and being pragmatic about decentralization

One of Turnkey's founding theses, he says, is that the industry overshot: "people were kind of shooting too far towards decentralization and to actually build very useful products a certain level of centralization around Key Management really unlocks ux for now longer term definitely want to decentralize that" [4]. He thinks the teams doing this well hide the machinery, and offers Heron Finance as the model, an easy-to-use investing platform where crypto appears as "just a single line in their FAQ" [4]. The seed phrase is his emblem of self-inflicted friction, alongside a punk 6529 thread on securing NFTs that ran to roughly 106 tweets: "nobody has time to put in 106 steps to actually secure you know the the couple hundred bucks of you know assets that they might have" [2]. His read of the market is that the tech-forward crypto consumer is largely tapped out, so growth now depends on abstracted onboarding for people with zero crypto knowledge, with Reddit's NFT project as the proof [2].

The same logic drives his view of multi-chain sprawl. Asking users to know which L2 their assets sit on is, in his analogy, like asking them to know "if their app is on AWS or Azure or gcp," and "that just doesn't make sense at scale so we need to abstract all of that away" [3]. Account abstraction is where he sees that unlock arriving, with integrations to zeroDev, Alchemy and others, and by his count "90% plus of TurnKey clients are using account abstraction in some way," gaining gas abstraction, batched transactions and session keys [3]. The consumer-facing questions he says the team thinks about "all day long" are which authentication methods can reach a non-custodial wallet and how to keep those wallets secure while staying usable, high performance and scalable [3]. Roadmap items he named at the time were OAuth authentication and further methods, deeper support for more networks, and an expanded policy engine [3].

On stablecoins, payments and the crappiness of traditional rails

He gives two drivers for stablecoin demand. The first is monetary: "hyperinflation and kind of weaker Fiat currencies are leading people to one a half digital dollars so they can protect the value of their money," because stablecoins are "just the easiest way to hold USD" [3]. The second is infrastructural, and comes from his own culture shock moving from Coinbase to Trade Republic and discovering that when a securities trade settlement went wrong "it could take upwards of seven plus days to settle a single trade" [3]. Stablecoins fix that with instant cross-border settlement on shared rails [3]. Beyond remittances he points to the democratization of ordinary financial services, debit cards powered by stablecoins behind the scenes and access to yield-bearing savings-type products, noting that Americans often cannot see the point until they look at how financial services work elsewhere [3].

He credits two structural shifts for making payments viable. The industry experimented with algorithmic and asset-backed designs and has "stabilized more or less on a Fiat backed stable coin like a usdc or a USD," which built trust [3]; and L2s took transaction fees toward zero, where previously "it didn't really make sense to be doing High throughput stable coin transactions when your transaction fee could be five 10 $20" [3]. Asked what excites him outside Turnkey, he named payments as the trend most likely to define the next cycle, "growing incredibly incredibly fast" with the strongest product market fit [4].

On institutions, regulation and where crypto is in its arc

He is deliberately unromantic about institutional adoption. The recurring narrative of a dam breaking is wrong: "institutions are slow and they take small steps at a Time," so he expects steady growth rather than an overnight flood [3]. The endpoint, though, is total absorption, by analogy to mobile: "companies are calling themselves crypto companies right now and eventually we're going to have this world where every financial company is a crypto company under the hood and it's just the infrastructure that powers things behind the scenes" [3]. He observes financial institutions moving from passive holding into active on-chain participation, staking, governance, running nodes, and increasingly DeFi strategies across DEXes and liquidity pools [2], while product builders move away from omnibus designs, where funds sit together and trades clear on an internal ledger, toward wanting every action to produce an on-chain state change, a shift he attributes to cheaper L2s and new L1s [2].

On regulation his instinct is that it helps: clarity is arriving unevenly and not always as the industry would like, but "you compare this to 5 years ago and we're worlds ahead," with comprehensive frameworks like MiCA in the EU, and over time it "is going to be a boon because it's going to allow financial institutions and Fortune 500 companies to actually enter into the space" [3]. Trade Republic supplied the demand-side proof: Europeans wanted "to own crypto with a regulated bank," viewing an exchange as riskier, and German banking regulation "known for being best of the best" gave them comfort, a preference sharpened by FTX, which "felt so solid" and then crumbled [1]. Asked whether the industry is still early, he distinguished 2018 from now: "this is the do die moment for crypto," still early in corners, but no longer a niche topic [2].

On product as art, and building a career around zero to one

He answers the art-versus-science question about product building without hedging: "I definitely think it is more art than science," noting that at some companies execution is everything because you are chasing a known product, "but especially in crypto, it's completely art. There's so much open space" [1]. His own path was assembled from deliberately varied parts, an arts high school then engineering at Georgia Tech [1][2], then consulting at Bain, which he describes as broad exposure but "just so slow," full of companies not innovating, "tens of thousands of people that were just keeping the core product alive" [1], and where you are "building these beautiful PowerPoints but kind of just throwing them over the wall and hoping for the best" [2]. Coinbase in 2018 was the opposite: 180 people, an opaque industry with few influencers explaining it, and therefore a place to learn how the foundational infrastructure was actually being laid [1][2]. He moved through the BizOps team into staking and then custody product work, launching Coinbase's first staking products and the first governance products for Compound and Uniswap [2].

The pattern in his moves is a preference for the zero-to-one phase. Once Coinbase custody was the dominant player, "rather than going zero to one, it was going like 10 to 100 or 10 to a thousand. And that scaling point was going to be a little bit more predictable for me," so he went looking for another zero to one [1]. That is what took him to Berlin mid-pandemic to build Trade Republic's crypto team and product from scratch, including a quarantine in a tiny apartment, an international move he does not recommend making during a pandemic [1][2]. The culture shift was real: at Coinbase there were "foundational truths that people believed," while at Trade Republic much of the job was explaining why crypto matters, though the company was genuinely trying to adopt Silicon Valley style company building [1]. Aesthetically he still admires Trade Republic for boiling the product down to "core UI flows," and Rainbow Wallet for having "transformed what it means to have a wallet and interact with crypto on chain" [1].

On choosing investors

He wants both kinds on the cap table and uses them differently. Crypto-native funds "can lean hard into hype cycles" and have been most useful on go-to-market, connecting Turnkey with early stage crypto companies as they get off the ground [1]. Less crypto-native funds "can really be thinking about the the core business underneath these and thinking about how this grows longer term," and help identify traditional firms starting to move into crypto that Turnkey should be pushing on [1]. The financing history he references runs from a $15 million Series A co-led by Galaxy and faction [3] to a $30 million Series B from BCV that he says came earlier in the year than the team expected to raise [1].

Takeaways

• Build below the use case: every prior generation of crypto key management broke because it was bundled to trading, then custody, then wallet-as-a-service, so Turnkey's bet is flexible API primitives that avoid being "painted into a corner with a specific use case" [1].
• Fully flexible signing is the core primitive, since only the ability to sign any transaction "is going to be able to keep up with the evolution of crypto," in contrast to deeply integrated per-action web-app flows [2].
• Permissioning should be borrowed from cloud IAM rather than from withdrawal limits: policies fine-grained enough to allow a tagged user to swap on Uniswap V3 up to a limit and do nothing else with that key [2].
• AI agents make permissioning existential, because an agent "could hallucinate about something and send it all to North Korea," so limiting what a given API key can do with a wallet is the control [1].
• Performance is a buying criterion: 50 to 100 millisecond signing latency and 2,000 transactions per second against providers at five to ten seconds and around 20 tps [1].
• Usability beats purity for now, because over-indexing on decentralization blocks good UX, and the best crypto products make crypto "a footnote in the FAQ" [4].
• Stablecoin demand has two roots, flight from weaker fiat currencies and the fact that traditional settlement can take seven or more days when something goes wrong [3].
• Expect institutional adoption to be gradual, not a dam breaking, ending in a world where "every financial company is a crypto company under the hood" [3].
• Use crypto-native investors for go-to-market into early stage crypto teams and non-crypto-native investors to think longer term and open doors to traditional entrants [1].