Where to earn crypto: picking a venue, not a rate

Deciding where to earn crypto is a different exercise from deciding how. The method sets your return; the venue sets what can go wrong. Two providers can offer the same 5% on the same stablecoin and give you completely different claims: one an unsecured debt against a privately held company in a jurisdiction you cannot sue in, the other a position in a smart contract with no company at all. Neither fact appears in the advertised rate.

This page is a taxonomy and a checklist. It does not review anyone — the fifteen provider reviews on our platforms page do that, and the standards we apply are set out in our review methodology. What follows is the structure you can apply to a venue we have never covered, which is the more durable skill.

The five venue types

Large exchanges

Exchanges bundle earn products alongside trading, which is why they dominate the category: the assets are already there and opting in is one tap. Binance advertises earn products across more than 300 cryptocurrencies. Coinbase runs staking across roughly seventeen networks. The breadth is real and so is the liquidity.

What they do well is scale and, at the better end, product separation. Coinbase distinguishes genuine staking pass-through from USDC rewards, which it funds as a promotional rate rather than presenting as yield. OKX is unusually direct that Simple Earn Flexible is lending. The characteristic failure is concentration: your trading balance, your earn balance and your custody all sit with one company whose solvency you cannot verify. Bybit lost roughly $1.4 billion in a cold-wallet exploit in February 2025, and OKX's Aux Cayes entity pleaded guilty to operating an unlicensed money transmitting business in the same month, agreeing to pay about $505 million. Neither event touched customer earn balances, but both illustrate the size of the entity risk you accept.

Dedicated savings and lending platforms

These firms do one thing, which tends to make them either much more transparent or much less. Ledn is at the transparent end: it discontinued bitcoin and ether Growth Accounts on 1 July 2025 rather than keep rehypothecating client assets, now pays only on stablecoins at 6.5% and 8.5% by balance, and names the loan book that funds it, as our Ledn review sets out. Nexo is at the other end, advertising up to 9.5% flexible on USDT while its consumer pages do not say how the yield is produced, and publishing three mutually inconsistent headline maxima — 15%, 16% and 13% — across its own site.

The characteristic failure here is a narrow business absorbing a shock it has no other revenue to cushion. This is the category that produced Celsius, BlockFi, Voyager and Genesis, and the full anatomy is on what happened to the CeFi lenders. Nexo itself settled with the SEC and state regulators for $45 million in January 2023 over its Earn Interest Product and left the US before returning in April 2025.

Broker and neobank apps

App-first providers reach people who would never open an exchange account. Revolut states it takes no commission on staking and passes on all the rewards it receives on-chain, which if accurate is the cheapest arrangement in our survey. SwissBorg names the individual strategies behind each yield — Lido, Morpho, Maple, Kyros — and labels each with a risk rating, which is genuinely better practice than most of the sector.

The characteristic failure is disclosure density. Revolut publishes its per-asset rates only in the app, so no rate table can be checked from outside. SwissBorg's tier system produces a spread of roughly two to ten times between the lowest and highest tier on the same strategy, and tiers depend on holding its own token. The form-factor problem is large enough that it has its own page: crypto earn apps.

Licensed banks

One venue in our research is a bank in the ordinary sense. Xapo Bank is licensed in Gibraltar under the Financial Services Act 2019, pays 3.35% variable on US dollar savings, and states that the yield comes from high-quality liquid assets including AAA-rated US Treasury bills and money market funds. Its US dollar deposits are covered by the Gibraltar Deposit Guarantee Scheme up to the US dollar equivalent of £120,000. Its crypto balances are explicitly not covered, and its BTC savings rate is 0.25%, which is an honest reflection of how little safe bitcoin yield exists — a structural point we examine on earning Bitcoin and in our Xapo Bank review.

The characteristic failure is subtler: a bank wrapper can make a lending product look like a deposit. Xapo's BTC Credit Fund targets up to 4% by lending pooled member bitcoin to financial institutions, with a minimum of $120,000 equivalent. That is the same category of risk that destroyed the 2022 lenders, capped at a modest rate and correctly described as a fund rather than a deposit — but it sits on the same page as a guaranteed dollar account.

Self-custodial DeFi

In a lending protocol there is no company to become insolvent, the rate is a published function of borrower utilisation, and you can verify the reserves yourself. Aave v3 on Ethereum was supplying USDC at 3.57% and USDT at 3.11% as of 16 September 2026. That transparency is the strongest argument for the venue type.

The characteristic failure is code and delegation. Curated vaults add a human decision-maker who sets allocations, caps and fees without a regulated duty to you: two vaults for the same asset under the same curator have charged 5% and 25% performance fees with 118 basis points of yield difference, and one has held 86% of its assets in a single market. Exploits remain routine — $972 million lost across 207 incidents in the first half of 2026 by Immunefi's count, $1.32 billion by CertiK's, with the two methodologies differing. There is nobody to complain to and no recovery process.

Reading a product page before you deposit

Most of what you need is on the provider's own site, spread across a marketing page, a help centre article and a set of terms that nobody links prominently. The marketing page carries the rate. The help centre carries the mechanics. The terms carry the entity name and the clause that lets the provider change everything. Reading them in that order takes about ten minutes and answers eight questions, in rough order of how much money each one can cost you.

Two habits make the pass faster. First, search the page for the word "discretion" and its relatives — voluntarily, may change, subject to change, at our sole discretion. Those words mark the boundary between what you are owed and what you are being offered. Second, search for a company name that is not the brand. If the only legal entity mentioned anywhere is the trade name on the logo, the terms have not been written for a reader who expects to enforce them.

Three of those eight questions are frequently unanswerable, and that is a finding rather than a dead end. Crypto.com does not publish its staking commission while stating that its quoted APRs exclude it. Bitpanda deducts a commission before distribution and does not publish the amount. Bybit's help centre and Easy Earn terms returned no readable content at all when we checked them, which means the legal nature of the product, the contracting entity and the insolvency disclosures cannot be confirmed from primary sources. Where a provider does not publish something material, the honest description is "not disclosed", and that belongs in your comparison as a data point rather than a blank.

Red flags worth walking away from

A headline rate with no base rate is the most common. Bitpanda is the honest version of this pattern because it publishes the split: 3% contractual plus a discretionary bonus to reach the advertised 7%. Most providers show only the top number. Where a platform will not tell you what the guaranteed component is, assume it is zero.

Native-token tier gating is the second. Nexo loyalty tiers depend on holding NEXO as a percentage of your portfolio; Crypto.com, Wirex and SwissBorg run comparable systems. The top Wirex tier requires locking 7.5 million WXT alongside an annual subscription, and the enhanced rate is itself paid in WXT. The structural problem is that a stress event at the platform hits the token, the tier, the rate and your recovery at the same moment.

Third, undisclosed counterparties. Wirex states that deposits are converted into DAI or another stablecoin and deployed in a DeFi protocol, without naming the protocol. That is three stacked risks — conversion, de-peg and smart contract — described in a single sentence with no way to verify any of them.

Fourth, insurance language that blurs two very different covers. Custody insurance pays out if a custodian is hacked. It does not pay out if the platform lends your assets to someone who does not repay. Celsius advertised a $750 million insurance policy that did not exist, and the FTC entered a $4.7 billion judgment against it in July 2023.

Fifth, geography that does not match the marketing. Uphold staking is unavailable in the UK, the EU, the EEA, Switzerland, Canada, Japan, Singapore and Australia, plus several US states — in practice it is a US product with exclusions. Coinbase cannot accept new staking principal in California, Maryland, New Jersey or Wisconsin. Binance failed to obtain MiCA authorisation, withdrew its Greek application on 24 June 2026 and told EU users it would restrict services ahead of 1 July 2026, without publicly enumerating which earn products stopped. What each regime actually requires of a yield product is covered on crypto yield regulation.

Matching the venue to the money

The framework that survives contact with reality is simple. Decide which failure you are least able to absorb, then pick the venue type that does not have it. If you cannot tolerate a company failing with your assets, the answer is self-custodial DeFi or a bank, not a lender offering two extra points. If you cannot tolerate a smart-contract exploit with no recourse, the answer is a regulated venue, and you pay for that in yield. If you cannot tolerate reading terms, the honest answer is that this category is not suited to you.

After that, size the position as though a total loss at any single venue were possible. This sounds severe until you notice that it is exactly what happened to the customers of four separate platforms in 2022, none of whom were warned in advance. Diversifying across venue types rather than across brands within one type is what actually reduces correlated risk — three exchanges all lending to the same class of borrower is one position, not three. The arithmetic behind that sizing rule is on passive income with crypto, and the underlying method choices are surveyed on how to earn crypto. If you are choosing purely on dollar yield, start instead with crypto savings accounts and check every figure against the Treasury bill.