I don't think I can possibly be more excited about the future of decentralized finance (DeFi).
Most people don't realize how close to disaster we were this fall in the U.S. when it comes to the free and open future of crypto's financial backbone. Let me tell you, the Crypto Twitter mob *did not* stop the DCCPA and its killer DeFi provisions. Instead, the public spectacle of the bill's leaked draft tightened the D.C. reviewer circle and could have put FTX in the driver's seat for the final revisions of future altering law for open crypto protocols.
We dodged a bullet with FTX's well-timed implosion.
I don't want to argue counterfactuals, but I mention this only to warn that DeFi's headwinds are still very real. We can rejoice that we live to fight and build another day. But we also have to keep our eyes on the prize, and there is no bigger political footrace right now than the one between DeFi builders, gridlocked legislators, and overreaching regulators.
If we want to take crypto to the next level, then the builders and the legislators will have to get simpatico inmediatamente, and hold the bureaucrats in check when it comes to DeFi oversight.
Salvation lies within. It's time for security and audit standards. Risk waivers. Comprehensive community disclosures. And apps that work beyond speculative teaser rates.
We need more apps to work for everyday users and to find use cases that make us proud. And we need to clearly explain what DeFi is and define what the policy solutions should be. Where should we invest in 2023 to strengthen crypto's financial services sector?
Revenge of the dApps
Are the L1 protocols still fat or are applications that make crypto interesting starting to bulk up and capture value?
Ethereum transaction fees are way down this year, and the competitive dynamics with dApps have shifted. Uniswap, Lido, and OpenSea (the three largest Ethereum-based apps) now generate more monthly fees, on a combined basis, than the entire Ethereum L1. More volume is shifting off Ethereum to its scaling solutions (Polygon and Optimistic Rollups, in particular), or to new customized appchains (dYdX on Cosmos) where the dApps have more control over their value capture mechanisms and consensus rules.
Aave and Uniswap launched their protocols on various emerging networks this year and came to quickly dominate their segments in terms of volumes and TVL. This showed that the Schelling point for most users isn't around chain-specific native versions of lending and decentralized exchange, but of the top applications by user. Just as Binance will likely steamroll the top domestic "Thai-first" exchange in Thailand upon its entrance to the country, Aave is likely to dominate in whichever digital country (L1) it enters.
Why then does DeFi still sit at all-time lows versus Ethereum? I'm bullish on DeFi dominance versus Ethereum in 2023, absent ham-fisted new regulations.
As I said in the intro, I don't want to gloss over regulatory risk.
Tornado Cash usage plummeted following the U.S. imposition of sanctions on the protocol and the arrest of one of its core developers. Aave decided to geo-fence its front end from U.S. residents shortly afterwards, effectively eliminating one of its largest user bases. Concerns about regulation over software and front ends are so acute that some investors are proposing compromises that call for the regulation of the front-end services. (You might not like it, but we'll be fighting for DeFi in court for many years.)
Still, the TAM for DeFi remains massive. The global financial services industry is valued at nearly $23trillion. So a $15 billion DeFi sector doesn't seem frothy to me. At all.
I wrote about a lot of the core plumbing in DeFi in the 2021 Theses (two reports ago!), covering things like automated market makers (AMMs), yield farms, vaults, flash loans, oracles, impermanent divergence loss, and more. It's pretty amazing that since then we're down to a single billion dollar DeFi protocol in Uniswap, despite the incredible progress that's been made in the two years since.
Here's a good refresher on DEXs and how they work, as well as a video on Uniswap V3 that covers why people were so excited about its release.I think it could be an unassailable protocol in the AMM category. V3 has proven to be a true 10x improvement when it comes to the working capital these DEXs require to run reliable, low slippage exchanges. The "concentrated liquidity" that professional market makers provide has arguably created more liquidity for major asset pairs than even the largest centralized exchanges. Paradigm is, of course, talking its book here, but I tend to agree that there's not "much alpha left in designing new AMM invariants."
That doesn't mean there aren't other areas in which DEX protocols can compete.
AMMs might compete around offering dynamic fees that adjust with volume or volatility, or improving the performance and reliability of their reference pricing oracles. Or we might yet see winning central limit order book models. 0x has become the backbone of several new NFT and gaming marketplaces, even as some skeptics speculate whether on-chain order books are dead thanks to the constant (expensive, high latency) stream of transactions they must update to match buyers and sellers around a central limit.
Otherwise, most AMMs will fail to stand out from the pack. PancakeSwap, dominant on BNB Smart Chain, settles transactions on fairly centralized infrastructure and relies on a close relationship to Binance. Various other AMM DEXs are confronting their own unique challenges. Curve has been hijacked by mercenary "board members" (remember veTokens?), who have brutalized its tokenomics, while Balancer has been dealing with hostage negotiations with an activist tokenholder. Sushi has had a nightmarish stretch in reorganizing its leadership, Osmosis was hit hard by the Terra collapse in Q2, and Serum was outright killed by FTX.
There's only one thing I know for sure in the DEX space:
Other DEXs are not going to dethrone Uniswap with tokenomic tricks or marginal price improvements. You don't compete on fees, you compete on value, and Uniswap has proven that Uniswap is Uniswap's biggest competitor when it comes to creating incremental value for its users. The only question I have for 2023 is how the community will design and implement the protocol's fee switch, in which UNI tokenholders will vote to (finally) direct a portion of protocol trading fees to the treasury.
I doubt most Uniswap liquidity providers will push back much at this point. Given Uniswap's dominance, the fee switch won't hurt them much, and I'm reminded of the Steve Jobs negotiation withHarperCollins before the iPad's launch. There's nothing wrong with charging sustainable and fair fees for a valuable service. In fact, there should be more of that in DeFi, and an 80/20 split between UNI LPs and governing tokenholders feels fair.
(We cover many of the top DEXs in depth for our quarterly reporting customers. Here's the State of Q3 reports for Uniswap, Balancer, 1inch, and Osmosis. Stay tuned in January for the full Q4 updates, and if you'd like your community covered in 2023, please get in touch.
It's shocking how fast we went from "most customers don't care about decentralization" to "the only customers who weren't rekt this year used DeFi lenders."
The irony is that the centralized lenders were the ones to blow up on exotic bets and toxic token collateral, while the biggest DeFi lenders appear to have been running much cleaner loan books. In MakerDAO's case at least, they've even been piling into <checks notes> U.S. Treasuries?
MakerDAO began dabbling in real-world assets (real estate, invoices, trade receivables, and commercial loans) in 2021, and they now count fully reserved dollar stablecoins as the majority of the collateral underpinning the Dai stablecoin. They added $500 million in exposure to U.S. Treasuries this fall as risk-free yields ticked higher and DeFi yields collapsed.
You would think these protocols would be upside down now that crypto risk premia are rising and real-world yield is a bit easier to come by. Instead, it's been the opposite.
It's a far cry from August, when MakerDAO's founder, Rune Christensen, claimed that he was considering proposing a rotation out of USDC and other real-world assets at risk of seizure in the wake of the Tornado Cash sanctions. That didn't happen, though. Real-world assets account for 57% of Maker's total protocol revenue, up from less than 10% in July.
As rates have risen, Aave decided to get in on the overcollateralized stablecoin game as well. Its GHOstablecoin, native to the Aave protocol, will allow protocol "facilitators" to mint a limited amount of GHO in a trustless manner and allow users to borrow GHO while earning yield on deposited collateral. The Aave protocol will capture 100% of interest revenue from GHO, compared to 10% on its other assets, so the success of GHO would be a boon for the project. The official deployment date for GHO and Aave V3 on Ethereum is expected soon™.
These decentralized lending protocols are a) more transparent, b) better collateralized, and c) seem to have a much firmer grasp on how to do risk management and rotate into whatever risk-free assets are best.
(We cover many of the top lenders in depth for our quarterly reporting customers. Here's the State of Q3 reports for MakerDAO, Aave, Compound, and Liquity. Stay tuned in January for the full Q4 updates, and if you'd like your community covered in 2023, please get in touch.)
Undercollateralized DeFi Lending
Overcollateralized DeFi lenders generate most of their yield from the demand for leverage from margin traders and market makers. Those yields have dried up as the risk appetite and opportunities for margin trading have collapsed this year. On the other hand, undercollateralized lenders like Goldfinch⭐ and Maple Finance offer yield on undercollateralized positions and have been able to generate double-digit yields from this riskier form of lending.
You might be asking, "Isn't this what just blew out a dozen large centralized lenders?" And you'd be correct. While it appears that DeFi's undercollateralized lenders maintained a higher standard of risk management than their centralized counterparts (we wrote over the summer about how 99% of Maple's loans were returned with interest despite the period of extremely high volatility for crypto in Q2), reality is starting to catch up with them.
Lending to delta-neutral crypto market makers worked well to avoid the fallout from Terra. Maple's delegates may have been more prudent with counterparties, even through the fall. But crypto delivers risk in unexpected ways. Contagion from FTX led Maple to liquidate loans worth $37 million, reducing its active loans to just $41 million. Of that, $10 million is now distressed.
Goldfinch still has a pristine track record, in part because it lends to non-crypto borrowers only (less reflexivity). But the protocol does have its own unique risks, such as higher exposure to developing economies and the fintech sector.
Maple Finance had a strong track record, as did Genesis Capital and BlockFi. But in a full risk-off environment with poor on-chain identity, reputation, and creditor infrastructure, this market has gone ice cold. Goldfinch is down 90%+ since its public launch at the beginning of the year. Maple is down 90%+ since its April high.
It's doubtful that crypto will ever compete with legacy banks at scale without undercollateralized lending.
The question is whether we have the right building blocks (credit scores, insurance, and yes, credit default swaps) that make this sector of DeFi viable in the short term. Can we use smart contracts, on-chain data, soulbound NFTs, and DeSoc identities to replace loan officers?
I hope so, but it feels like it may simply be too early.
Liquid staking protocols made it more palatable for thousands of Ethereum investors to stake ETH and help bootstrap the security of Ethereum's post-Merge Proof-of-Stake blockchain.
A game for whales (a 32 ETH minimum is required to stake), with significant duration and technical risk (you didn't know when your staked ETH would become available again in the leadup to The Merge, which had been fraught with technical risks and years-long delays), and high opportunity cost (you can't use your ETH in the NFT or DeFi markets), was opened to all thanks to protocols like Lido and Rocket Pool.
These staking protocols were able to catalyze thousands of incremental ETH stakers thanks to the creation of new synthetic assets, staked ETH (Lido's stETH and Rocket Pool's rETH) that accrued Ethereum's staking rewards (for a 10% fee) atomically, and were liquid and tradable as soon as they were minted.
Post-Merge, the economics of staking are even more compelling.
With ETH inflation falling to near zero and validators earning real revenue (tips and MEV) on each processed block, the "real returns" to stakers have spiked to 6%. Further, when withdrawals go live on the Ethereum staking contract as part of the Shanghai upgrade planned for sometime in 2023, the risk of staking ETH (and synthetic instruments) will fall considerably as the minimum duration for staking will fall to just 27 hours.
We expect DAOs, exchanges, and other institutional ETH holders to stake en masse post EIP-4895. DAOs have a combined $1 billion in treasury assets. Coinbase has even more under custody. There aren't many more automatic annuity streams in crypto. Lido and Rocket Pool strike me as the next major long-term blue chip DeFi assets. I expect Lido will be the top fee generating dApp in 2023 across all of crypto. And Rocket Pool's market share could 5–10x in the new year.
Word of warning! Until ETH holders can withdraw funds from the staking contracts, there is ever-present technical and duration risk to these products. You'll recall from earlier that 3AC was decimated by forced selling of stETH after its peg to ETH broke amidst the market turbulence in May. I think about those as dramatically reduced risks going forward, but it's likely that there will be things that break in the synthetic staked token market as more exotic structures get proposed to compete with Lido. (Though I'm excited about Eigenlayer's potential.)
Final thought: remember last year, when JPMorgan projected that staking could be a $40 billion/year industry by 2025? Right now, there's a mere $13 billion in staked ETH circulating at 6% per year, a figure which would have to grow 50x or more, to hit JPMorgan's $40 billion staking revenue threshold. A 50x on Lido's current revenue would be $1.6 billion annually. Nice.
dYdX has strong product-market fit despite botching its decentralization and token value accrual scheme out of the gate (though, I think silly regulation is partly responsible). Q3 marked the fourth consecutive quarter in which token rewards paid out of dYdX outpaced revenues earned, this time to the tune of $22.6 million.
Things are soon changing for the better. The launch of dYdX V4, a dedicated appchain on Cosmos, will provide a prime opportunity to decentralize the protocol and fix the leading on-chain derivative protocol's tokenomics. Validators in V4 will run the dYdX order book instead of the centralized dYdX Trading entity, realigning tokenholders and protocol revenue.
The community is aware of the coming tailwinds, too. In Q3, governance voted to wind-down two different token incentive programs and lowered trading rewards to reduce token inflation.
And the sheer number of innovations dYdX is looking to insource and customize as a vertically integrated appchain (base L1, custom modules, off-chain orderbook network, oracle network, Alchemy-like indexer, mobile applications, and a custom wallet) will be something other communities will be watching intently.
The coming migration will be risky, but it could also create the highest quality perpetuals product on the market and compete effectively with centralized alternatives. That's timely, given that one of the world's largest marketplaces for crypto futures just disappeared.
I was surprised by how poorly on-chain asset managers have done this year, though I suppose I shouldn't be.
The fact that I haven't thought about them since last year's report is telling, and with DeFi yields plummeting and investors focusing on capital preservation and minimizing counterparty risk, there aren't many value-adds to yield optimization services. We're back to where we were in 2019 and early 2020, when this sort of DeFi application felt like picking up pennies in front of a Zamboni of risk (smart contract, governance, and counterparty).
It feels like the puck has moved over to investment DAOs instead (covered in the next Chapter), though I'm scratching my head at how protocols like Enzyme and Index Coop have struggled so mightily in "creating customized indices for crypto." That feels like a trend whose time has come, and I wrote last year about how ETFs were one of the most successful financial innovations in the past 30 years, with some $6 trillion in net assets, lower management fees, and higher net returns for their investors.
Good active managers could make a killing this cycle as memes run out of steam and fundamentals take over. Rules-based asset managers are now much easier to create with code, and protocols that allow for a proliferation of on-chain funds and indices should have a stronger showing in 2023, provided they are built overseas. (The SEC *definitely* won't like these.)
I'm going to run back this prediction for one more year:
"There's opportunities for smart beta products, sector specific plays, portfolio copy-trades, and more. The biggest near-term opportunity could be shadow stonks, like we've already seen on Synthetix, Mirror, UMA, etc. Consider that the "total value secured" by Chainlink oracles (smart contracts that leverage their data infrastructure) [is now $16 billion, down 80% year over year, but still 2x higher than it was in 2020] and you have the foundation for something big. Reliable oracle data, synthetic stocks, and indices smart contracts…all we need are CNBC talking heads for distribution, and we're full-stack, fam."
New Novel Markets: Two Truths & A Lie
Can you spot the lie?
a. Anyone who follows me knows that my side hustle to crypto is ESG investing. Whenever you can invest in a provably sustainable, green, and socially conscious organization, you should. And I'm glad more people are starting to agree with me, even if it's only on a voluntary basis.
So you can imagine how excited I am about crypto protocols like Nori, Flowcarbon, KlimaDAO and Toucan, which are laying the groundwork for corporations and individuals to reduce their carbon foot- prints, by revamping the broken carbon trading markets. By tokenizing carbon offsets and establishing reputable on-chain carbon marketplaces, these protocols have the power to bring transparency, liquidity, and aggregation to the global green markets.
b. Real estate finance is one of the largest markets in the world. So it should be expected that there will be a massive opportunity for crypto protocols to innovate in this area. Physical real estate plays that leverage crypto haven't taken off yet, such as Propy (tokenized real estate), Milo Credit (collateralized mortgage lending), and Vesta Equity (fractionalized ownership), and I'm skeptical that there is a near-term path to crypto's meaningful participation in such a heavily regulated and physical asset class.
That said, there are pockets that could be interesting. Housing prices are up 33% since the start of the pandemic, and the average 30-year fixed-rate mortgage is at its highest level in over a decade (up 87% since last December). The majority of commercial institutions have ceased accepting new applicants for home equity loans since the 2008 recession, and I can tell you from first-hand experience that the lenders hate when the majority of your net worth is in crypto. I wonder if this is the sector where undercollateralized lending could actually work. Buy a crypto-friendly bank, deposit holdings, and secure loans against your house AND crypto. It's not necessarily a cost saver, but it could boost the accessibility of the mortgage market to more home buyers.
I've got to admit, I'm much more excited about crypto protocols that fractionalize NFTs and virtual property than I am about physical real estate plays. When in doubt, natively digital property will be easier to secure on natively digital ledgers. The physical markets will always come last.
c. I believe in prediction markets.
Ok, now guess the truths and the lie.
Fine, it's the prediction market one.
Prediction markets feel like a use case for crypto that's always one year away. I'm surprised they yielded essentially nothing of value in a high-stakes election year this year. Today, there's a paltry $2 millionat stake on bets surrounding the 2024 Republican Presidential nomination in the US on Polymarket. I think Avichal is largely right: "Prediction markets converge to the two big use cases: 1/ speculating on the price of assets & 2/ sports betting. #1 is DeFi. I do think someone will do a sports betting 'DEX' but it requires sports specific oracles." Legal sports betting is growing without the added risks of crypto, and no one wants to gamble on tokens right now. Nothing to see here.
As I said up top, I'm not sure that we should take for granted that DeFi will survive the global regulatory gauntlet next year. DeFi is seen as the highest-risk subsector of crypto by members of Congress, and it will be challenging to effectively communicate a different story. If the value of crypto writ large is a complicated story to tell, just add DAOs, and it ratchets up the complexity to 11 for people who do not live and breathe the details of our industry.
I predicted last year that "we'd see a bifurcation of DeFi into CeDeFi (known teams), and AnonFi (pseudonymous developers)." With the Tornado Cash sanctions and detention of one of their core developers, that is unfortunately holding up, all too well.
Some teams are already preparing for a world in which DeFi protocols are regulated at the interface and "Labs" levels. It's a hard compromise that might have to be temporarily swallowed in some jurisdictions (to protect the underlying protocols themselves from crackdown), then fought like hell in courts.
It isn't new for the front ends of major protocols run by centralized teams to implement certain controls and restrictions on their sites. Uniswap Labs delisted tokens last year, Aave geofenced U.S. users recently, and most browser wallets and front ends are doing *some level* of IP tracking for AML compliance purposes. Because some teams have already set precedent, the expectation might be that all communities have the ability to comply with laws deemed applicable to DeFi, and should do so without fuss or incident.
Last year, I wrote about the Sia Skynet team's efforts to secure unstoppable frontends in Web3 via a project called "Homescreen." I still hope something like that succeeds, but it feels like we're a couple of core building blocks of privacy tech (and maybe a couple of brave legal victims) away from seeing that become reality.
As bravely as I'd like to flout laws I disagree with, I'm not spending a single day in Bahamian prison, American prison, or serving time as a guest of any state.
There are smarter fights to pick and stronger allies to solidify first. There's no glory in defeat.
The majority of DeFi users and volume in DeFi may be KYC'd within the next several years, but we can take comfort from the Tornado Cash case study that other non-KYC'd transactions will always be processed. Despite the furor around network-level censorship following the OFAC sanctions, 30–40% of Ethereum's blocks are still processing transactions to and from addresses on the U.S.'s naughty list.
There's no reason to believe this 70/30 dynamic won't continue elsewhere in DeFi. Just like the broader economy, the whitelist users will drive the majority of volume while we figure out how to weed out the blacklist and protect the gray area.
There's no such thing as bullish unlocks. That was a bull market meme. I can't believe I said this last year: "There's more trust in VCs to be professional secondary sellers on the way up, than panic retail sellers on the way down, too, so FDV likely matters more in well-distributed tokens than ones with big, long-term oriented backers."
That is such a horrendously ice cold take. My bad.