Recently, major conferences in Hong Kong have been heating up, the primary market is recovering, and ABCDE reviewed over 100 projects in Q1—giving us firsthand exposure to several especially hot sectors, with Appchain, ZK, and Gaming topping the list.
This year, notable shifts have occurred across these three sectors, prompting us to revisit our thinking. We’ve documented our updated perspectives in this memo for your reference.
I. Appchain (especially RAAS — Rollup-as-a-Service)
RAAS emerged at the end of last year, closely tied to the launch of OP Stack. However, “Appchain-as-a-Service” has existed for much longer—exemplified by Cosmos SDK—and began gaining traction after Celestia introduced the concept of modular blockchains. RAAS can be viewed as a particularly hot recent subset within this broader category.
Why might the Appchain-as-a-Service sector currently be overhyped?
First, if you’re a developer aiming to build your own Appchain, here are your current options:
If your chain is EVM-compatible, you can:
1. Build a pure ETH sidechain like Ronin (though few would choose this path today)
2. Use Skale to build an ETH sidechain
3. Launch a chain on Avalanche and connect it to Avalanche’s P-Chain
4. Deploy an EVM chain using Polygon Supernet
5. Build a BNB Chain-based sidechain using BAS
6. Build a Rollup Appchain using OP Stack
7. Build a Rollup Appchain using Caldera (which is fundamentally built on OP Stack)
8. Build an L3 using zkSync (expected to go live this year)
9. Arbitrum recently launched Orbit—a similar L3 infrastructure framework to OP Stack—while this memo was being written
10. Additional projects under active development include Opside, Stackr, Sovereign SDK, and others
If your chain is non-EVM, you can:
1. Build a chain using Cosmos SDK—either independently or by sharing ATOM’s security (via the newly approved ICS proposal)
2. Launch a chain using Substrate—either by bidding for a Polkadot parachain slot, connecting to Octopus Network, or operating entirely independently
3. Build a Rollup Appchain using Celestia’s Rollkit—with DA provided by Celestia and optional settlement layer
4. Build a Rollup Appchain using Dymension
5. Build a Rollup Appchain using Saga
6. Build an L3 based on Starkware (expected to go live this year)
7. Many more projects are undoubtedly in development—some unknown to us or omitted here
Doesn’t this feel like—“slightly too many options?!”
Second, which applications are truly suited to becoming Appchains? Recall the online debate from a few months ago about whether Uniswap should become its own Appchain.
In summary, turning Uniswap into an Appchain would bring advantages such as transaction fee capture, token value accrual, MEV resistance, and dedicated resource allocation—but would also incur significant trade-offs: degraded user experience, reduced security, and diminished composability.
To date, we have seen no official plan—or ambition—from Uniswap to become an Appchain.
Compound initially considered building a chain using Substrate but later abandoned the idea. Today, both Compound and Aave V3 deploy across multiple chains—making an Appchain transition appear unlikely for now.
Curve, meanwhile, has never expressed any intention to pursue this path.
DYDX stands out as the sole major DeFi protocol that chose to build on Cosmos—an Appchain with minimal reliance on composability. Its migration is expected to complete in Q3 this year. Perhaps since Luna’s collapse, DYDX will be the most anticipated Appchain launch. It will demonstrate a viable path for builders: if you dominate a sector *and* don’t heavily depend on composability, launching a sovereign, high-performance Appchain makes sense. Prior to that, consider launching your dApp on a general-purpose chain ecosystem to grow organically—then “go solo” once maturity and scale are achieved.
This path is relatively feasible. For example, if a SocialFi dApp built on Lens Protocol grows significantly—achieving high daily active users—beyond what Polygon’s throughput can support, launching its own Rollup Appchain using one of the above infrastructures would make perfect sense. That said, in the short term, the number of projects meeting all these criteria—and thus qualified to launch their own chain—is likely still far smaller than the number of available infrastructure solutions.
Then there are sectors inherently suited to launching Appchains from day one—these will become the largest users of the aforementioned infrastructure. Until cross-chain heterogeneous composability is fully solved, non-Cosmos DeFi protocols are unlikely to follow this route. The most natural fit remains GameFi—including Onchain Autonomous Worlds. Examples include DFK and Crab on Avalanche, the recently popular OPcraft, and Curio (which launched its chain via Caldera).
Looking at my MetaMask and Keplr wallets, post-Luna collapse, I’ve barely used any Appchains besides DFK and Osmosis—prompting me to wonder: does this cycle truly require *dozens* of Appchains—or Rollup-as-a-Service offerings—at such scale?
Finally, regarding the recently trending RAAS (Rollup-as-a-Service), we can offer a light critique.
1.1. OP-Based Rollups
Currently, most OP-based RAAS offerings essentially involve forking the OP Stack—there’s virtually no technical barrier involved. The OP Stack’s codebase and documentation are exceptionally clean and clear; our ABCDE technical experts can set up an OP Rollup appchain in under a day by simply following the official docs. As such, the value proposition of these RAAS providers lies primarily in ancillary services—such as sequencer infrastructure, block explorers, and rapid deployment—not in core technical innovation. Marketing prowess far outweighs technical capability here.
Moreover, TPS, block time, and transaction fees on these appchains are functionally identical to Optimism’s—no performance or cost optimizations have been implemented. Thus, theoretically, unless Optimism itself is congested, users gain no meaningful UX advantage over Optimism’s shared L2. And all of Optimism’s current shortcomings—including its long-delayed, still-unlaunched fraud proof—are fully inherited by these RAAS offerings.
That said, credit must be given to OP Stack’s “Superchain” vision: the more Rollup chains launched using OP Stack (or its forks), the closer we get to realizing a Polkadot-like architecture—slotless, with built-in cross-chain communication protocols and fully asynchronous composability. Though this vision remains distant, it’s undeniably compelling. Leveraging the OP Stack narrative, Optimism has successfully matched Arbitrum’s red-hot DeFi innovation head-on. Midway through drafting this report, Arbitrum announced Orbit—I knew Arb wouldn’t let such a lucrative opportunity slip away. Sure enough, token launch + new projects rolled out seamlessly. OP vs. Arb? The rivalry just got even more exciting!
1.2. ZK-Based Rollups
Theoretically, ZK-based RAAS could improve appchain user experience. Projects like zkSync and Scroll—pursuing the ZK-EVM path—prioritize EVM compatibility above all else, meaning their circuit designs inevitably sacrifice some efficiency and cannot be optimized for specific dApps. If RAAS providers could tailor circuit designs or optimizations per dApp, ZK appchains would undoubtedly outperform generic ZK-EVM rollups in both performance and UX.
However, talent fluent in both ZK cryptography and blockchain engineering remains extremely scarce—and most of it is concentrated at Starknet, zkSync, Scroll, and Polygon. Among currently available ZK-based RAAS offerings, nearly all simply deploy forked chains using zkSync’s open-source Alpha release. Once Polygon and Scroll go fully live and open-source, RAAS providers will likely only offer customers a choice among ZK-EVM implementations—e.g., “Do you prefer zkSync, Polygon, or Scroll?”—reminiscent of selecting between Red Hat, CentOS, or Debian when spinning up a Linux VM on AWS.
So again, there’s little technical differentiation—BD (business development) dominates, and the ecosystem remains less mature than OP’s. Several major ZK rollups haven’t yet launched mainnets or fully open-sourced their code; what’s available today are mostly testnet releases riddled with bugs and subpar UX compared to OP’s relatively polished offering. We hope to see future ZK-based RAAS solutions that deliver truly customized circuit design and optimization per appchain.
II. Zero-Knowledge (ZK)
If Appchain-as-a-Service represents the flagship modular blockchain trend, then ZK and modularity are the two dominant themes shaping blockchain this year.
ZK-related challenges can be examined across several dimensions:
2.1. Layer 2
Scalability
No need to elaborate here—the major ZK-Rollups have all launched mainnets this year. Yet “launching” doesn’t equate to operational maturity; numerous issues persist post-launch.
Completion & Maturity
Currently, zkSync, Scroll, and Polygon have launched EVM-compatible mainnets; Starknet stands alone as non-EVM-compatible—but even Starknet hosts Kakarot, its “in-house” ZK-EVM implementation. According to ZK experts within the industry, the mainnet launches of these leading ZK-Rollups all carry a distinct “rushed-to-market” vibe. Their product completion and maturity levels fall short of traditional definitions of “mainnet readiness.” Post-launch, expect ongoing performance bottlenecks and bugs requiring continuous upgrades and patches. This is evident from repeated delays in testnet timelines—from last year into this one. The root cause? ZK-EVM is genuinely *hard*—so hard that even top-tier engineers require significantly more time than initially anticipated to solve it. Why the rush to launch this year? Likely due to mounting pressure from OP’s rapidly expanding ecosystem, steady mainnet upgrades, and growing stability. If ZK rollups didn’t ship now, they risked irrelevance. “Good enough to launch” sufficed—continuous iteration could follow later.
Performance
At present, ZK-based rollups’ performance is ≤ OP-based alternatives. From the end-user perspective, this difference may be imperceptible: both confirm transactions on the sequencer within seconds. ZK proof generation—which typically takes 10–20 minutes per block—can happen asynchronously. Users rarely care about, nor notice, when transactions achieve true “finality” on Ethereum L1. Current circuit optimizations and hardware acceleration efforts target precisely this 10–20 minute proof window—but they yield zero tangible UX improvement.
Costs
OP’s fraud proofs incur near-zero costs, whereas ZK proofs demand massive computational resources—i.e., significant expense. Yes, ZK rollups upload less data to L1 (lower calldata gas costs), but this savings is unlikely to offset the prover’s added expense—especially once EIP-4844’s blob transactions go live, slashing L1 data costs and amplifying OP’s cost advantage.
Security
This is subjective. Conventional wisdom holds that ZK’s mathematically sound proofs are inherently safer than OP’s game-theoretic fraud proofs—math > economics.
Long-term, this is unquestionably valid.
But not necessarily today.
“Security” ultimately means transaction finality on Ethereum L1. Arbitrum submits proofs every 2–3 minutes; Optimism does so roughly every 10 minutes. Due to ZK proof generation’s computational intensity, finality currently sits at ~10–20 minutes—even under full block utilization. With lower activity, delays worsen.
Thus, while mathematics indeed trumps economics, ZK-based finality times remain substantially longer than OP’s today. Closing this gap requires advances in ZK algorithms, circuit optimization, and hardware acceleration. Should proof times shrink to 10–20 seconds (perhaps in 5–10 years?), ZK rollups would decisively surpass OP in security.
2.2. Middleware & Other Applications
This sector garners even stronger enthusiasm than scaling-focused ZK applications—because building ZK-based rollups is an extremely “heavy” undertaking, as evidenced by multi-year delays before mainnet launches. In contrast, middleware solutions are comparatively lightweight while perfectly leveraging ZK’s unique strengths.
The hottest middleware segment is interoperability—using ZK proofs to eliminate reliance on third-party validators, dramatically enhancing bridge security, and even connecting previously isolated or incompatible ecosystems (e.g., between L2s, or between EVM and Cosmos IBC). Key teams actively building here include Succinct Labs (which recently launched Telepathy—a unidirectional ZK bridge to Ethereum), Electron Labs (early proponents of ZK-IBC), and Polyhedra (building ZK bridges, ZK-DID, and ABCDE’s first project).
Although this sector is significantly lighter than Rollups, it remains an extremely hardcore—and time- and labor-intensive—domain. It’s highly unlikely we’ll see a bidirectional ZK bridge this year that delivers both satisfactory performance and robust security. Fully ZK-native interoperability will likely require a timeline of 2–3 years for realization.
As for other ZK-related sectors, this year has seen something of an explosion—anyone who attended ETH Denver knows exactly what we mean. Teams are applying ZK to virtually everything: onchain vaults, DID, oracles, and even AI and machine learning. Some use cases genuinely make sense; others leave you wondering: “Couldn’t this be done without ZK?” Overall, the vibe feels reminiscent of the 2017 ICO boom—where blockchain was treated like a hammer, and teams scoured the world for nails. Back then, we saw decentralized ride-hailing services and decentralized Airbnb clones—projects that now seem absurd. This time around, ZK has replaced blockchain as the universal hammer, and every sector seems ripe for a ZK-powered overhaul…
III. GameFi
GameFi is arguably the hottest sector right now—no contest. Counting just the projects we’ve reviewed and discussed over the past few months, if broken down by sub-sector, GameFi leads in sheer quantity.
The reasons are straightforward. First, Vitalik Buterin himself stated years ago that finance and gaming would be the two earliest real-world applications of blockchain.
Second, despite numerous attempts across DeFi, storage, SocialFi, and other domains, mass adoption remains elusive. Although the X-to-Earn narrative has collapsed, Axie Infinity and StepN demonstrated tangible “cross-over” potential—so much so that many remain convinced large-scale mainstream adoption hinges on gaming.
Consequently, Web2 game developers—including technical teams from major studios and renowned game development houses—are increasingly turning their attention here. Simultaneously, native Web3 ecosystems—including NFT and DeFi projects—are actively exploring ways to “wrap” their offerings with a GameFi layer. Ape’s recent Dookey Dash launch exemplifies one relatively successful attempt.
Yet the current GameFi landscape sits in a somewhat awkward phase. The painful aftermath and death spiral of X-to-Earn have only just subsided. No one yet knows how to strike the right balance between incentives and gameplay—everyone is literally feeling their way forward in the dark. One emerging consensus is clear: Free-to-Play (F2P) is now the baseline. Models requiring upfront NFT purchases—like StepN’s sneakers or Axie’s creatures—are effectively obsolete.
We currently observe exploration across the following models:
AAA-tier games—
This represents a swing from one extreme to another: whereas Axie emphasized “Earn,” AAA-tier games emphasize “Play”—albeit to varying degrees. Lighter variants target the Web3 community, using NFTs and similar mechanisms to attract Web3-native players. Heavier variants target the broader Web2 audience, adopting conventional mobile-gaming production, operations, and grassroots marketing practices—essentially porting only the transaction system onto-chain, sometimes even embedding wallets seamlessly into the experience.
Casual & social games—
Web2 users remember the era of farm games, virtual pet care, and parking-space battles. Could Web3 GameFi enter a similar phase—blending social interaction, casual play, and light earning mechanics? The future remains uncertain—but at least this direction remains largely unexplored. A new X-to-Earn paradigm—currently the most visible variant is “Bet-to-Earn” (or “Risk-to-Earn”), exemplified by PSI. In essence, earnings become directly tied to player skill and proficiency. Imagine a Web3 battle-royale game where 100 players each pay $1 to join, and the winner takes home the $100 prize pool. This model fundamentally shifts economics toward PvP, avoiding the Ponzi-like dependency on ever-increasing new-user inflows—and its inevitable death spiral. However, like AAA games, this model demands high engagement and replayability to retain users, since direct economic incentives for average players shrink dramatically.
NFT-based Free-to-Own—
DigiDaigaku serves as the leading example here; Ape’s Dookey Dash also shares some similarities. The core idea is to attract players to acquire NFTs either for free or at minimal cost, then progressively enhance those NFTs’ utility over time. This path demands exceptional early-stage storytelling, mystery-building, and marketing prowess—and equally strong game-development capabilities later on. As such, it’s an exceptionally high-barrier route. Moreover, initial NFT supply is often capped at 10,000 units, posing a nontrivial challenge for scaling community size.
“Nintendo-style” games—
TreasureDAO and Gala represent this category. Gala’s titles tend toward heavier, more immersive experiences, while TreasureDAO focuses predominantly on lightweight mini-games. Following the breakout success of Beacon, TreasureDAO’s ecosystem is poised to attract more similar mini-games. While mini-games have enjoyed enduring popularity in Web2, whether Web3 can successfully replicate this remains unknown.
DeFi gamification—
DeFi Kingdoms stands nearly alone here—and its ambitions are enormous. Its system is extraordinarily complex, evoking visions of an “onchain Journey to the West.” Yet after peaking with 100x hype in 2021, it has made little material progress since; its token price has languished near its original level. Such deeply integrated, complex DeFi–GameFi hybrids face a steep and uncertain road ahead.
Fully Onchain Games—
This may be the most hyped GameFi category at ETH Denver. Why? Because all other categories bear at least some Web2.5 traits—only Fully Onchain Games embody truly Web3-native characteristics. Some aren’t even “games” per se, but rather Onchain Autonomous Worlds. This could represent the third truly “blockchain-native” innovation wave—after DeFi (MakerDAO) and NFTs (CryptoKitties). Just as DeFi and NFTs emerged in 2017 but didn’t explode until 2020 and 2021 respectively, Fully Onchain Games remain in an extremely early exploratory phase—and may similarly require 3–4 years before reaching their prime.
Among these seven directions, it’s still unclear which—if any—will ultimately prevail. Furthermore, GameFi faces a fundamental tension: although games offer the most natural on-ramp for attracting users from outside crypto, their core purpose is to create a world detached from reality—a sanctuary where people temporarily escape daily stress. Yet integrating Web3—or blockchain infrastructure—inevitably re-links GameFi to the real world via financialized mechanisms. Does this linkage undermine gaming’s essential function as a “sanctuary”? That’s a question worth deep reflection.
Summary:
Re-examining these three hottest sectors today doesn’t signal skepticism—they remain highly promising. In fact, ABCDE is bullish on all three long-term and has already invested in—or plans to invest in—each. We often overestimate the short-term impact of new technologies while underestimating their long-term value. These sectors may not deliver commensurate—or immediate—returns for investors or public markets. But for long-term VCs like us—with 5-year LP fund horizons—time remains our greatest ally.