Author | @archetypevc
Compiled by | Odaily Planet Daily (@OdailyChina)
Translator | Dingdang (@XiaMiPP)
Editor's Note: As we enter 2026, the focus of industry narratives is quietly shifting. Capital, infrastructure, user demands, and content distribution methods are undergoing structural adjustments. Archetype gathers insights from multiple industry researchers to outline several emerging trends that may evolve into key investment windows in the coming year.
This article is compiled by Odaily Planet Daily, aiming to present readers with the underlying driving clues of the next phase of crypto applications as seen by Archetype, as well as the structural turning points to watch in 2026.
When "chains born for applications" finally make sense: The best time for Appchains has arrived
Summary: Blockchains that are deliberately designed, finely crafted, and optimized based on underlying primitives for specific application scenarios will truly experience an explosion in the next year or two.
The influx of developers, users, institutions, and capital onto the chain is markedly different from previous cycles. They each have distinct cultures and preferences (i.e., their definitions of "user experience"), which often matter more than abstract concepts like "decentralization" and "censorship resistance." Sometimes, these demands align with existing infrastructure; other times, they require entirely different chain structures.
For cryptocurrency applications aimed at non-professional users, such as Blackbird and Farcaster, user experience is particularly crucial. Designs that would have been considered heretical three years ago, such as node deployment, single sequencers, and customized databases, are now the most reasonable solutions. The same applies to stablecoin chains or trading scenarios (like Hyperliquid and GTE) that are highly dependent on latency and price accuracy.
However, not all new applications are like this.
Today, a significant counterforce is growing: the preference for "privacy" from institutions and mainstream users. The experience demands of different applications vary greatly, and thus their underlying chain architectures should not be uniform.
The good news is that assembling a chain that meets application needs from scratch is now much simpler than it was two years ago. It resembles "building a custom PC."
You can choose each hard drive, fan, and power cable yourself—but in most cases, this is not necessary. You can easily select from a range of highly compatible, customizable preset configurations, just like choosing Digital Storm or Framework. If you need a certain degree of personalization, you can add your own components based on what they have already optimized. This ensures both stability and high flexibility.
Similarly, when applications can freely assemble and adjust primitives such as consensus mechanisms, execution layers, data storage, and liquidity structures, they can create chain forms with different cultural characteristics, allowing the application's own "experience definition" to receive native support, thus forming differentiated competitive advantages. This differentiation is akin to the stark differences yet shared underlying commonalities between ToughBook, ThinkPad, desktop towers, and MacBook.
More importantly, each component becomes a reversible "knob," eliminating concerns about causing widespread issues and no longer being constrained by the upgrade pace of the parent protocol.
With Circle's acquisition of Informal Systems' Malachite, the importance of independently controlling "customized block space" has become an industry trend. In the coming year, I expect to see more applications quickly building their own chains and block spaces based on "default templates" and underlying modules provided by Commonware, Delta, etc., much like using HashiCorp or Stripe Atlas.
Ultimately, this will allow applications to truly control their cash flow and establish a moat through chain structures that better align with user experiences.
Prediction markets will continue to innovate (but only some players will break through)
By Tommy Hang
In this cycle, prediction markets are undoubtedly one of the most outstanding application categories. The total trading volume of prediction markets across the chain has repeatedly hit new highs, with weekly trading volumes exceeding $2 billion, clearly indicating that this category has taken a significant step toward becoming a mainstream consumer product.
Amid the hype, numerous new projects attempting to challenge Polymarket and Kalshi have emerged. However, identifying "true innovation" amid the noise is key to determining which projects are worth focusing on in 2026.
From a market structure perspective, I am most interested in solutions that reduce spreads and enhance open interest. Although market creation remains permissioned and selective, the liquidity of prediction markets is still relatively thin for both market makers and traders. The best paths include optimizing routing systems, introducing different liquidity models, and improving collateral efficiency based on lending products.
Categorical trading volume is a key factor influencing platform success. For example, over 90% of Kalshi's trading volume in November came from sports markets, indicating that certain platforms are naturally better suited to compete for specific liquidity. In contrast, Polymarket's trading volume in crypto and political markets is 5-10 times higher than Kalshi's.
Of course, on-chain prediction markets still have a long way to go before achieving true "mass adoption." For instance, the Super Bowl alone generated $23 billion in off-market betting volume in 2025, more than ten times the total daily trading volume of all current on-chain markets.
Bridging this gap requires teams that can truly tackle the underlying challenges of prediction markets. I will continue to monitor these players in the coming year.
Agentic Curators: The next layer of DeFi expanded by "smart agents"
Today's DeFi "asset curation and risk allocation layer" is heading toward two extremes: either fully algorithmic (fixed interest rate curves, preset rebalancing rules) or entirely reliant on human input (risk control committees, active managers).
Agentic curators (smart agent curators) represent a third path: AI agents (LLM + tools + cyclical scheduling) manage the risks and strategies of vaults, lending markets, and structured products, not by executing fixed rules, but by "reasoning"—logical deductions about risks, returns, and position strategies.
Taking Morpho's market as an example: to design an attractive yield product, one needs to define collateral asset rules, LTV limits, and various risk parameters. Currently, this remains a human bottleneck, while smart agents can expand this process. Soon, you will see smart agent curators competing head-on with algorithmic models and human managers.
So, when will DeFi's "Move 37" arrive?
Many fund managers have extreme views on AI: either they believe LLMs will automate all trading desks, or they think they will collapse immediately in real markets. However, both overlook the real structural changes: AI agents combine emotionless execution, systematic strategies, consistent policy constraints, and reasoning capabilities, while human noise is high, and pure algorithms are too fragile. In the future, LLMs will act as "architects," designing risk frameworks, strategy constraints, and portfolio structures, while true high-frequency and sensitive computations will still be executed by deterministic code.
When the cost of deep reasoning drops to "a few cents," the strongest vaults will no longer be managed by the smartest people, but driven by the strongest computational power.
Short videos will become the new "traffic entry point"
By Katie Chiou
Short videos are becoming the default interface for "global users to discover and purchase content." TikTok Shop's GMV exceeded $20 billion in the first half of 2025, nearly doubling year-on-year, gradually acclimating global users to "watching and buying," where entertainment becomes the storefront.
Instagram has shifted Reels from a defensive product to a core revenue engine. This format has brought more exposure and is expected to occupy an increasingly large share of Meta's projected advertising revenue in 2025. Whatnot has already proven that real-time, personalized sales methods can achieve conversions at speeds unattainable by traditional e-commerce.
The logic is simple: real-time viewing allows users to make decisions faster. Every swipe is a potential purchase point. Therefore, platforms are quickly blurring the lines between "recommended information streams" and "payment processes," with the information stream itself becoming the new sales point, and every content creator acting as a distribution channel.
AI will accelerate this trend: lowering video production costs, increasing content quantity, and allowing creators and brands to test ideas in real-time. The more content and touchpoints there are, the more motivated the platform will be to optimize conversion efficiency every second.
And cryptocurrency perfectly aligns with this trend: faster content requires faster, cheaper, and programmable payment channels. As shopping becomes frictionless and directly embedded in the content itself, we need a system that can settle micro-payments, programmatically allocate and distribute revenue, and track the contributions of various parties in a complex impact chain. Cryptocurrency is born for such processes, making it hard to imagine how the era of massive-scale streaming-native commerce would develop without it.
Blockchain will drive new AI Scaling Laws
In recent years, the AI narrative has been dominated by giants and unicorns, while decentralized innovators have long been overlooked. However, outside the spotlight, several crypto-native teams have made astonishing progress in "decentralized training and reasoning," moving from whiteboard demonstrations to real testing and production environments.
Today, teams like Ritual, Pluralis, Exo, Odyn, Ambient, and Bagel are ready to "take the main stage" and welcome the arrival of a golden age. This new generation of competitors is expected to have an explosive orthogonal impact on the foundational development trajectory of artificial intelligence.
Distributed training environments are breaking existing scaling limitations, with asynchronous communication and parallel solutions proving feasible in real-scale training tasks.
New consensus mechanisms and privacy technologies are making "verifiable" and "confidential" reasoning a reality.
The new generation of chain architecture combines a "truly intelligent contract system" with a more general computing model, enabling AI agents to use crypto assets as a unified medium of exchange, forming a complete autonomous computing closed loop.
The foundational work has been completed.
The next challenge is to scale these underlying facilities into mass production and prove that blockchain can drive fundamental innovations in AI, rather than just remaining a conceptual slogan or a funding story.
RWA: Real World Assets Will Truly Land in the Real World
RWA (Real World Assets) has been discussed for years, and it is finally ushering in large-scale adoption—widespread use of stablecoins, well-established on-and-off ramps, and clearer regulatory frameworks have made this inevitable. According to RWA.xyz data, over $18 billion in various assets have been issued on-chain, compared to just $3.7 billion a year ago. This trend will continue to accelerate in 2026.
It is important to note that "Tokenization" and "Vaults" are two different RWA design patterns: the former digitizes real-world assets, while the latter directs on-chain capital toward offline yield scenarios.
In the future, I expect to see asset tokenization covering a broader range of areas: from precious metals like gold, to short-term credit for business operations, to public and private equity, and even more global fiat currencies. We can even be "bolder"—eggs, GPUs, energy derivatives, salary advances, Brazilian government bonds, yen… all can be tokenized.
Essentially, RWA is not about "putting more things on-chain," but about upgrading the global capital allocation method. The high barriers, low transparency, and decentralization of traditional markets can be redefined on public chains and combined with DeFi primitives to achieve composability.
Of course, many assets will still face challenges such as transfer restrictions, insufficient transparency, lack of liquidity, and inefficiencies in risk management and distribution, so infrastructure that addresses these issues will also be crucial.
The Revival of Agent-Driven Products is Coming
By Ash Egan
The core of the next generation of the internet is no longer the applications we swipe through, but the "smart agents" we converse with.
We all know that the share of bots and agents in all online activities is rapidly increasing. Rough estimates suggest that this proportion has now reached about 50%, including both on-chain and off-chain activities. In the cryptocurrency space, bots are increasingly involved in trading, management, assistance, scanning contracts, and executing various operations on our behalf, from token trading and fund management to smart contract auditing and game development.
This marks the beginning of a "programmable, agent-driven internet." And 2026 will be the first year where crypto product design is truly "agent-centric" (in a positive, non-dystopian way).
The future form is still gradually emerging, but at least for me, I hope to spend less time clicking between pages and more time managing on-chain agents in a chat-like interface: like Telegram, but the conversation partners are "agents for specific applications/tasks." They will be able to formulate and execute complex strategies, automatically gather the most relevant information from the network, and present trading results, risks and opportunities to watch, and curated information. I assign them tasks, and they will track opportunities, filter out all irrelevant information, and execute at the optimal moment.
The infrastructure needed for this vision is already prepared on-chain. By combining default open data graphs, programmable micropayments, on-chain social graphs, and cross-chain liquidity tracks, we have everything needed to support a dynamic agent ecosystem. The "plug-and-play" nature of the crypto world means that the cumbersome processes and barriers agents face will be significantly reduced. Compared to Web2 infrastructure, the blockchain's readiness for this agent revolution cannot be overstated.
This may be the most critical point here: this is not just automation, but a true liberation from the data silos, various frictions, and unnecessary waiting of Web2. We are already witnessing this transformation in the search domain: about 20% of Google's searches now directly provide AI overviews, and data shows that once users see this overview, their likelihood of clicking traditional search result links drops significantly. Manually flipping through pages of information is becoming redundant. Programmable agent-driven networks will extend this experience to all applications we use daily, which I believe is a tremendous benefit.
In this new era, we will: reduce mindless scrolling, decrease emotional panic trading, and completely eliminate time zone differences (no more saying "wait for the Asian markets to wake up"). Whether for developers or ordinary users, interactions with the on-chain world will become simpler and more expressive.
As more assets, systems, and users enter the chain, this cycle will continuously reinforce itself and accelerate growth. The more opportunities on-chain, the more smart agents deployed, the more value unlocked. It is a cycle. However, what we build now and how we build it will determine whether this "agent-driven network" ultimately becomes a thin layer of noise and automation or truly ignites a revival of empowered users, filled with vitality and innovation.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
