Archetype: Explore the 2026 investment perspective, focusing on the changes in these sectors

Writing by: Aadharsh Pannirselvam, Tommy Hang, Eskender Abebe, Katie Chiou, Danny Sursock, Dmitriy Berenzon, Ash Egan, Archetype

Compiled by: JW, Techub News

Looking ahead to 2026, the Archetype team is focusing on key trends, forces, and structural changes.

Building application chains finally makes sense

The core judgment is quite clear: truly competitive blockchains will increasingly be born for “specific applications.”

It’s not about having a general-purpose chain first and then trying to fit various applications; rather, from the beginning, designing, building, and continuously adjusting around the needs of the application itself. Such chains will perform very strongly over the next year.

The reason is that the new wave of developers, users, institutions, and capital entering the crypto world is completely different from the early days. They have clear cultural preferences and very specific requirements for user experience, no longer primarily concerned with abstract value propositions. In reality, these needs can sometimes be met by existing infrastructure, but often cannot.

Take applications like Blackbird and Farcaster, which greatly de-emphasize crypto perception and target mainstream users. Some design choices that were considered “unacceptable” three years ago—such as node centralization, a single sequencer, or even fully customized data systems—are now becoming reasonable solutions to enhance experience.

The same applies to trading or stablecoin-related applications like Hyperliquid and GTE. The competition in these systems fundamentally depends on latency, matching efficiency, and price quality. In scenarios where milliseconds decide life or death, many “principled issues” naturally take a backseat to user experience.

Of course, not all applications are suitable for this approach.

An important emerging counterbalance is the obvious rise in privacy needs among institutions and retail investors. Different user groups, use cases, and risk preferences mean the infrastructure they rely on should also take different forms.

The good news is that customizing a chain for a specific application is no longer a high-threshold engineering task. Compared to two years ago, it’s more like assembling a custom PC.

You can choose to fully configure each component yourself or base adjustments on mature solutions. The model provided by Digital Storm or Framework is essentially: using verified, workable combinations, replace or streamline modules as needed, ensuring performance while avoiding unnecessary complexity. This approach offers greater modularity and controllability. Applications can retain only the components they truly need, while ensuring overall system stability and scalability.

When consensus mechanisms, execution layers, data storage, and liquidity modules become primitives that can be freely combined and adjusted, applications will naturally form highly differentiated “chain forms.” These forms will continually reflect their understanding of user experience and serve very specific target groups.

This differentiation is similar to different types of computing devices: ToughBook, ThinkPad, desktop, or MacBook—looking very different on the surface, but sharing a lot of common logic underneath. The key is that each component becomes an adjustable parameter, not a restrictive inheritance. The acquisition of Informal Systems’ Malachite by Circle clearly signals a trend: the emphasis on sovereignty over proprietary blockchain spaces is becoming consensus.

In the coming year, we may see roles similar to “HashiCorp or Stripe Atlas in the blockchain space,” provided by teams like Commonware, Delta, offering standardized primitives and default configurations, making it easier for applications to define and control their own chain resources.

Ultimately, this model enables applications to do one thing truly: own their own blockchain space and cash flow directly, turning the chain itself into a long-term competitive advantage.

Prediction markets will continue to evolve

In this cycle, prediction markets are undoubtedly one of the most closely watched application categories.

When the weekly trading volume of all crypto prediction platforms exceeds $2 billion, the data proves that this track is no longer just a niche experiment.

As popularity rises, many projects attempt to copy, replace, or even directly challenge top platforms like Polymarket and Kalshi. But beyond emotions, there’s only one truly important question: which teams are solving core structural issues, and which are just riding the wave.

From a market structure perspective, the most important factors remain reducing spreads and increasing open interest. Although current market creation is still permissioned, liquidity on both maker and taker sides remains thin.

There is clear room for improvement in areas such as better order routing mechanisms, more suitable liquidity models, or leveraging lending to improve capital efficiency. These factors are crucial for scaling the product. The structure of trading categories also directly affects platform competitiveness. For example, Kalshi in November saw over 90% of trading volume concentrated in sports markets, indicating a natural advantage in certain liquidity structures. Conversely, Polymarket leads significantly in crypto and political markets, with trading volumes several times higher than Kalshi. Still, on-chain prediction markets are far from mainstream scale.

The 2025 Super Bowl is a vivid comparison: in just one day, traditional offline betting platforms handle over $23 billion in trading volume, far exceeding all current on-chain prediction markets combined.

To narrow this gap, the solution isn’t marketing or narratives but teams capable of truly solving structural issues. This remains the most important area to monitor over the next year.

Agent-based curators will push DeFi toward scalability

DeFi asset management exists in two extremes: purely algorithmic (hardcoded interest rate curves, fixed rebalancing rules) or purely manual (risk committees, active fund managers). Agent-based curators represent a third mode: they do not simply execute preset rules but use AI Agents (LLM + tools + feedback mechanisms) to continuously judge risks, returns, and strategies, and participate in parameter setting.

Take Morpho markets as an example: to build sustainable yield products, clear collateral policies, LTV caps, and risk parameters must be defined. Currently, this process heavily relies on human judgment, inherently limiting scalability. The introduction of Agents is fundamentally an attempt to address this issue.

In the future, we are likely to see agent-based curators directly compete with traditional algorithmic models and human managers within the same markets.

Market opinions on AI’s role in trading and asset management tend to polarize: some believe it will rapidly replace human traders, others think it cannot handle real market uncertainties.

But the real change isn’t about “replacement” itself; it’s about architectural adjustments. Agents are more likely to take on roles in strategy design, constraint setting, and portfolio management rather than participating directly in latency-sensitive execution layers. As inference costs continue to decline, computing power itself will become a new competitive factor.

In such an environment, the most advantageous DeFi products may not come from the smartest individuals but from teams capable of deploying scalable intelligent decision systems.

Short videos are becoming the new entry point for trading

Short videos are increasingly becoming the primary way for people to discover, understand, and ultimately purchase content.

TikTok Shop surpassed $20 billion GMV in the first half of 2025 and continues to grow rapidly, already demonstrating the strength of this trend.

Instagram is gradually shifting Reels from a defensive feature to a core commerce engine. Whatnot’s practice further proves that real-time, personalized content significantly improves conversion efficiency compared to traditional e-commerce models.

The logic behind this is straightforward: watching real-time content makes it easier to make quick decisions. As recommendation feeds and checkout processes merge, content itself becomes a trading interface, and creators naturally evolve into distribution nodes. AI accelerates this process. With decreasing content production costs and increasing testing frequency, platforms are optimizing conversion rates per second of video.

In this environment, payment systems must be fast, cheap, and highly composable. Microtransactions, automatic revenue sharing, and contribution attribution become fundamental capabilities.

These are scenarios where crypto systems are naturally suited. In streaming-based business models, it’s hard to imagine a reliable operation without crypto as the underlying settlement and incentive tool.

Blockchain is driving new paths for AI expansion

In recent years, AI attention has mainly focused on competition between large cloud providers and top startups. Meanwhile, a group of crypto-native teams are making substantial progress in distributed training and inference.

These efforts have moved from theory to testing and even into production environments. Teams like Ritual, Pluralis, Exo, Odyn, Ambient, and Bagel are at the forefront of this wave of exploration. By training models in globally distributed environments and combining asynchronous communication and parallel mechanisms, traditional scaling bottlenecks are being redefined.

Simultaneously, new consensus mechanisms and privacy technologies are making verifiable, confidential inference practically feasible. Moreover, some new blockchain architectures are attempting to truly integrate smart contracts with more general computing structures, providing a foundation for autonomous agents.

The foundational capabilities are in place.

The key now is whether this can scale to production levels and demonstrate that this approach isn’t just a conceptual experiment but a real driver of AI capability evolution.

RWA is moving toward real-world scale

The industry has discussed RWA (Real-World Assets) for years. But with the proliferation of stablecoins, mature deposit and withdrawal channels, and increasingly clear regulatory environments, tokenization is finally entering a scale phase.

According to data from RWA.xyz, the total tokenized assets issued on-chain have exceeded $18 billion, up from less than $4 billion a year ago.

It’s important to distinguish two modes:

Tokenization maps off-chain assets onto the blockchain; Vaults allow on-chain capital to directly participate in off-chain yields. In the future, the types of on-chain assets will become more diverse—from commodities, private credit, stocks, foreign exchange, to some non-traditional assets.

But the key isn’t just “more asset types.”

The real significance lies in using blockchain to make the inefficient, opaque capital allocation process more programmable and liquid.

Of course, this process still faces issues like transfer restrictions, lack of liquidity, and risk management, so the infrastructure supporting it is equally important.

Agent-driven product cycles are imminent

The next-generation internet interaction core is shifting from “platform” to “Agent.” Whether on-chain or off-chain, autonomous agents already handle a significant portion of network activity. In crypto, they participate in trading, asset management, information filtering, contract auditing, and even content creation.

2026 is likely to be a clear turning point.

Crypto product design will start prioritizing Agents over human interfaces. The ideal form isn’t more buttons but fewer operations. Users will give goals via conversational interfaces, while Agents handle information filtering, strategy execution, and feedback. The supporting infrastructure already exists: open data, programmable payments, on-chain identities, and cross-chain liquidity.

Compared to Web2, blockchain is more friendly to Agents because it provides open interfaces rather than closed systems. This isn’t just about efficiency—it’s a shift in interaction paradigm. As search, trading, and execution are gradually taken over by Agents, humans can focus on higher-level judgment.

As more assets and activities go on-chain, this cycle will amplify: opportunities increase, Agents multiply, and value gets released.

The only real question is: are we building systems that amplify value or just noise?