Looking for the main investment themes in 2026: Which trends are worth paying attention to?

null

Author: Aadharsh Pannirselvam

Translation: Blockchain Talk

It’s simple: chains designed, built, and tuned for applications will shine. Next year’s top application chains will be carefully assembled from primitives and first principles.

The recent wave of developers, users, institutions, and capital entering on-chain is different from before: they have specific cultures (understood here as: definitions of user experience), and they value these cultures more than abstract ideals like decentralization and censorship resistance. In practice, sometimes this aligns with our existing infrastructure, and sometimes it doesn’t.

For crypto abstractions like Blackbird or Farcaster—applications aimed at non-experts—the particularly important aspects of user experience include centralized design decisions that even three years ago might have been considered heretical—such as colocated nodes, single sequencers, and custom databases—which are actually very reasonable. The same applies to stablecoin chains and trading venues like Hyperliquid* and GTE that depend on millisecond, minimal price fluctuations (ticks), and optimal prices.

But this isn’t true for every new application.

For example, balancing this comfort with centralization is the growing interest in privacy from institutions and retail users. The demands and expected experiences of crypto applications can differ sharply, so their infrastructure should reflect that.

Fortunately, assembling chains from scratch that cater to these specific user experience definitions is far less complex than two years ago. Today, it’s very much like assembling a custom PC.

Of course, you can select every drive, fan, and cable yourself. But if you don’t need that level of granularity (which is likely the case), you can use services like Digital Storm or Framework, which offer a range of pre-built, customizable PCs tailored for different needs. If you’re somewhere in between, you can add your components to what they’ve already chosen and know will work well together. This provides higher modularity, flexibility, and the ability to strip out components you don’t actually need—while ensuring the final product runs at a high level.

By assembling and tuning primitives like consensus mechanisms, execution layers, data storage, and liquidity, applications create culturally unique forms that continually reflect different needs (understood here as: conceptions of user experience), catering to their specific target audiences and ultimately retaining value. These forms may look different—like ToughBooks, ThinkPads, desktop towers, or MacBooks—but they also tend to converge and coexist to some extent—not every such device has its own unique OS. More importantly, each essential component becomes a “knob” that applications can iterate on and tune as needed, without worrying about destructive changes to parent protocols.

Given Circle’s acquisition of Malachite from Informal Systems, owning sovereignty over custom block spaces is clearly a broader priority now. Over the next year, I look forward to seeing applications and teams define and own their chain resources around primitives and sensible defaults offered by companies like Commonware and Delta—much like how HashiCorp or Stripe Atlas do for blockchain and block space.

Ultimately, this will enable applications to directly own their cash flows and leverage their unique forms to provide the best user experience as a lasting moat.

Prediction Markets Will Continue to Innovate

One of the most acclaimed applications this cycle is prediction markets. With weekly trading volumes across all crypto venues reaching a record $2 billion, it’s clear this category has made meaningful strides toward becoming a mainstream consumer product.

This momentum creates tailwinds for adjacent projects aiming to supplement or replace current market leaders like Polymarket and Kalshi. But amidst hype, distinguishing true innovation from noise is crucial in deciding what will matter in 2026.

From a market structure perspective, I am particularly excited about solutions that reduce spreads and deepen open interest. While market creation remains permissioned and selective, liquidity in prediction markets remains relatively thin for makers and takers. There are real opportunities to improve the optimal routing systems, various liquidity models, and collateral efficiency through products like borrowing derivatives.

Trading volume by category is also a key driver of dominance among venues. For instance, in November, over 90% of Kalshi’s trading volume came from sports markets, highlighting that some venues are naturally better positioned to compete for favorable liquidity. By contrast, Polymarket’s trading volume in crypto-related and political markets is more than 5 to 10 times that of Kalshi.

That said, on-chain prediction markets still have a long way to go before achieving truly large-scale adoption. A good benchmark is the 2025 Super Bowl; just this event generated over $23 billion in off-chain betting volume, more than 10 times the total daily trading volume of all on-chain markets today.

Closing this gap will require sharp, inspired teams to solve core prediction market challenges, and I’ll be watching closely in the coming year.

Agentic Curators Will Expand DeFi

The curation layer of DeFi sits at two extremes: purely algorithmic (hardcoded interest rate curves, fixed rebalancing rules) or purely human (risk committees, active managers). Agentic curators represent a third paradigm: AI agents (LLMs + tools + loops) that manage curation and risk strategies in treasuries, borrowing markets, and structured products. They do not merely execute fixed rules but reason about risk, yield, and strategy.

Think of the curator role in Morphos markets—someone must define collateral policies, loan-to-value (LTV) limits, and risk parameters to generate yield products. Today, this is a human bottleneck. Agents can extend it. Soon, you’ll see agentic curators competing head-to-head with algorithmic models and human managers.

When will we see a “Move 37” moment for DeFi? (referring to the surprising move made by AlphaGo against Lee Sedol in Go)

When I talk to crypto fund managers about AI, I get two answers: either LLMs will soon automate every trading desk, or they’re “illusion toys” that can never withstand real markets. Both miss the architecture shift. Agents will bring emotionless execution, systematic strategy adherence, and flexible reasoning to domains where humans tend to generate noise and pure algorithms are too fragile. They are likely to supervise and/or compose lower-level algorithms rather than replace them. LLMs act as architects designing safe shells, while deterministic code remains in the high-speed (hot latency) path.

As the cost of deep reasoning falls to mere cents, the most profitable treasuries will not be those with the smartest humans, but those with the most computational resources.

Short-form Video Is the New Storefront

Short videos are rapidly becoming the default interface for discovering (and ultimately buying) content people love. In the first half of 2025, TikTok Shop achieved over $20 billion in gross merchandise volume (GMV), nearly doubling from the previous year, quietly training a global audience to view entertainment as a storefront.

In response, Instagram has transformed Reels from a defensive feature into a revenue engine. This format drives more impressions and is capturing an increasing share of Meta’s projected advertising revenue in 2025. Whatnot has already proven that real-time, personalized-driven sales conversions outperform traditional e-commerce.

The core message is simple: when people watch content in real-time, they make faster decisions. Every swipe becomes a decision point. Platforms understand this well, which is why the boundary between recommendation feeds and checkout processes is disappearing. The feed is the new sales point—every creator is a distribution channel.

AI further accelerates this shift. It lowers the cost of making videos, increases content volume, and makes it easier for creators and brands to test ideas in real-time. More content means more conversion surface area, with platforms optimizing every second of video to drive purchase intent.

Cryptocurrency aligns perfectly with this shift. Faster content requires faster, more cost-effective payment rails. As shopping becomes frictionless and embedded directly into content, you need a system capable of settling micro-payments, programmatically allocating and splitting revenue, and tracking contribution on the messy on-chain chain of effects. Crypto is built for this, and it’s hard to imagine a super-scale streaming-native commerce era without it.

Blockchain Will Drive New AI Scaling Laws

Over the past few years, AI focus has been on billion-dollar arms races among mega-companies and startups, with decentralized innovators working in the shadows.

But as attention shifts elsewhere, some crypto-native teams are making huge advances in decentralized training and reasoning. The frontlines of this quiet revolution are gradually moving from whiteboards to testing and production environments.

Now, teams like Ritual*, Pluralis, Exo*, Odyn, Ambient, Bagel are ready for prime time. A new generation of competitors is poised to unleash explosive orthogonal impacts on the foundational trajectory of AI.

By training models in distributed global setups, leveraging new methods for asynchronous communication and parallelism validated at scale in production, we can break scaling constraints.

The combination of new consensus mechanisms and privacy primitives makes verifiable and confidential reasoning very realistic options in the on-chain toolkit.

Revolutionary blockchain architectures will combine (true) smart contracts with expressive computational structures, simplifying the use of cryptocurrencies as exchange mediums for autonomous AI agents.

The groundwork has been laid.

The current challenge is to scale this infrastructure into production and demonstrate why blockchain can push AI innovation beyond philosophy, ideology, or mimetic fundraising experiments.

Real-World Assets (RWAs) Will Reach Real-World Adoption

We’ve been hearing about tokenization for years, but with the mainstream adoption of stablecoins, smooth and powerful on/off ramps, and clearer regulation/support globally, we’re finally seeing large-scale RWA adoption. According to data from RWA.xyz*, the total tokenized assets issued have surpassed $18 billion so far, up from $3.7 billion a year ago—I expect this momentum to accelerate into 2026.

It’s important to note that tokenization and vaults are different design patterns for RWAs: tokenization creates on-chain representations of off-chain assets, while vaults build a bridge between on-chain capital and off-chain yields.

I’m excited to see tokenization and vaults providing access to a broad range of physical and financial assets—from commodities like gold and rare metals to credit raised for operational and payment financing, to equity raises and public offerings, and more global currencies. We should also open our imagination. I want to see eggs, GPUs, energy derivatives, earned-wage access, Brazilian government bonds, Japanese yen—all on-chain!

It’s important to clarify that this isn’t just about putting more things on-chain. It’s about upgrading how the world allocates capital through public blockchains, making opaque, slow, and siloed markets accessible, programmable, and liquid. Once on-chain, we’ll enjoy the composability benefits of the DeFi primitives we’ve already built.

Finally, many of these assets will undoubtedly face challenges related to transferability, transparency, liquidity, risk management, and distribution—building infrastructure to alleviate these challenges will be equally important and exciting!

An Agent-Driven Product Renaissance Is Coming

The next-generation network will be less influenced by our scrolling platforms and more by the intelligent agents we converse with.

We all know that bots and agents are rapidly increasing their contribution to all network activity. Rough estimates, including on-chain and off-chain activity, suggest they now account for about 50%. In crypto, bots increasingly trade for us, curate content, assist, scan contracts, and take actions—covering everything from token trading and treasury management to auditing smart contracts and developing games.

This is the era of programmable, agent-driven networks. While we’ve been in it for some time, 2026 will be the year when crypto product design begins to cater more to robots than humans (in a positive, liberating, non-dystopian way).

What it looks like is still taking shape, but I personally want to spend less time clicking websites and more time interacting with a simple chat interface where I manage on-chain bots. Imagine Telegram, but the conversational partner is a specific smart agent for an application/task. They’ll be able to form and execute complex strategies, search the web for relevant data, report trading outcomes, risks, opportunities, and curated information. I’ll give them a task, and they’ll track opportunities, filter out noise, and execute at the right moments.

The infrastructure to make this happen already exists on-chain. Combining open data graphs and programmable micro-payments with on-chain social graphs and cross-chain liquidity rails, we have everything we need to support a dynamic agent ecosystem. The plug-and-play nature of crypto means fewer bureaucratic hoops and dead ends for these agents. Compared to Web2 infrastructure, blockchain is more prepared than ever.

And this might be the most important point: it’s not just about automation, but freeing ourselves from Web2 silos, friction, and waiting. We see this shift happening in search: about 20% of Google searches now produce an AI overview, and data shows that when people see this overview, they are far less likely to click traditional search links. Manually sifting through pages is becoming unnecessary. Programmable, agent-driven networks will extend this further into the apps we use, and I think that’s a good thing.

This era will reduce “doomscrolling.” Reduce panic trading. Eliminate time zone delays (“waiting for Asia to wake up”). Interacting with the on-chain world will become easier and more expressive for every developer and user.

As more assets, systems, and users find ways onto the chain, this cycle will compound.

More on-chain opportunities → Deploy more agents → Unlock more value. Repeat.

But what we build now—and how we build it—will determine whether this agent-driven network merely becomes noise and automation or sparks a renaissance of empowering and dynamic products.