Author: Aadharsh Pannirselvam
Compiled by: Plain Talk Blockchain
It's simple: Chains designed, built, and tailored for applications will shine. The best application chains next year will be meticulously assembled from primitives and first principles.
The recent wave of developers, users, institutions, and capital entering the on-chain space is different from before: they possess a specific culture (understandable as: the definition of user experience), which they value more than abstract ideals like decentralization and censorship resistance. In practice, sometimes this aligns with our existing infrastructure, and sometimes it does not.
For crypto-abstracted, non-professional-oriented applications like Blackbird or Farcaster, particularly important aspects of the user experience—centralized design decisions that would have seemed heretical even three years ago—such as co-located nodes, single sequencers, and custom databases—are actually quite reasonable. The same goes for stablecoin chains and trading venues like Hyperliquid* and GTE, which rely on milliseconds, minimal price fluctuations (ticks), and optimal pricing.
But this doesn't apply to every new application.
For example, balancing this comfort with centralization is the growing interest in privacy from both institutions and retail. The demand and expected experience for crypto applications can be vastly different, so their infrastructure should be too.
Fortunately, assembling chains from scratch to cater to these specific user experience definitions is far less complex than it was two years ago. Today, it's practically no different from assembling a custom PC.
Of course, you could pick every drive, fan, and cable yourself. But if you don't need that level of granularity (which is likely the case), then you can use services like Digital Storm or Framework, which offer a range of pre-built custom PCs for different needs. If you're somewhere in between, you can add your own parts to components they've selected and know work well together. This gives you greater modularity, flexibility, and the ability to strip out components you don't actually need, while ensuring the final product performs at a high level.
By assembling and adjusting primitives like consensus mechanisms, execution layers, data storage, and liquidity, applications create culturally unique forms that continuously reflect different needs (understandable as: the concept of user experience), cater to their unique target audiences, and ultimately capture value. These forms can look as different as ToughBooks, ThinkPads, desktop tower PCs, or MacBooks, but they also converge and coexist to some extent—not every such computer has its own unique operating system. More importantly, each necessary component becomes a 'knob' that the application can iterate and adjust as needed, without worrying about making disruptive changes to the parent protocol.
Given Circle's acquisition of Malachite under Informal Systems, owning sovereign, custom block space is clearly a broader priority currently. In the coming year, I'm excited to see applications and teams define and own their chain resources around primitives and sensible defaults provided by companies like Commonware and Delta, somewhat like a HashiCorp or Stripe Atlas for blockchains and block space.
Ultimately, this will enable applications to directly own their cash flow and leverage the unique forms they've built to provide the best user experience in their own way, as a lasting moat.
Prediction Markets Will Continue to Innovate
One of the most acclaimed application categories this cycle is prediction markets. With weekly trading volume across all crypto venues reaching a record $2 billion, it's clear that the category has taken meaningful steps towards becoming a mainstream consumer product.
This momentum creates a tailwind for adjacent projects aiming to complement or replace current market leaders like Polymarket and Kalshi. But amidst the hype, distinguishing true innovation from noise is ultimately key to deciding what's worth watching in 2026.
From a market structure perspective, I'm particularly excited about solutions that reduce spreads and deepen open interest. Although market creation is still permissioned and selective, liquidity in prediction markets remains relatively thin for both makers and takers. There is a real opportunity to improve optimal routing systems, different liquidity models, and collateral efficiency through products like lending.
Volume by category is also a major driver for why some venues outperform others. For example, over 90% of Kalshi's trading volume in November came from sports markets, highlighting that some venues are naturally better equipped to compete for favorable liquidity. In contrast, Polymarket's volume on crypto-related and political markets is 5 to 10 times higher than Kalshi's.
Nevertheless, on-chain prediction markets still have a long way to go to achieve true mass adoption. A good reference point is the 2025 Super Bowl; this single event alone generated $23 billion in trading volume in off-chain betting markets, which is over 10 times the current total daily trading volume of all on-chain markets combined.
Bridging this gap will require sharp, inspired teams to solve core prediction market problems, and I will be closely watching these players in the coming year.
Agentic Curators Will Expand DeFi
The curation layer in DeFi exists at two extremes: purely algorithmic (hard-coded interest rate curves, fixed rebalancing rules) or purely human (risk committees, active managers). Agentic curators represent a third institution: AI agents (LLMs + tools + loops) that manage curation and risk strategies in vaults, lending markets, and structured products. They don't just execute fixed rules; they reason about risk, yield, and strategy.
Think of the curator role in a Morpho market, where someone must define collateral policies, loan-to-value (LTV) limits, and risk parameters to create yield products. Today, this is a human bottleneck. Agents can scale it. Soon, you will see agentic curators competing head-to-head with algorithmic models and human managers.
When will we see DeFi's "Move 37" (referring to the surprising brilliant move made by the Go AI AlphaGo against Lee Sedol)?
When I talk to crypto fund managers about AI, I get one of two answers: either LLMs are about to automate every trading desk, or they are "hallucinating toys" that can never withstand real market tests. Both views miss the architectural shift. Agents bring emotionless execution, systematic strategy adherence, and flexible reasoning to areas where humans are prone to noise and pure algorithms are too brittle. They are likely to supervise and/or compose lower-level algorithms rather than replace them. The LLM acts as the architect designing the safety shell, while deterministic code remains in the hot latency path.
When the cost of deep reasoning drops to a few cents, the most profitable vaults won't be the ones with the smartest humans, but the ones with the most computational resources.
Short-Form Video is the New Storefront
Short-form video is rapidly becoming the default interface for people to discover (and ultimately purchase) content they like. TikTok Shop achieved over $20 billion in Gross Merchandise Volume (GMV) in the first half of 2025, nearly doubling year-over-year, and is quietly training a global audience to see entertainment as a storefront.
In response, Instagram has transformed Reels from a defensive feature into a revenue engine. This format drives more impressions and accounts for an increasing share of Meta's projected ad revenue for 2025. Whatnot has proven that live, personality-driven sales conversions are unmatched by traditional e-commerce.
The common thread is simple: when people watch content in real-time, they make faster decisions. Every swipe becomes a decision point. Platforms know this well, which is why the line between the recommendation feed and the checkout process is blurring. The feed is the new point of sale, and every creator is a distribution channel.
AI further accelerates this shift. It lowers the cost of producing videos, increases the volume of content, and makes it easier for creators and brands to test ideas in real-time. More content means more surface area for conversion, and platforms respond by optimizing every second of a video for purchase intent.
Cryptocurrency fits perfectly into this shift. Faster content requires faster, more cost-effective payment rails. As shopping becomes frictionless and directly embedded into the content itself, you need a system that can settle micro-payments, programmatically distribute and split revenue, and track messy influence chains on-chain. Cryptocurrency is built for such processes, and it's hard to imagine the era of hyper-scale streaming-native commerce without it.
Blockchains Will Drive New AI Scaling Laws
Over the past few years, the focus in AI has been on the multi-billion dollar arms race between hyperscale companies and startup giants, while decentralized innovators fumbled in the shadows.
But while attention was elsewhere, some crypto-native teams made significant strides in decentralized training and inference, and the frontier of this quiet revolution has slowly moved from whiteboards to testing and production environments.
Now, teams like Ritual*, Pluralis, Exo*, Odyn, Ambient, Bagel are ready for prime time. This new generation of contenders promises to unleash explosive orthogonal effects on the fundamental trajectory of AI.
By training models in globally distributed setups, leveraging new methods for asynchronous communication and parallelism validated at production scale, it's possible to break scaling constraints.
The combination of new consensus mechanisms and privacy primitives makes verifiable and confidential inference a very real option in the on-chain builder's toolkit.
And revolutionary blockchain architectures will combine (true) smart contracts with expressive computational structures, thereby simplifying autonomous AI agents using cryptocurrency as a medium of exchange.
The foundational work is done.
The challenge now is to scale this infrastructure to production and prove why blockchains can drive fundamental AI innovation beyond philosophy, ideology, or skeuomorphic fundraising experiments.
Real World Assets (RWAs) Will See Real World Adoption
We've heard about tokenization for years, but with the mainstream adoption of stablecoins, the emergence of smooth and robust on/off ramps, and clearer regulation and support globally, we are finally seeing mass adoption of RWAs. According to RWA.xyz*, over $18 billion in tokenized assets have been issued at the time of writing, up from $3.7 billion a year ago, and I expect this momentum to accelerate in 2026.
It's important to note that tokenization and Vaults are different design patterns for RWAs: tokenization creates an on-chain representation of an off-chain asset, while Vaults create a bridge between on-chain capital and off-chain yield.
I'm excited to see tokenization and Vaults provide access to a wide range of physical and financial assets, from commodities like gold and rare metals to raised credit for working capital and payment financing, to raised and public equity, and more global currencies. Let's also free our imagination. I want to see eggs, GPUs, energy derivatives, earned-wage access, Brazilian government bonds, Japanese Yen, all on-chain!
To be clear, this isn't just about putting more stuff on-chain. It's about upgrading how the world allocates capital through public blockchains, making opaque, slow, and isolated markets accessible, programmable, and liquid. Once they are on-chain, we will enjoy the benefits of composability with 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, so infrastructure that mitigates these challenges is equally important and exciting!
An Agent-Driven Product Renaissance is Coming
The next generation of the web will be less influenced by the platforms we scroll through and more by the agents we talk to.
We all know that bots and agents are contributing a rapidly growing share of all web activity. Rough estimates, including on-chain and off-chain activity, put it at around 50% today. In crypto, bots are increasingly trading, curating, assisting, scanning contracts, and acting on our behalf, covering everything from trading tokens and managing vaults to auditing smart contracts and developing games.
This is the era of the programmable, agent-driven web. While we've been in it for a while, 2026 will be the year crypto product design starts catering more to bots than humans (in a positive, liberating, non-dystopian way).
What this looks like is still taking shape, but personally, I hope to spend less time clicking on websites and more time interacting with a simple chat-like interface where I manage on-chain bots. Imagine Telegram, but the conversations are with application/task-specific agents. They will be able to form and execute complex strategies, search the web for information and data most relevant to me, and report trading results, risks and opportunities that need attention, and curated information. I will give them a task, and they will track opportunities, filter out the noise, and execute at the optimal moment.
The infrastructure to achieve this already exists on-chain. Combining the default open data graph and programmable micro-payments with on-chain social graphs and cross-chain liquidity rails, we have everything needed to support a dynamic agent ecosystem. The plug-and-play nature of cryptocurrency means agents face less red tape and dead ends. The readiness of blockchains for this, compared to Web2 infrastructure, cannot be overstated.
And this might be the most important point here. This isn't just about automation; it's about liberation from Web2 silos. Liberation from friction. From waiting. We're already seeing this shift happening in search: about 20% of Google searches now generate an AI Overview, and data shows that when people see this overview, they are much less likely to click on traditional search result links. Manually sifting through pages is becoming unnecessary. The programmable agent-driven web will extend this further into the applications we use, and I think that's a good thing.
This era will allow us to reduce 'doomscrolling'. Reduce panic trading. Time zone differences will be eliminated (no more "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 their way on-chain, this cycle compounds.
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 web becomes just another layer of noise and automation, or ignites a renaissance of empowering and dynamic products.
