Author: Aadharsh Pannirselvam
Compiled by: White Paper Blockchain
It's simple: chains designed, built, and tuned for applications will shine. The best application chains next year will be carefully assembled around primitives and first principles.
The recent influx of developers, users, institutions, and capital into the chain is different from before: they possess a specific culture (understood as the definition of user experience), and they value this culture over abstract ideals like decentralization and censorship resistance. In practice, this sometimes aligns with our existing infrastructure and sometimes does not.
For crypto abstractions like Blackbird or Farcaster, which are aimed at non-experts, aspects of user experience that were once seen as heretical centralization design decisions—such as colocated nodes, single sequencers, and custom databases—actually make a lot of sense. This is also true for stablecoin chains and trading venues like Hyperliquid* and GTE that rely on milliseconds, minimal price ticks, and optimal pricing.
But this does not apply to every new application.
For example, balancing this comfort with centralization is the growing interest in privacy from institutions and retail. The demands and expected experiences of crypto applications may differ vastly, and so should their infrastructure.
Fortunately, assembling chains that cater to these specific user experience definitions from scratch is far less complicated than it was two years ago. Today, it is essentially no different from assembling a custom PC.
Of course, you can pick every driver, 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 tailored to different needs. If you're somewhere in between, you can add your own parts to the components they've already selected and know work well together. This provides you with higher modularity, flexibility, and the ability to eliminate components you don’t actually need, while ensuring the final product runs at a high level.
By assembling and adjusting primitives such as consensus mechanisms, execution layers, data storage, and liquidity, applications create culturally unique forms that continuously reflect different needs (understood as the concept of user experience), catering to their unique target audiences and ultimately retaining value. These forms can look as different as ToughBooks, ThinkPads, desktop towers, or MacBooks, but they also tend to converge and coexist to some degree—not every such computer has its own unique operating system. More importantly, every necessary component becomes a 'knob' that applications can iterate and adjust as needed, without worrying about making destructive changes to the parent protocol.
Given Circle's acquisition of Malachite under Informal Systems, having sovereign custom block space is clearly a broader priority at present. In the coming year, I look forward to seeing applications and teams define and own their chain resources around primitives and reasonable defaults provided by companies like Commonware and Delta, somewhat like HashiCorp or Stripe Atlas for blockchain and block space.
Ultimately, this will enable applications to own their cash flows directly and leverage their uniquely constructed forms to provide the best user experience as a lasting moat.
Prediction markets will continue to innovate
One of the most well-received applications of this cycle is prediction markets. With weekly trading volume across all crypto venues hitting a record $2 billion, it's evident that this category has made meaningful strides towards becoming a mainstream consumer product.
This momentum creates a tailwind for adjacent projects aimed at complementing or replacing current market leaders like Polymarket and Kalshi. But in the hype, distinguishing true innovation from noise will ultimately be the key to determining what is worth paying attention to in 2026.
From a market structure perspective, I am particularly excited about solutions that reduce spreads and deepen open interest. While market creation is still permissioned and selective, the liquidity of prediction markets remains relatively thin for makers and takers. There is a real opportunity to improve optimal routing systems, different liquidity models, and collateral efficiency through products like derivatives.
Trading volume by category is also a major driver of some venues outperforming others. For instance, over 90% of Kalshi's trading volume in November came from sports markets, which highlights that some venues are inherently more capable of competing for favorable liquidity. In contrast, Polymarket's trading volume in crypto-related and political markets is over 5 to 10 times that of Kalshi.
Nevertheless, on-chain prediction markets have a long way to go to achieve true mass adoption. A good reference point is the Super Bowl in 2025; that alone created $23 billion in transaction volume in the off-chain betting market, which is over 10 times the total daily transaction volume of all on-chain markets currently.
Bridging this gap will require sharp, inspired teams to address core prediction market issues, and I will closely monitor these participants in the coming year.
Agentic Curators will expand DeFi
The curation layer of DeFi sits at two extremes: pure algorithm (hardcoded interest rate curves, fixed rebalancing rules) or pure human (risk committees, active managers). Agentic curators represent a third institution: AI agents (LLMs + tools + loops) that manage curation and risk strategies in treasuries, derivatives markets, and structured products. They do not just execute fixed rules, but reason about risk, reward, and strategy.
Think about the curator role in the Morpho market, where 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 scale it. Soon, you will see agentic curators competing directly with algorithmic models and human managers.
When will we see DeFi's 'Move 37' (referring to the unexpected 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 'hallucination toys' that will never withstand the test of real markets. Both views miss the architectural shift. Agents will bring unemotional execution, systematic adherence to strategy, and flexible reasoning into areas where humans tend to create noise and pure algorithms are too brittle. They will likely supervise and/or compose lower-level algorithms rather than replace them. LLMs act as architects for designing safe shells, while deterministic code remains on the hot latency path.
When the cost of deep reasoning drops to a few cents, the most profitable treasury will not be the one with the smartest humans, but the one with the most computational resources.
Short videos are the new storefronts
Short videos are rapidly becoming the default interface for people to discover (and eventually purchase) the content they love. TikTok Shop achieved over $20 billion in gross merchandise value (GMV) in the first half of 2025, nearly doubling year-over-year, and is quietly training global audiences to view entertainment as a storefront.
In response, Instagram has transformed Reels from a defensive feature into a revenue engine. This format has led to more impressions and is taking up an increasingly larger share of Meta's projected advertising revenue for 2025. Whatnot has already proven that real-time, personalized sales conversion rates are unmatched by traditional e-commerce.
The thread running through all of this is simple: when people watch content in real-time, they make decisions faster. Every swipe becomes a decision point. Platforms are well aware of this, which is why the boundary between recommendation feeds and checkout processes is disappearing. The information stream is the new point of sale, and every creator is a distribution channel.
AI is further driving this shift. It lowers the cost of producing videos, increases the amount of content, and makes it easier for creators and brands to test ideas in real-time. More content means a larger conversion surface area, and platforms respond by optimizing every second of video to achieve purchase intent.
Cryptocurrency aligns perfectly with this shift. Faster content requires faster, more cost-effective payment rails. As shopping becomes frictionless and directly embedded in the content itself, you need a system that can settle small payments, programmatically allocate and split revenues, and track chaotic impacts on on-chain contributions. Cryptocurrency is built for such processes, making it hard to imagine a hyper-scale streaming native commerce era without it.
Blockchain will drive new AI scaling laws
Over the past few years, the focus of AI has been on the billions of dollars arms race between super-scale companies and startup giants, while decentralized innovators have been groping in the shadows.
But as attention shifts elsewhere, some crypto-native teams have made significant strides in decentralized training and reasoning, and the forefront of this quiet revolution has slowly moved from whiteboards to testing and production environments.
Now, teams like Ritual*, Pluralis, Exo*, Odyn, Ambient, Bagel, etc., are ready for prime time. This new generation of competitors is poised to unleash explosive orthogonal impacts on the foundational trajectory of AI.
By training models in globally distributed settings and leveraging new methods of asynchronous communication and parallelism validated in production-scale operations, scalability constraints can be broken.
The combination of new consensus mechanisms and privacy primitives makes verifiable and confidential reasoning a very realistic choice in the toolkit for on-chain builders.
And the revolutionary blockchain architecture will combine (true) smart contracts with expressive computational structures, simplifying the use of cryptocurrency as a medium of exchange for autonomous AI agents.
The foundational work has been done.
The current challenge is to scale these infrastructures to production environments and demonstrate why blockchain can drive foundational AI innovations beyond philosophy, ideology, or commodified fundraising experiments.
Real World Assets (RWAs) will usher in real-world adoption
We have been hearing about tokenization for years, but with the mainstream adoption of stablecoins, the emergence of smooth and robust on/off-ramp channels, and clearer global regulation and support, we are finally seeing large-scale adoption of RWAs. According to RWA.xyz*, as of the time of writing, the issued tokenized assets have exceeded $18 billion, up from just $3.7 billion a year ago, and I expect this momentum to accelerate by 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 create bridges between on-chain capital and off-chain yields.
I am 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 credit raised for working capital and payment financing, to crowdfunding and public equity, and more global currencies. We also need to unleash our imagination. I want to see eggs, GPUs, energy derivatives, earned-wage access, Brazilian government bonds, yen, etc., all on-chain!
It needs to be clear that this is not just about putting more things on-chain. It is 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 have already built.
Ultimately, many of these assets will undoubtedly face challenges in transferability, transparency, liquidity, risk management, and distribution, so infrastructure that alleviates these challenges is equally important and exciting!
An agent-driven product renaissance is on the horizon
The next generation of networks will be less influenced by the platforms we scroll through and more by the agents we converse with.
We all know that the contribution of bots and agents to all network activity is rapidly increasing. Roughly estimated, including on-chain and off-chain activity, today it accounts for about 50%. In the crypto space, bots increasingly represent us in trading, curating, assisting, scanning contracts, and taking action, covering every aspect from trading Tokens and managing treasuries to auditing smart contracts and developing games.
This is the era of programmable, agent-driven networks. While we've been in it for a while, 2026 will be the year when crypto product design starts to cater more to bots rather than humans (in a positive, liberating, non-dystopian way).
What this looks like is still taking shape, but I personally hope to spend less time clicking around websites and more time interacting with a simple chat-based interface where I manage on-chain bots. Imagine Telegram, but the conversation partners are specific agents for applications/tasks. They will be able to form and execute complex strategies, search the web for information and data most relevant to me, and report back on trade outcomes, risks, and opportunities to watch, as well as curated information. I will give them a task, and they will track opportunities, filter out all the noise, and execute at the optimal moment.
The infrastructure to achieve this already exists on-chain. By combining the default open data graphs and programmatic micropayments 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 fewer bureaucratic hurdles and dead ends for agents to navigate. The readiness of blockchain for this, relative to Web2 infrastructure, cannot be overstated.
And this might be the most important point here. It's not just about automation, but about being liberated from Web2 silos, freed from friction. Freed from 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 significantly less likely to click on traditional search result links. Manual filtering of pages is becoming unnecessary. Programmable agent-driven networks will further extend this to the applications we use, and I believe 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 onto the chain, this loop 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 a layer of noise and automation or ignites a renaissance of empowered and dynamic products.
