As we approach 2026, having moved beyond the infrastructure-focused approach of the past four years, the crypto industry is undergoing a profound paradigm shift. OKX Ventures defines this as the dawn of the "Kinetic Finance" era, where the core focus is no longer on network speed, but rather on the liquidity and earning efficiency of on-chain assets.
To get straight to the point, we believe that Crypto's future opportunities will focus on three core changes:
- Asset Transformation: From "On-Chain" to "Global Settlement". RWA will enable the seamless 24/7 on-chain circulation of everything in the real world (US Treasury bonds, real estate, IP), bringing about a qualitative change in capital efficiency.
- The main shift: from "human" to "AI Agent." The protagonist of transactions will change from humans to AI. DeFi protocols will become "financial APIs" invoked by AI, allowing funds to proactively seek the best global returns, as if they possess intelligence.
- Rule-changing: From "post-event regulation" to "code compliance." Privacy and compliance are no longer obstacles, but rather infrastructure embedded in the code, paving the way for the entry of large-scale institutional funds from Wall Street and other institutions.
We firmly believe that projects that can use code to address real-world trust costs and improve capital efficiency will become the cornerstone of the new era. OKX Ventures' investment strategy in infrastructure over the past few years has primarily focused on the robustness of underlying protocols and the expansion of network capacity. We will continue to seek out and support these future-defining builders.
In 2025, the industry saw many significant developments: in terms of compliant funding channels, the approval of BTC spot ETFs opened up traditional capital pathways, and its cumulative net inflows exceeded $50 billion, making crypto assets a standard choice for global macro hedging.
At the underlying technology level, Ethereum reduced the consensus layer communication load by more than 90% and increased the network Blob data throughput by 4 times through the Pectra upgrade in May and the subsequent Fusaka phase. Combined with the native account abstraction capability, it cleared the obstacles to high-frequency interaction of hundreds of millions of users.
On-chain transaction performance has seen a qualitative leap, with high-performance DEXs such as Hyperliquid repeatedly setting records of $20 billion in daily trading volume.
In terms of asset scaling, RWA has made a key leap, with the size of BlackRock's BUIDL fund alone exceeding $2.5 billion by the end of the year, proving that the "two-way valve" for on-chain and off-chain liquidity has been fully operational.
In 2025 alone, OKX Ventures invested in dozens of projects across multiple sectors, including RWA, Infra, DeFi, AI, stablecoins, and consumer applications, continuously committed to supporting innovation in the industry.
I. Deep Financialization of RWA
RWA is no longer simply about issuing a "digital receipt" for real-world assets (such as houses and bonds). We are entering the RWA 2.0 phase, the core of which is to turn the blockchain into a global 24/7 clearinghouse. Imagine that in the past, selling an asset took two days to process (T+2); now, using blockchain, it can be done instantly (T+0). This is not just a matter of speed; it fundamentally changes the efficiency of global capital operations.
RWA Asset Layering: From On-Chain US Treasuries to Synthetic US Equities and Non-Standard Credit
The US dollar is a globally accepted currency, which has led to the rapid development of stablecoins like USDT and USDC, which are mapped to US stocks. Consequently, we see many DEXs and CEXs venturing into US stock tokenization. However, a significant portion of global assets are still non-US dollar assets. These assets naturally exhibit liquidity gaps. For example, US Treasury bonds have extremely high liquidity, while real estate and private lending are deeply non-standard assets. The core of RWA 2.0 lies in abandoning the "one-size-fits-all" AMM model and building adaptive issuance and trading architectures for different asset tiers. Standardized assets are the easiest category to put on-chain and achieve scale.
According to data from RWA.xyz, the scale of tokenized US Treasury bonds has exceeded $7.3 billion (a year-on-year increase of over 300%). On-chain US stocks are becoming the second largest growth engine for standardized assets after US Treasury bonds, currently valued at approximately $500 million. Their core value lies in breaking the trading hour restrictions (achieving 24/7 trading) and geographical barriers of traditional stock markets. This trend indicates that on-chain finance will not only possess a "risk-free rate" (US Treasury bonds) but also "equity-like risk assets" (US stocks), thereby constructing a complete on-chain investment portfolio.
In contrast, active lending in non-standard assets, such as private lending, remains in the $8 billion range. This significant gap indicates that high-yield non-standard assets are still constrained by pricing and liquidity challenges. BCG predicts that the RWA market will reach $16 trillion by 2030, with 2026 marking a critical inflection point in this growth curve, where the on-chain non-stablecoin RWA market is expected to surpass $100 billion. We believe this is significant because it signifies RWA's transition from a niche experiment to a mainstream narrative of a trillion-dollar market.
RWA has moved beyond a simple mapping phase and evolved into a layered architecture based on asset liquidity characteristics.
Stablecoins are reshaping the global settlement network
Undoubtedly, stablecoins are the killer product for cryptocurrencies. They are far more than just trading pairs on exchanges; they represent a viable alternative for cross-border payments, with the potential to gradually replace the traditional SWIFT system. Traditional cross-border payments typically involve fees of 3-5% and a settlement cycle of 2-3 days. In contrast, on-chain stablecoin payments have fees below 1% and are almost instantaneous.
As of November 2025, the total annual settlement volume of on-chain stablecoins has exceeded $12 trillion, officially surpassing Visa's annual settlement volume. Currently, the market capitalization of stablecoins has stabilized above $210 billion, with more than 40% of the transaction volume occurring outside of trading hours (when traditional banks are closed), filling the "liquidity vacuum" in the global financial infrastructure.
Furthermore, the composability of full asset tokenization is also noteworthy: Leading DeFi protocols (such as Aave and MakerDAO) have completed the integration of RWA assets, creating a "Lego effect." Whether it's government bonds (such as BUIDL and USDY), real estate, or private lending, assets have been successfully used as the underlying collateral for DeFi lending protocols.
By the end of 2025, BlackRock's BUIDL fund and Ondo Finance's USDY had officially integrated with the Aave V4 and Sky (formerly MakerDAO) protocols. Approximately 30% of the tokenized government bonds (about $2.2 billion) on-chain were being used directly as underlying collateral for lending protocols, rather than sitting idle in wallets.
Traditional financial institutions will leverage the real-time T+0 clearing capabilities on the blockchain to increase capital utilization by 2-3 times, completing a substantial migration of their backend infrastructure to a decentralized ledger.
OKX Ventures' key focus projects
OKX Ventures focuses on and invests in infrastructure projects that can reduce trust costs and efficiency frictions at scale, addressing the three core bottlenecks of the current on-chain capital market (including RWA 2.0): (1) settlement depth and capital liquidity, (2) verifiability and auditability, and (3) sustainable returns—that is, a reliable on-chain money market interest rate.
- Axis is building a verifiable, multi-strategy on-chain delta-neutral (market-neutral) arbitrage engine, encapsulating execution capabilities previously exclusive to institutions into composable yield primitives. Unlike products that rely on a single trade (such as a basis), Axis dynamically allocates resources across a basket of market structure opportunities: funding/basis arbitrage, cross-exchange spot spreads, CEX–DEX spreads, and cross-currency/regional premiums, making yield sources more diversified and more resilient as AUM expands. Its first product, USDx (and its staking counterparty USDx), packages this engine as a USD-pegged asset and provides machine-readable transparency (high-frequency PoR/NAV disclosure, segregated custody/segregated proof-of-stake, and independent third-party verification), thereby reducing the discount on black-box yields and counterparty risk premiums. In the longer term, the same execution infrastructure will be extended to scenarios such as OTC/RFQ execution, quotation/pricing API, and liquidity supply, making Axis the on-chain arbitrage infrastructure layer for USD (and will later be extended to BTC and gold-related products).
- Accountable is building a privacy-preserving verification infrastructure, translating institutional trust into machine-readable engineering primitives. Its Data Verification Network (DVN) can be deployed in a client's own environment, connecting data sources such as exchanges, wallets, custodians, and banks, and outputting cryptographically verifiable proofs. This allows counterparties to verify key states such as assets, liabilities, and asset restrictions/staking without exposing original positions or relinquishing sensitive account permissions, supported by cryptographic capabilities such as zkTLS and secure computation. This is equivalent to upgrading PoR from periodic reporting to continuous, insurable verification: stronger audit granularity, lower information asymmetry, and ultimately, lower counterparty risk premium. DVN already provides real-time PoR/NAV workflow support for legitimate issuers in production environments. In the long term, this verification framework will extend to trusted distribution surfaces such as Vault-as-a-Service, making verification no longer just a compliance cost, but a key advantage in improving capital efficiency and product go-to-market.
II. The Deep Integration of AI with the Crypto World
The AI wave is sweeping the globe, and as the most watched technological wave today, its every move has a profound impact on all aspects of society. What sparks will AI ignite in the field of encryption in the future? We believe that AI has great potential in areas such as proxy services and automated payments.
AI Agent Economy and M2M Payment Network
In multi-agent collaborative networks, different agents (such as data analysts, trade executors, and risk control officers) need to interact frequently. Blockchain smart contracts provide a permissionless trust foundation and payment pathway for this machine-to-machine (M2M) collaboration. This is mainly reflected in the following three aspects: The AI Payment sector has entered an early stage of explosive growth. Four major players—Google AP2, OpenAI × Stripe ACP, Visa Agentic Commerce, and x402—are simultaneously deploying agent payment infrastructure. Google launched the AP2 protocol to standardize agent payment interfaces, and Stripe ACP (Agentic Checkout Protocol) processes over 2 million API calls daily.
Visa's Agentic Commerce pilot program shows that AI agents autonomously complete e-commerce payments with a success rate of 98.5%, far exceeding traditional automated scripts. M2M payments are also poised for rapid growth. VanEck predicts that with the widespread adoption of Web3 native agent payment protocols such as x402, on-chain automated transactions driven by AI agents will reach $5 billion per day by 2027, with a projected compound annual growth rate (CAGR) exceeding 120%. Service call costs will be significantly reduced.
By leveraging blockchain micropayments to invoke agent services on demand, the cost of service invocation is reduced by 60% compared to the traditional API subscription model (SaaS) of the Web2 era. The cost per interaction can be as low as $0.0001, which greatly reduces economic friction and losses in the process of multi-agent collaboration.
Once Agent A completes a specific task, Agent B can make a USDC micropayment in milliseconds via the Lightning Network or Layer 2 protocol, without any human intervention, thus establishing an automated value transfer system.
AI and Verifiable Data Layer
As artificial intelligence evolves, "world models" such as JEPA and Sora, proposed by deep learning pioneer Yann LeCun, are replacing simple LLMs. Their core requirement has shifted from text generation to the accurate simulation of causality in the physical world. AI needs more realistic and reliable data from the real world. Gartner predicts that by 2026, 75% of global AI training data will consist of synthetic data. The lack of a data loop with real physical feedback will greatly increase the probability of "model collapse."
According to Messari's market analysis, verifiable real-world datasets, due to their scarcity and authenticity, are typically valued at 15 to 20 times the market value of ordinary web crawler data. High-fidelity training of world models heavily relies on high-dimensional physical data containing 3D space, depth information, and motion trajectories. Blockchain, through its cryptographic signature technology, provides immutable on-chain proof for every data point collected by sensors, fundamentally solving the problems of "data pollution" and "synthetic fabrication" commonly found in AI training, and building a trusted bridge between the physical world and digital models.
As of the third quarter of 2025, the number of active edge sensor nodes on the blockchain network has exceeded 4.5 million. These nodes provide approximately 20 PB of verifiable physical data for various world models every day, becoming the cornerstone supporting the next generation of AI cognition.
zkML and Decentralized Edge Computing: Trusted Inference on the Edge with Privacy Protection
The performance leaps of high-performance small parameter models (SLMs) such as Llama 3-8B and Phi-3 indicate a paradigm shift in AI inference computing power, moving from centralized cloud to edge devices (mobile phones, PCs, IoT devices). Market data shows that decentralized edge inference networks built using idle consumer devices (such as io.net or Akash) have a unit computing power cost of approximately $1.49/hour at the H100 level, which is 60% to 75% lower than AWS or Nvidia cloud inference services ($4.00 - $6.50/hour), presenting a significant economic arbitrage opportunity.
Thanks to the technological advancements of projects such as Accountable and Modulus Labs, the demand for zkML verification services in on-chain prediction markets, insurance protocols, and high-value asset management saw a 230% quarter-on-quarter increase in Q3 2025, indicating that high-value DeFi scenarios have developed a rigid demand for "trustworthy reasoning".
To address the risks of data forgery or model tampering caused by untrusted edge devices, zkML (zero-knowledge machine learning) has become a crucial trust infrastructure. Emerging protocols such as Accountable are building standardized verification layers that allow nodes to generate mathematical proofs. These proofs rigorously verify on-chain, without disclosing input data (such as medical images or financial private keys), that "the reasoning result was indeed generated by the correct operation of a specific model on the edge device," thus achieving a "trustless" closed loop for decentralized computing power.
OKX Ventures' key focus projects
- Aspecta: In essence, it builds "digital passports" for multi-agent systems. In the future, AI will collaborate and transact with each other like humans. But the question is, how can one AI trust another unfamiliar AI? What Aspecta does is give each AI a "digital passport." By analyzing its past behavior and code records, it assigns it a credit score. In this way, AIs can establish trust and even achieve unsecured lending. In the economic network where agents call services from each other, Aspecta generates a verifiable credit score for each agent by parsing the on-chain interaction graph and GitHub code contributions. This is a prerequisite for realizing M2M unsecured lending and trusted collaboration.
- LAB: It is an AI intent compiler in the Web3 field, utilizing the latest multimodal understanding capabilities to transform ambiguous human natural language (such as "arbitrage with the lowest risk") into structured on-chain execution instructions. It solves the "last mile" problem of connecting AI technology with complex DeFi protocols, significantly lowering the barrier for non-technical users to use advanced DeFi strategies.
- Hyperion serves as the physical anchor for AI world models, providing verifiable physical world data. It leverages a decentralized map network to collect geospatial data and combines it with AI inference to provide on-chain agents with zero-knowledge proof-verified "location services." This is crucial for RWA asset management and embodied intelligence (bot) scheduling that rely on real physical states.
III. Institutional Strategies: Macro Hedging, Privacy Infrastructure, and Smart Compliance
In the eyes of institutions, crypto assets have transformed from "speculative assets" to "global macro hedging tools." A very significant change is that in the previous cycle, retail investors might have ignored macroeconomic events when trading cryptocurrencies, but in this cycle, neglecting data such as the Federal Reserve, US-China tariffs, and CPI can easily lead to a passive position. Compliance is no longer an obstacle, but rather a moat for institutions. The issuance of digital banking licenses will allow for seamless exchange between crypto assets and fiat currency. Currently, three innovative products have emerged in the market.
First, with basis arbitrage and volatility products, institutions are no longer content with passively holding positions. Open interest in CME Bitcoin futures has repeatedly hit new highs, and institutional long positions have increased significantly.
Second, basis trading, which uses the price difference between spot ETFs and futures contracts to conduct risk-free arbitrage, has become a mainstream strategy for hedge funds, with an annualized return that is stable in the range of 8%-12%, far exceeding the yield of US Treasury bonds.
Thirdly, there are structured notes, which package "BTC spot + Ethereum staking yield" to provide institutions with an allocation option similar to "dividends + appreciation" in traditional finance, without having to deal with complex DeFi interactions.
Institutional portfolios have expanded from a single BTC allocation (as digital gold) to structured combinations of "BTC + ETH/SOL + DeFi blue chips"—where BTC acts as a store of value, and the staking yield of the POS chain is increasingly being regarded as a risk-free benchmark interest rate in the digital economy.
Privacy renaissance: A rigid demand for institutional entry
The Privacy Revival: The Rigid Demand for Institutional Entry With the deep involvement of TradFi giants in the crypto market in 2026, the double-edged sword effect of transparent ledgers began to emerge. In a completely public on-chain environment, it is difficult for institutions to execute complex arbitrage strategies or complete large transactions, because the exposure of any trading intentions could lead to serious risks of front-running and strategy leaks. This structural contradiction makes "privacy" an unavoidable prerequisite for institutional funds to be on-chain.
Against this backdrop, the meaning of privacy is being redefined. It is no longer seen as a means of evading regulation, but rather as a tool for protecting trade secrets through compliance—neither weakening regulatory capabilities nor sacrificing the institution's own strategic security. Specifically, institutions are shifting their investment and technological focus towards "programmable privacy." Privacy-preserving computing protocols based on zero-knowledge proofs (ZK) and trusted execution environments (TEEs) allow institutions to demonstrate the solvency and compliance of their assets without exposing their trading strategies and position details, thus achieving a balance between transparency and confidentiality.
Meanwhile, "compliant privacy pools," similar to the dark pool trading mechanisms in traditional financial markets, are forming on-chain. These liquidity pools hide transaction details from the public but grant viewing access to regulatory nodes, enabling large-scale institutional funds to complete low-impact, high-efficiency transaction execution in DeFi. They are seen as a "last mile" solution for institutional capital to enter the on-chain financial system.
Privacy is not a denial of the transparency spirit of blockchain, but rather an upgrade for the institutional era. Future on-chain finance will no longer solely pursue "everyone seeing everything," but rather, under the premise of compliance, properly conceal what shouldn't be seen. This privacy capability is transforming from a peripheral need into the infrastructure that truly enables institutional funds to be on-chain.
The rise of the on-chain compliance track
In the future, AI agents will take over on-chain interactions on a large scale, posing a risk of collapse to traditional financial compliance systems. Multiple institutions predict that by 2026, the daily number of on-chain interactions will have grown exponentially, with over 45% of transactions initiated by non-human entities. Faced with tens of thousands of high-frequency machine transactions per second, the traditional KYC/AML (Know Your Customer/Anti-Money Laundering) vetting model, which relies on sheer manpower, may become completely ineffective. Institutions will be unable to hire enough compliance officers to handle this throughput.
Therefore, the focus of compliance is shifting from "post-event accountability" to "code-level blocking." The next generation of compliance architecture requires embedding regulatory rules into smart contracts to achieve millisecond-level automated risk control. This is not only a regulatory requirement but also a prerequisite for the safe entry of institutional funds into DeFi. CipherOwl introduces an AI-driven on-chain audit and compliance layer, focusing on on-chain forensics and transaction tracking. It utilizes AI-assisted analytics tools to identify money laundering risks and illicit fund flows, providing institutional investors and regulators with necessary security barriers and due diligence tools. Its SR3 technology stack, through screening, reasoning, reporting, and research, uses Large Language Models (LLM) to parse complex on-chain transaction graphs, automatically identifying money laundering risks or sanctioned entities.
Furthermore, CipherOwl provides an API that allows trading savvy agents on Hyperion to query the compliance score of counterparty addresses milliseconds before executing a trade. If the risk is too high, the smart contract will automatically refuse to interact. This makes regulatory rules no longer an ex-post punishment, but a code constraint embedded in the trading process. For Wall Street institutions hoping to enter DeFi by 2026, CipherOwl is an indispensable compliance middleware.
IV. DeFi Proactive Intelligent Services and Prediction Markets
The open finance revolution that erupted in 2020 has shaken the blockchain industry, with its elegant AMM and permissionless features painting a picture of the future potential of Crypto finance.
DeFi 3.0 Proactive Intelligent Services: Intent-Based Kinetic Finance
We believe that DeFi is undergoing a leap from DeFi 1.0 (passive smart contracts) to DeFi 3.0 (active smart services). If the core of the 2020 DeFi Summer was the "democratization of asset issuance," then the transformation in 2026 will be dominated by the "active circulation of funds." The participation logic of institutional funds is evolving from simple RWA to "strategic on-chaining," using customized institutional-grade agents to execute 24/7 programmatic market making and risk control.
The market is abandoning the old "prescribed path" model. Data shows that monthly trading volume of CoW Swap based on the "Solver" model has routinely exceeded $3 billion, demonstrating the overwhelming advantage of intent-centric approaches in liquidity transfer. Investment logic is shifting from general-purpose DeFAI terminals to autonomous agents in vertical scenarios.
Compared to general-purpose chat UIs facing implementation bottlenecks, vertical agents specializing in yield optimization and liquidity management possess complete closed-loop execution capabilities and verifiable cash flow, making them key to controlling the underlying pricing power of the intelligent agent economy. In terms of investment, we believe the trading paradigm is shifting from "human-to-machine (H2M)" to "machine-to-machine (M2M)." Given that large AI models (LLMs) cannot directly parse complex Solidity bytecode, the market urgently needs to build a DeFi Adapter Layer. By introducing standards such as MCP (Model Context Protocol), heterogeneous protocols can be encapsulated into standardized, semantic "toolkits," enabling AI to access financial services like calling an API.
Under this framework, assets evolve into "smart packages" that can automatically seek returns, and the core metric shifts from TVL (Total Value Locked) to TVV (Total Value Velocity).
Prediction Market: Global Information Infrastructure in 2026
We believe that in an era of information overload with high signal-to-noise ratios, prediction markets are not only gaming platforms but also high-resolution, high-timeliness "truth oracles." In October 2025, the compliant platform Kalshi, with its CLOB architecture, surpassed Polymarket with a 60% market share and a weekly trading volume of $850 million, and the market's open interest (OI) rebounded to $500-600 million, marking the entry of long-term non-speculative capital.
Investments should focus on projects that can address capital utilization issues at the protocol level: Polymarket's NegRisk mechanism improves capital efficiency in multi-outcome markets by 29 times by automatically converting "NO" shares into mutually exclusive "YES" combinations, contributing 73% of the platform's arbitrage profits; Kalshi's "collateral return" frees up capital tied up in hedging positions. Whoever can turn money around faster can capture liquidity. Polymarket's strategy of seizing liquidity with extremely low fees of 0-0.01% is essentially building a data factory, ultimately selling "sentiment indicators" to institutions through ICE's (NYSE's parent company) $2 billion investment intention and channels. This data narrative supports its $12 billion valuation.
In contrast, Kalshi maintains a high fee rate of approximately 1.2% by leveraging its compliance moat and employs an "embedded" expansion strategy, achieving 400,000 monthly active users through integration with Robinhood. Myriad, on the other hand, captured 30,000 active trading users by embedding itself in Decrypt's media stream. This demonstrates that the embedded model has a lower customer acquisition cost than a standalone app. The legal battle over regulatory jurisdiction is the biggest variable in this sector, with the core conflict revolving around whether the prediction market falls under the jurisdiction of the CFTC as a "product" or under the jurisdiction of individual states as "gambling."
Kalshi opted for a "federal-first" approach, holding a DCM license from the CFTC in an attempt to leverage the exclusive jurisdiction of federal law to override state laws. However, this also exposed it to litigation challenges from at least eight state gaming commissions. While this compliance status granted it significant pricing power (such as a 1.2% fee premium), it also came with substantial legal costs.
On the other hand, Polymarket chose the "offshore circumvention" route, using DeFi architecture and geographic blocking to circumvent US jurisdiction, but constantly faces the threat of SEC's long-arm jurisdiction and ISP bans from EU countries. OKX Ventures believes that future investment opportunities will focus on the following three areas:
- Long on middleware: Pay attention to underlying middleware (such as Azuro) and dedicated oracles (such as Pyth, EigenLayer AVS). They are not restricted by a single regulatory jurisdiction and can capture the value generated by all front-end applications, making them the "infrastructure bet" with the highest risk-reward ratio.
- Find embedded traffic entry points: Customer acquisition costs for standalone prediction market apps are extremely high. Focus on projects that develop Telegram bots or modularly embed prediction markets into news/social platforms*. These projects can reach users with zero friction and have a stronger potential for viral growth.
- Arbitrage opportunities in vertical sectors: Avoid the general political/macro markets that have formed duopolies, and look for leading companies in the sports and high-frequency crypto asset vertical sectors. The sports sector has a huge product gap due to the complexity of parlay functionality, while crypto high-frequency prediction is a necessity for DeFi traders. Neither of these areas has yet emerged an absolute dominant player.
V. Summary
Looking ahead to 2026, we anticipate the industry's focus will shift from "supplying network capacity" to "releasing asset efficiency." We will no longer solely focus on ledger storage and verification capabilities, but rather on the speed, intelligence, and settlement efficiency of on-chain capital flows. We define this as the era of "kinetic finance," a macro-level shift from "asset on-chain" to "economy on-chain." Under this new paradigm, traditional financial boundaries are blurring.
OKX Ventures believes that 2026 will be a critical turning point for the crypto industry as it sheds its speculative bubble and returns to value creation. Projects that can solve real-world trust costs and circulation efficiency through code will become the cornerstone of the new era.
We are firmly optimistic about this transformation and will continue to invest in cornerstone projects that can reduce trust costs and improve capital efficiency through code. At the singularity of deep integration between digital and reality, whoever can define the rate of asset flow and the boundaries of truth will hold the pricing power of the new era.
