Source: Fintech Blueprint, compiled by: BitpushNews
Structural problems in the crypto market
The adoption of on-chain financial instruments and the trend of machine economy are booming.
In the past year, we have seen a huge expansion of blockchain-native finance in the following five dimensions: (1) stablecoins, (2) decentralized lending and trading, (3) perpetual contracts, (4) prediction markets and (5) digital asset vaults (DATs). The regulatory environment in the United States has become extremely favorable, which has led to an increase in both the number of projects and risk appetite.
Leaving aside the uncertainties brought about by taxation and market structure, a tolerant macroeconomic environment has also provided fertile ground for crypto innovation to take root. These trends are well-known and need no further elaboration with data. However, 2025 will be an extremely difficult year for long-term investors in tokens other than Bitcoin and crypto assets. If you're a trader or banker, things might be going well—we've seen record commissions pushing DATs to market, and exchanges like Binance raking in huge fees during the listing process. But for those of us with 3-5 years of investment experience, the market structure has been terrible. We're caught in a negative prisoner's dilemma: token holders anticipate future selling pressure and therefore sell any and all assets; while market makers and exchanges that underpin the crypto economy take short-term speculative positions. Token unlocking mechanisms and issuance prices often cripple projects before they even become profitable or find a market fit. Furthermore, the structural market failure on October 10th this year clearly dealt a heavy blow to several major market players, and although the losses have not yet been disclosed, the aftershocks of the liquidation continue. The correlation among all crypto assets has risen to near 1, indicating industry-wide deleveraging by participants, despite their vastly different fundamental logics. It's easy to retreat and become cynical at this moment. However, we prefer to conduct "Market-to-market" comparisons as clearly as possible to plan for future investments. The 2025 decline in crypto investment is information, but not a definitive conclusion. It is very likely that we will see a large-scale liquidation in the private secondary market in 2026, at which time we will analyze how so many special purpose vehicles (SPVs) were issued at high valuations during the crypto boom. Meanwhile, the vision of programmable finance and "robot money" continues to materialize, and we must continue to strive to find the best position for their inevitable rise. For background, see the image below. This image zooms in over the past decade, showing the market capitalization creation in several regions and industries. When we look at this history, the value creation in this field is astonishing compared to the rest of the world. The European capital markets (approximately $2-3 trillion across various countries) have made virtually no progress, merely maintaining the status quo. You'd be better off investing in government bonds, earning 3% interest annually, which would likely create more value. On the right side of the chart, India and China show a compound annual growth rate (CAGR) of 5-10%, with net market capitalization increases of approximately $3 trillion and $5 trillion respectively during the same period. Having understood this scale, let's look at our definition of "robot currency": (1) The "Magnificent 7" representing technology and AI has increased its market capitalization by approximately $17 trillion at a rate of 20% per year; (2) The crypto asset market, representing the modern financial track, has increased by $3 trillion during the same period, with a compound annual growth rate of 70%. This is the financial center of the future. But being logically correct is not enough. We must delve deeply and meticulously into the parts of the value chain that have not yet been noticed by the world. Recall the discussions about robo-advisors in 2009, Neobanks in 2011, or DeFi in 2017; the vocabulary and connections were still nascent, and it wasn't until 2-5 years later that these insights solidified into clear business opportunities. As a kind of "self-torture" exercise, we've compiled a 158-page summary report covering the most relevant players in the machine economy by 2025. In the open market, 2025 will be a year of "the strong getting stronger and the weak falling behind". The clear winners are the owners of physical and financial bottlenecks: electricity, semiconductors, and scarcity. Bloom Energy, IREN, Micron, TSMC, and NVIDIA have all significantly outperformed the market as capital chases assets that are “must be passed through by machines.” Bloom and IREN are typical examples: they directly capitalize on the AI capital expenditure boom, translating urgency into revenue. In contrast, traditional infrastructure like Equinix has underperformed, reflecting the market's perception that general-purpose capacity is far less valuable than reliable power supply and high-density, customized computing power. In the software and data sector, performance diverged along another dimension: (1) mandatory versus (2) optional. Platform-like enterprise systems with embedded workflows and mandatory renewals (such as Alphabet and Meta) continued their compound growth, both rising year-to-date as AI spending strengthened their existing distribution moats. ServiceNow and Datadog, despite their strong product capabilities, suffered from valuation pressures, bundling pressures from hyperscale cloud providers, and slower AI monetization. Elastic illustrates the downside: strong technical capabilities, but squeezed by cloud-native alternatives, and deteriorating unit economic returns. The private equity market also exhibits similar screening mechanisms. Foundational model companies are the protagonists of the story, but their vulnerability is increasing. OpenAI and Anthropic are experiencing rapid revenue growth, but their neutrality, capital intensity, and margin compression are now clear risks. Scale AI serves as a cautionary tale this year: Meta's partial acquisition destroyed its "neutral" position and triggered customer churn, demonstrating how quickly service-heavy business models can crumble once trust is broken. In contrast, companies controlling value (Applied Intuition, Anduril, Samsara, and emerging fleet operating systems) appear better positioned, even though value realization remains largely private. Networks were the weakest performing sector. With very few exceptions, decentralized data, storage, agents, and automation protocols underperformed as usage failed to translate into token value capture. Chainlink remains strategically important but struggles to align protocol revenue with its token economic model; Bittensor is the biggest bet in crypto AI but doesn't yet pose a substantial threat to Web2 Labs companies; Giza and similar agent protocols have shown real activity but remain hampered by dilution and meager fees. The market no longer rewards "collaboration" without mandatory fee mechanisms. Value is accumulating in areas where machines are already paying for it—electricity, silicon, computing contracts, cloud bills, and regulated balance sheets—not in areas they might choose someday. In 2025, the market rewarded ownership of "choke points" and punished projects with grand ideals but lacking control over cash flow or computing power. The key to the future lies in identifying where economic power already exists and betting on assets that machines cannot bypass. Core takeaway: The realization of AI value is "one step deeper" than most people realize. Neutrality is now a first-class economic asset (see Scale AI). A "platform" is only effective when combined with a control point, not just a function. AI software is deflationary (pricing pressure); AI infrastructure is inflationary. Vertical integration is only important when it locks in data or economic effects. Token networks are repeatedly undergoing the same market structure tests. Simply having exposure to AI isn't enough; positioning quality is everything. Robotics hardware and software will be the next hype cycle, and we may see similar investment waves and selective winners. Positioning in 2026. Over the past two years, we've built a core portfolio covering the key themes discussed here. Looking ahead to 2026, our positioning and investment execution will be further strengthened. Next, let's discuss our holding strategy. While the long-term vision for autonomous intelligent agents, robotics, and machine-native finance is on the right track, the market is currently experiencing extremely outrageous valuations in the private AI and robotics sectors. Aggressive secondary liquidity and implied valuations exceeding $100 billion mark a transition from the "discovery phase" to the "exit phase." As an early-stage fund with a Fintech perspective, we must target these downstream expenditures: Machine Transaction Surfaces: Machines or their operators already host layers of economic activity, such as payments, billing, metering, routing, and the orchestration, compliance, custody, and settlement primitives of capital or computing power. Returns are derived from transaction volume, acquisitions, or regulatory status, not speculative narratives. Walapay and Nevermined in our portfolio are examples. Applied Infrastructure With Budgets: Infrastructure that enterprises or platforms are already procuring, such as computing power aggregation and optimization, data services embedded in workflows, and tools with recurring expenditures and switching costs. The focus is on ownership of the budget and the depth of integration. Examples include Yotta Labs and Exabits. High-novelty opportunities: A few opportunities with asymmetric growth but uncertain timing: basic research, cutting-edge science, and AI-related cultural or IP platforms. Our recent investment in Netholabs (a lab dedicated to extrapolating the complete digital brain of mice) fits this description. Furthermore, we will be more actively investing in equity until the token market structure issues are resolved. Previously, our exposure was 40% tokens and 40% equity, with the remaining 20% allocated flexibly. We believe the token space needs 12-24 months to digest the current predicament. Key takeaways: You don't need to be a venture capitalist to learn from and benefit from these market dynamics. Massive capital expenditures are flowing from tech giants to energy and component suppliers. A handful of companies are expected to be trillion-dollar winners in the public markets, but they are choosing to remain private and divest their special purpose vehicles (SPVs). Publicly traded companies are struggling to defend themselves. Political power is centralizing and nationalizing these initiatives—whether it's Musk and Trump, or China and DeepSeek—rather than supporting their decentralized alternatives in Web3. Robots are intertwined with national manufacturing and war industries. In creative industries (from games to movies and music), there is resistance to AI, with those engaged in "human craftsmanship" rejecting robots that pretend to do the same. In software, science, and mathematics, however, AI is seen as a great achievement that can help discover and build efficient business architectures. We need to stop believing in this collective illusion and return to reality. On the one hand, dozens of companies have achieved annual revenues exceeding $100 million by serving users; on the other hand, the market is also rife with falsehoods and scams. These two phenomena coexist and are not mutually exclusive. The new year will bring a complete reshuffle, but it also contains tremendous opportunities. Only by cautiously navigating the tightrope of opportunity can we achieve success. Let's meet again on the other side!
Anais
Weatherly