Author: Jiawei @0xjiawei
In the mid-to-late 1990s, investment in the Internet centered around infrastructure. The capital markets were betting almost exclusively on fiber optic networks, ISPs, CDNs, and server and router manufacturers. Cisco’s stock soared to a market value of more than $500 billion by 2000, making it one of the most valuable companies in the world, and fiber-optic equipment makers such as Nortel and Lucent became hot, attracting tens of billions of dollars in financing.
As a result of this boom, the U.S. added millions of kilometers of new fiber-optic cable between 1996 and 2001, building far more than was actually needed at the time. The result was significant overcapacity around 2000 — transcontinental bandwidth prices fell by more than 90 percent in just a few years, and the marginal cost of Internet access approached zero.
Although this round of infrastructure boom so that the later birth of Google and Facebook can take root in the cheap, ubiquitous network, but for the fanatical investors at the time, it also brought pain: infrastructure valuation bubble quickly burst, Cisco and other star companies in a few years more than seventy percent of the market value of the shrinkage.
Doesn’t that sound a lot like Crypto in the last two years?
Is the era of infrastructure perhaps over for now?
Block space goes from scarcity to proliferation
The expansion of block space and the exploration of blockchain’s “impossible triangles” have largely dominated the early years of crypto for several years now, so it’s appropriate to talk about them as a landmark element.
In the early stages, public chains had extremely limited throughput, and block space was a scarce resource. During DeFi Summer, for example, DEX interactions often cost $20-$50 per transaction with all types of on-chain activity overlapping, and hundreds of dollars in extreme congestion. During the NFT period, the demand and clamor for capacity expansion was at its peak.
Etherrum’s composability is a big advantage, but overall it increases the complexity and gas consumption of a single call, and limited block capacity is prioritized for high-value transactions. As investors, we often talk about L1’s fee and burn mechanism and use this as an anchor for L1 valuation. During this time, the market has priced infrastructure very high, and the so-called “fat protocols, thin applications” argument that infrastructure captures most of the value has been accepted, leading to a series of capacity expansion programs and even a bubble.
As a result, key Ethereum upgrades (e.g., EIP-4844) migrated L2 data availability from expensive calldata to lower-cost blobs, resulting in a significant reduction in the unit cost of L2. Transaction costs for mainstream L2 are generally down to the order of a few cents. The introduction of modularity and Rollup-as-a-Service solutions has also significantly reduced the marginal cost of block space. Various Alt-L1s supporting different virtual machines have also emerged. The result is that block space has gone from being a single scarce asset to a highly fungible commodity.
The chart above shows the evolution of the cost of various types of L2 uplinks over the last few years. As you can see, in 23 to early 24, Calldata dominated the costs, even approaching $4 million per day. Then in mid-24, the introduction of EIP-4844 allowed blobs to gradually replace Calldata as the dominant cost, and the overall on-chain cost dropped significantly. As we move into ’25, the overall overhead tends to be lower.
As a result, more and more applications can put their core logic directly on the chain, rather than using the complex architecture of off-chain processing followed by on-chain processing.
From this point on, we see value capture begin to migrate from the underlying infrastructure to the application and distribution layers that can directly take traffic, drive conversions, and close the cash flow loop.
Revenue Level Evolution
We can visually validate this view at the revenue level. In a cycle dominated by infrastructure narratives, the market’s valuation of L1/L2 protocols is based on expectations of technical strength, ecosystem potential, and network effects, the so-called “protocol premium”. Token value capture models tend to be indirect (e.g., through staking, governance, and vague expectations of fees).
Application value capture is more direct: verifiable on-chain revenues are generated through fees, subscriptions, service charges, etc. These revenues can be used directly for token recovery. These revenues can be used directly for token buyback destruction, dividends, or reinvested in growth, creating a tight feedback closure loop. The app’s revenue stream becomes solid — more from actual service rate revenue than token incentives or market narratives.
The chart above provides a rough comparison of protocol (red) and app (green) revenues from 2020 to date. We can see that the value captured by the application is gradually increasing and reaching about 80% this year. The table below lists the 30-day protocol revenue rankings from TokenTerminal’s statistics, with L1/L2 accounting for only 20% of the 20 projects. Particularly prominent are applications such as stablecoins, DeFi, wallets and trading tools.
In addition, the correlation between the price performance of app tokens and their revenue figures is gradually increasing due to the market reaction brought about by buybacks.
Hyperliquid’s daily buyback size of around $4 million has provided significant support to the token price. Buybacks are considered to be one of the key factors driving the price rally. This suggests that the market is beginning to correlate agreement revenue and buyback behavior directly to token value, rather than relying solely on sentiment or narrative. And I expect this trend to continue to strengthen.
Embracing a new cycle dominated by apps
A Golden Age for Asian Developers
Electric Capital’s 2024 Developer Report shows that for the first time, the share of blockchain developers in Asia has reached 32%, surpassing North America as the world’s largest concentration of developers.
Over the past decade, globalized products such as TikTok, Temu, DeepSeek, etc. have proven the outstanding capabilities of Chinese teams in engineering, product, growth and operations. Asian teams, especially Chinese teams, have a strong iterative rhythm, the ability to quickly validate requirements, and localize and expand through localization and growth strategies, and crypto fits these characteristics: the need to quickly iterate and adapt to market trends, and the need to serve a global user base, cross-language communities, and multi-market regulation.
As a result, Asian developers, especially Chinese teams, have a structural advantage in the crypto adoption cycle: they have strong engineering capabilities, as well as sensitivity to speculative market cycles and strong execution.
Against this backdrop, Asian developers have a natural advantage in being able to deliver globally competitive Crypto applications faster. This cycle we have seen Rabby Wallet, gmgn.ai, Pendle and others represent Asian teams on the global stage.
We expect to see this shift soon: a shift from the US narrative to a new path of product launch in Asia, followed by expansion into Europe and the US. Asian teams and markets will have more say in the application cycle.
Primary market investment under the application cycle
I would like to share some ideas on primary market investment:
- Trading, asset issuance and financialization applications still have the best PMFs and are almost the only products that can traverse bulls and bears. The counterparts are perp such as Hyperliquid, Launchpad such as Pump.fun, and products like Ethena. The latter packages money rate arbitrage into a product that can be understood and used by a broader user base.
- For investments in niche tracks where there is a high degree of uncertainty, consider investing in the beta of the track, thinking about what projects will benefit from the growth of the track. A typical example is the prediction market — there are about 97 public prediction market projects on the market, Polymarket and Kalshi are the more obvious winners, at this point the probability of betting on the middle and long-tail projects to bend the curve is very low. Investing in prediction market tool-based projects such as aggregators, analysis tools, etc., on the other hand, is much more certain to reap the dividends of the track’s growth and go from doing a difficult multiple choice question to a single choice question.
- Once the product is in place, the next step centers on how to get these apps to the masses. In addition to common entrances such as social login provided by Privy and others, I believe that the front-end and mobile end of aggregated transactions are also quite important.
Under the application cycle, no matter perp or prediction market, mobile will be the most natural reach scenario for users, whether it is the first deposit or daily high-frequency operation, the experience will be smoother on the mobile terminal. The value of the aggregation front-end lies in the distribution of traffic. The distribution channel directly determines the conversion efficiency of users and the cash flow of the project.
Wallet is also an important part of this logic.
I believe that the wallet is no longer a simple asset management tool, but has a positioning similar to the Web2 browser. The wallet directly captures the order flow and distributes it to block builders and searchers, thus realizing the traffic; at the same time, the wallet is also a distribution channel, through the built-in cross-chain bridge, built-in DEX, access to third-party services such as staking and so on, it becomes a direct entry point for users to contact other applications. In this sense, the wallet holds the order flow and traffic distribution right, and is the first entrance of user relationship.
- For the infrastructure under the whole cycle, I think some public chains created out of thin air have lost their significance of existence; while the infrastructure that does basic services around applications can still capture value. A few specific points are listed below:
- Infrastructure that provides customized multi-chain deployment and application chain building for applications, such as VOID;
- Companies that provide user onboarding (covering login, wallet, deposit/withdrawal, deposit/withdrawal, etc.) services, e.g., Privy, fun.xyz; this can also cover the wallet and payment layer (fiat-on/off ramps, SDK, MPC, etc.)
- Cross-chain bridges: As the multi-chain world becomes a reality, the influx of application traffic will create an urgent need for secure and compliant cross-chain bridges.
Application Cycle: A Golden Age for Asian Developers was originally published in IOSG Ventures on Medium, where people are continuing the conversation by highlighting and responding to this story.
