ArkStream Capital: From AI Synergies to RWA Emergence, Fund Migration in the 2026 Crypto Market

Original Title: "ArkStream Capital: From AI Siphoning to RWA Emergence | The Capital Migration of the 2026 Crypto Market"
Original Source: ArkStream

TL;DR

The following analysis and insights are based on market data as of the end of Q2 2026 and the market conditions at that time.

In the first half of 2026, the weakness of the Crypto market did not stem from industry implosions but was driven by the Fed's hawkish policy shift, supply-side inflation triggered by the US-Iran conflict, and the continued siphoning of liquidity by AI stocks. Meanwhile, profits in the AI sector are transitioning focus from chip leaders, cloud providers, and software applications to physical supply chain bottlenecks like storage and PCBs; the three mega IPOs will further test the overvaluation and high burn rate model of the AI sector.

Against the backdrop of the overall DeFi contraction, RWAs have become one of the few areas in Crypto to experience net inflows: on-chain activity has grown from around $216 billion at the beginning of the year to $330 billion, with tokenized stocks and RWA perpetual contracts seeing the fastest growth. The demand for "coins" is waning, but there is a rapid increase in demand for trading traditional assets such as stocks and commodities through Crypto infrastructure. Our assessment is that the stock market's siphoning effect on Crypto may have peaked in June 2026; barring any major negative developments, the market may have entered a phase of slow recovery.

Macroeconomic Environment: Policy Shifts and Geopolitical Impacts

This section revolves around three main threads: the complete reversal of Fed expectations on monetary policy from "rate cuts" to "rate hikes" after the Fed Chair change, the ongoing disruption of the energy and inflation chain due to the US-Iran conflict, and the risk-off choices of global funds in the backdrop of US-China rivalry. These three factors collectively determined the pricing environment of all risk assets this quarter, explaining why the Crypto market weakened while global stock markets surged.

The macroeconomic environment in the first half of 2026 is defined as "asset pricing differentiation under a triple impact": the policy shift, ongoing geopolitical conflicts, and the extreme concentration of funds in the AI sector. These three factors have combined to create the largest performance differentiation among different assets under the same interest rates and liquidity conditions in this cycle. While the Dow Jones index hit a historic high at the end of the quarter and the South Korean KOSPI rose by over 100% in the first half of the year, Bitcoin saw a monthly decline of around 20.5% in June, with the Crypto Fear and Greed Index dropping to 15—such a degree of deviation is rare in the comparison of major asset classes over the past decade. Understanding the reason for this deviation is crucial to understanding the performance of the Crypto market this quarter.

New Fed Leadership: From “When to Cut Rates” to “Whether to Hike”

By the fourth quarter of 2025, the market had only experienced a slowdown in rate cuts; by the first half of 2026, the entire rate-cutting narrative was completely overturned. This reversal was not caused by a single meeting or a specific piece of data; it took a full two quarters to erode the early-year optimistic expectations bit by bit.

· Three-Month Expectation Reversal

1. In January, the market's mainstream pricing for 2026 was still “two to three rate cuts,” with CME rate futures at one point indicating that the first rate cut of the year would fall in the second quarter; at the April policy meeting, the dissenting votes reached a new high since 1992, with at least five officials publicly stating that the next policy direction might be a rate hike; in May, Kevin Warsh officially succeeded Powell as Fed Chair; in June, more than half of the officials on the dot plot moved to the rate hike camp. In six months, the market shifted from “two to three rate cuts” to “at least one rate hike this year.”

Kevin Warsh's attitude shift was particularly notable: at his confirmation hearing, he criticized the Fed for “fatal policy mistakes” made between 2021 and 2022, and had previously expressed willingness to lower rates, believing that high rates were harming economic vitality—based on this background, the market initially labeled him as dovish. However, the data did not give him that opportunity: in April, U.S. real wages fell by 0.5% month-on-month, the first time in nearly three years, and the impact of inflation on household purchasing power had already become apparent; May's price data showed a broad-based deterioration. His first decision was not “cut or not cut rates,” but “how to avoid getting caught in the rate hike debate.”

· June FOMC: Inflection Point in the Macroeconomic Narrative

The June 16th to 17th meeting was an inflection point in the macroeconomic narrative for the quarter. The resolution itself was uneventful: the federal funds rate remained unchanged at 3.50%–3.75% for the fourth consecutive time, unchanged since the last rate cut in December 2025. The real ripples were caused by three unconventional signals.

June 2026 FOMC Dot Plot Distribution (Fed SEP, ArkStream compilation)

In the simultaneously released Summary of Economic Projections (SEP), officials lowered the median forecast for 2026 real GDP growth to 2.2%, made a slight adjustment to the unemployment rate forecast to 4.3%, and raised short-term inflation expectations. With growth downgrades, inflation upswings, and rate hikes, all three arrows pointed to one word: stagflation concerns.

ArkStream Capital believes that the combination of these three signals is more important than the "will they or won't they hike rates" question itself. Over the past decade, the market has relied on dot plots, forward guidance, and chair press conferences to piece together a full suite of rate path calibration tools. The first thing Powell did when he took office was systematically dismantle the credibility of these tools. With the calibration tools gone, the volatility of rate expectations can only passively rise, leading to the turmoil in risk assets this quarter, especially high-beta assets (in previous quarterly reports, we classified crypto assets led by BTC under this category).

· The Dual Nature of Data: Inflation Stickiness and Employment Swings

What underpins the hawkish pivot is the substantial deterioration in price data. May's CPI rose by 4.2% year-on-year and 0.5% month-on-month, with energy being the major driver; PPI for final demand surged by 6.5% year-on-year and 1.1% month-on-month, with final demand goods prices increasing by 2.8% month-on-month, and the war-induced production cost pressures are still being transmitted downstream.

However, the structure of inflation is not uniform, which is crucial in determining the policy path. Core CPI rose by 2.9% year-on-year and 0.2% month-on-month, still manageable; research from the Dallas Fed shows a clear deviation between core PCE (around 3%) and trimmed mean PCE excluding extreme items (around 2.3%), indicating that the pressure in this round is highly concentrated in categories affected by supply shocks such as energy and transportation, and has not yet spread comprehensively to services and wages.

Distribution of U.S. Inflation Subcomponents in the first half of 2026 (BLS, Dallas Fed, compiled by ArkStream)

The catalyst for this round of inflation was the U.S.-Iran war and the closure of the Hormuz Strait.

On the employment front, a reversal drama played out during the quarter:

The report in May that was "so strong that it ruled out rate cuts" directly fueled rate hike expectations; by June, disrupted by the World Cup and holiday mismatches, market pricing fluctuated back and forth: after the June FOMC, CME data showed the probability of a rate hike in October briefly rose to 60.7%, with the money market fully pricing in a 25-basis-point hike in December; after the June nonfarm payroll report, coupled with Powell's softening statement at the ECB Forum that "in the past four weeks, inflation expectations have somewhat receded," rate hike pricing quickly retreated.

Market divergence is equally rare: Bank of America predicts a 25 basis point rate hike in September, October, and December, with the year-end interest rate range reaching 4.25%-4.50%; Huatai Securities assigns close to a 50% probability to a December rate hike; CICC and Citic maintain their stance that there will be no interest rate adjustments during the year. The same set of data can lead top institutions to three different conclusions, a rarity in the history of Fed watching over the past decade.

In our first-quarter report, we pointed out that "the rate inflection point itself faces the risk of being overturned, and a scenario of zero rate cuts or even rate hikes throughout the year needs to be taken seriously." This scenario became a reality in the second quarter and evolved beyond our initial estimates. Continuing the analysis of the correlation between interest rates and Bitcoin: the three rate cuts from the end of 2024 to 2025 constitute the "loose trading period," the fourth quarter of 2025 enters the "expectation digestion period," and the second quarter of 2026 officially enters the fifth stage — the rate hike expectation period. The most troublesome aspect of this stage lies in the two-way volatility: every swing in rate expectations is mirrored by a magnified fluctuation in high-beta assets, increasing the cost of both long and short positions, causing back-and-forth swings, and forcing capital to remain on the sidelines.

· Abandoning Forward Guidance: Why the Impact on the Crypto Market Is Particularly Severe

This point is worth discussing separately because its impact on different assets is asymmetrical.

For traditional assets, the disappearance of forward guidance mainly implies increased difficulty in duration management. Institutions can respond by shortening durations, increasing hedges, and raising cash levels; financial tools are readily available. However, the pricing of crypto assets heavily relies on the "direction of discount rate expectations" and lacks interest rate tools for hedging — there is no Bitcoin interest rate swap in the market, no crypto fixed-income products with duration matching, and institutions can only reduce their exposure. In other words, when macro visibility declines, the response for traditional assets is structural adjustments, while the response for crypto assets can only be position reduction.

This explains a counterintuitive phenomenon this quarter: the June FOMC did not actually raise rates, not even by a single basis point, yet the reaction in crypto assets far exceeded that in the U.S. stock market. Market trading is never about the current interest rate level but the confidence interval of future rate paths. When this interval is actively widened, the first assets to be sold off are always those at the edge of the risk budget table, lacking hedging tools. This also further validates our proposition in the first quarter that "the macro pricing power of crypto has not yet been regained": before the completion of the Fed's communication framework restructuring, the crypto market will find it hard to reclaim its pricing power.

US-Iran War: Uncontrollable Variables and Supply Shock-type Inflation

The U.S.-Iran military action that began on February 28 did not end in the second quarter; instead, it turned into a war of attrition.

For the financial markets, the transmission path of this war was singular but impactful: Strait of Hormuz blockade → Forced production cuts in Middle Eastern oil-producing countries → Surge in oil prices → Energy component of CPI spikes → Rate cut expectations fade, rate hike expectations rise → Global risk assets sell-off.

Every link in the chain has specific numbers that are clearly reflected in financial market pricing. On the supply side, oil production in southern Iraq plummeted from 4.3 million barrels per day to 1.3 million barrels per day, with Rumaila field alone cutting production by 700,000 barrels. Officials warned that if exports remained constrained, the production cut would escalate to over 3 million barrels within a few days. Kuwait Petroleum Corporation announced force majeure production cuts, Qatar Energy halted LNG production and transport, and Saudi Arabia's largest Ras Tanura refinery shut down following a drone attack. The details on the shipping side further illustrate the issue: multiple international insurance companies canceled war risk insurance for vessels in high-risk waters, some routes were forced to circumnavigate the Cape of Good Hope, adding an extra two weeks to the journey. Several very large crude carriers, each loaded with 6 million barrels of oil, were stranded in the Persian Gulf for over two months until around May 20 when they began slowly exiting the strait. Asian refineries had to find alternative solutions — India's MRPL issued force majeure notices, reducing production by 20% to 30%; Japan and Indonesia increased purchases of U.S. crude oil, and the U.S. Treasury granted Indian refineries a 30-day waiver from Russian oil sanctions.

2026 Key Oil Price Milestones in the First Half of the Year (CME Group, ArkStream compilation)

The efforts to hedge the supply gap have been ongoing: the IEA's 32 member countries announced a joint release of 400 million barrels from strategic reserves, with 164 million barrels already released as of May 8; OPEC+ has announced production increases for four consecutive months, with July's increment at 188,000 barrels per day. However, with the strait blocked, the production increases are almost futile — oil cannot be shipped out, rendering quotas meaningless. On the demand side, the high oil prices are starting to bite back. The IEA revised global oil demand growth from a pre-war rate of +1.2 million barrels per day to a decline of 420,000 barrels per day in 2026. South Korea even reinstated odd-even license plate restrictions for public vehicles.

A comparison with the 2022 Russia-Ukraine conflict highlights the uniqueness of this event. In 2022, after the oil price surge, the Fed's response was aggressive rate hikes, from near-zero to over 5%, with ample policy space and a clear direction. Although the market was in pain, at least there was clarity on the script. This time around, the situation is much more awkward: rates are already at 3.5%–3.75%, but inflation is resurging due to external shocks. The Fed lacks sufficient room for rate hikes and has lost the tailwind script of "inflation easing — rate cuts." What's more troublesome is that the 2022 supply shock mainly affected Europe's natural gas and grains, while the U.S.'s energy self-sufficiency provided a buffer. The Strait of Hormuz, however, impacts around one-fifth of global oil shipping volumes, leaving no room for buffer.

For the cryptocurrency market, the experience of the two rounds of impact is also different. In 2022, Bitcoin's decline was accompanied by a chain reaction within the industry (Terra, Three Arrows, FTX), making it difficult to distinguish between internal and external causes. This time, there was no equivalent credit event within the industry, nor were there enough heavyweight defaults to cleanse, making the decline almost purely the result of external pricing.

This round of inflation is being driven by a supply shock. Raising interest rates cannot stop the US-Iran ceasefire or open the Strait of Hormuz, but the Federal Reserve only has interest rate hikes in hand. The mismatch between policy tools and inflation triggers is highly likely to prolong the period of high interest rates, which is the most important judgment on the macro tone for the second half of the year.

The US-China Game and Global Funds' "Flight to Safety"

The Trump administration introduced a "secondary tariff" policy during the quarter, where tariffs and wars constituted a dual-source cost shock. The commonality between the two is that they both act on the supply side and cannot be resolved by monetary policy.

China remained neutral in the conflict, controlling the direct impact of the blocked strait through diversified energy procurement.

The US-China game is happening in the supply chain rather than in energy, and this layer is directly related to the AI theme in the second part of this report. In the first half of the year, export controls on advanced processes, semiconductor equipment, and key materials continued to tighten, while China accelerated its efforts in AI autonomy—DeepSeek V4 Pro achieved full-process training on Huawei's Ascend chip, and ZhuiPu GLM-5 became the first cutting-edge model trained entirely on domestically produced chips. The outcome of the game is not a win for one side, but a shift for the global AI supply chain from a "single optimal solution" to "two parallel systems," where both systems need to stockpile, expand production, and hoard key links in their capacity. This is precisely one of the deep underlying reasons for the historic shortage of storage in this round: the demand side is not only the growth of AI itself, but also the repetition and strategic inventory caused by geopolitical uncertainties.

In such an environment, the choice given by global funds is to "flight to safety"—avoiding assets with high valuations, distant cash flows, and sensitivity to liquidity, and embracing tangible bottlenecks. In the market, a new term has even emerged: HALO Trades (Heavy Assets, Low Obsolescence), which are long on the energy, raw materials, and infrastructure industries that have physical barriers and cannot be replaced by AI. The reason why the global AI market rally in this round was ultimately "squeezed dry" by storage and similar upstream sectors, rather than being taken by cloud providers and software applications, is due to the underlying logic being aligned with HALO Trades.

As for crypto assets, in the cognitive framework of most traditional investors, it stands at the other end of this asset class: no cash flow, no physical barrier, highly sensitive to interest rates. The ultimate result of the triple macro impact overlay is that crypto assets experienced a standalone decline amid the global rally of risk assets. Bitcoin saw a monthly decline of about 20.5% in June, ending the month at $60,760, just a stone's throw away from the 52-week low of $58,100, with a retracement of over half from the 52-week high of $125,500; Ethereum ended the month at $1,606, with a 60% decline from the 250-day high; Bitcoin spot ETF saw net outflows for 13 consecutive trading days from May 15 to June 3, totaling $4.4 billion, marking the longest outflow period on record, with a net outflow of $4.51 billion for the entire month of June, the worst performing month since the spot ETF's listing; the Crypto Fear and Greed Index dipped into the extreme fear zone with a reading as low as 15—while in the same week, the U.S. stock market was hitting all-time highs.

At the macro level, this quarter can be summarized by four points:

1. The rate-cut narrative has collapsed for the second time, and this time it is structural. The change in leadership has brought not only a shift in stance but also a reconstruction of the policy communication framework, significantly reducing the predictability of the interest rate path.

2. Inflation is supply-shock-driven, with a mismatch between policy tools and inflation triggers. As long as the situation in the Strait of Hormuz remains uncertain, the Fed does not have a dovish data foundation, and the duration of high interest rates will be prolonged.

3. Risk budget flows toward "physical bottlenecks". The combined impact of war and AI has seen capital embrace visible and tangible scarce links such as energy and storage, with crypto assets positioned on the opposite side of this preference.

4. The macro pricing power of crypto has yet to be regained. This quarter, the movement of crypto assets was almost entirely determined by external variables, and there has not been an incremental narrative within the industry strong enough to hedge against macro headwinds—but there have been signals of change on the RWA track, which we will explore in Part Three.

Global AI Technology and Stock Market Dynamics

This section revolves around three main themes: the pace and pattern change of large-scale model technology evolution in the second quarter, the sector rotation features of the global stock market under the AI hype, and the possible impact of the mega IPOs of OpenAI, Anthropic, and SpaceX on market liquidity and the AI valuation system. Together, they constitute a complete picture of this quarter's AI narrative transitioning from "technology faith" to "value validation."

Q2 Progress and Developments of AI Large Models

The second quarter of 2026 was one of the most intense periods of model releases in the large model industry in recent years. Following the first quarter where models such as GPT-5.3 Codex, Claude Sonnet 4.6, and Gemini 3.1 Pro intensely refreshed benchmark records, from April to June, the three major companies OpenAI, Anthropic, and Google continued their almost monthly iteration release cadence, while domestic companies also kept up with the trend.

Global Advanced Large Model Release Timeline Q2 2026

From a technical perspective, this quarter exhibited three distinct features:

• Parallel Base Model Retraining and Efficiency Optimization: GPT-5.5 is the first base model retrained from scratch by OpenAI since GPT-4.5, followed by the release of the GPT-5.5 Instant for everyday use scenarios, reflecting the manufacturers' dual focus on "capability ceiling" and "cost exploration."

• Inference and Agent Capability Become Competitive Focus: The Claude Opus series iterated continuously to versions 4.7 and 4.8 from April to May, with programming and long-context consistency steadily improving; on June 9, the release of Claude Fable 5 and Mythos 5 further raised the SWE-bench Verified benchmark to around 95%, representing one of the industry's highest levels in inference and long-chain intelligent agent tasks.

• Coexistence of Native Multimodality and Price Wars: Google's Gemini 3.5 series released at the May I/O conference processed text, image, audio, and video uniformly from day one of training, representing the "native multimodal" route; meanwhile, companies like OpenAI and DeepSeek significantly reduced prices, with the industry API costs continuing to decline, as "technology catch-up" and "price competition" unfold simultaneously.

It is worth noting that the presence of domestic companies significantly increased in this quarter: Zhipu GLM-5 became the first cutting-edge model trained entirely on Huawei's Ascend chip, reducing the hallucination rate to 1.2%; Kimi from Dark Side of the Moon introduced K2.5, the first open-source model to top the LMSYS Chatbot Arena. The simultaneous advancement of domestically controllable and open-source ecosystem-compliant computing power is reshaping the landscape previously dominated by U.S. manufacturers in standard setting. For the cryptocurrency market, the continuous rise in model capabilities is also directly accelerating the pace of implementing new narratives such as AI Agent integration with on-chain payments and stablecoin settlements.

Global Stock Market AI Frenzy: Sector Rotation Landscape

If the second quarter's flurry of big AI model releases embodied the "technology narrative," then the concurrent global stock market performance reflected the "capital narrative."

In the first half of 2026, the major global stock markets exhibited a distinct tiered divergence: the Korea Composite Stock Price Index (KOSPI) surged by 101.1%, galloping far ahead as the only major market index to double; the Taiwan Weighted Index (TAIEX) followed closely behind with a 59.3% gain; the Nikkei 225 Index rose by 39.2%; while the S&P 500 Index (+9.6%) and the CSI 300 Index (+7.5%) significantly lagged behind, with gains less than one-tenth of the Korean market. The magnitude of the disparity between these five markets—from the highest 101.1% to the lowest 7.5%—exceeded 13 times, marking an exceptionally rare level of differentiation among the major global markets in the past decade.

Comparison of Global Major Stock Market Gains in the First Half of 2026

The core logic behind this rare difference: the fluctuation of major market indices fundamentally depends on the weighting of the AI/semiconductor industry chain in that market. South Korea and Taiwan, the two markets with the most significant gains, happen to be the two economies with the highest concentration in the global semiconductor supply chain—Samsung Electronics and SK Hynix jointly accounted for 39% of the KOSPI's total market value at the beginning of the year, a proportion that climbed to 61%, while TSMC alone represented 42.87% of the TAIEX's total market value. This means that KOSPI and TAIEX are to a considerable extent no longer "diversified major market indices" in the traditional sense but rather closer to "proxy indices of the semiconductor industry chain." In contrast, due to their broader coverage of constituent stocks (encompassing a large number of traditional industries unrelated to AI such as finance, consumer goods, healthcare, and energy), the S&P 500 and CSI 300 saw the excess returns of AI hardware significantly diluted by other sectors of the indices, resulting in a notably more subdued performance.

To better illustrate the AI market situation, we have selected representative stocks from some of the representative sectors within AI, employing a free-float market capitalization-weighted index (USD-denominated, base period = 100, with the start of 2026 as the base) from both regional and sectoral perspectives. This method aligns with the compilation logic of mainstream benchmark indices such as the S&P 500 and KOSPI 200, providing a more accurate reflection of "how much money the entire market has gained", rather than being skewed by the extreme performance of individual small-cap stocks or heavyweights.

· Regional Representative Stock Mapping (Volume Priority Principle)

Weighted Index of Regional Representative Stocks' Free-Float Market Cap in the First Half of 2026 (Base Year: 2026 = 100)

In the first half of 2026, the world's top five stock markets, calculated by the free-float market cap weighted index (USD-based, 2026 = 100), showed a clear tiered differentiation: Korean Stock Exchange +195.8%, Japanese Stock Exchange +107.1%, A-share +83.1%, Taiwan Stock Exchange +57.5%, US Stock Exchange +12.9%—with a difference of over 15 times between the highest and lowest.

Looking at the monthly trend, these five markets from the end of February to the end of March experienced an almost synchronous collective pullback (Korean Stock Exchange dropped from 164.3 to 119.4, Japanese Stock Exchange from 117.7 to 95.5, A-share and Taiwan Stock Exchange also weakened, and the US Stock Exchange remained almost flat), and then entered a new uptrend channel from April. This synchronized pullback period aligns highly with macro disturbances—intensified geopolitical conflicts at the end of the first quarter, exacerbated divergence in the Federal Reserve's policy path, indicating that even the strongest global theme of AI was not completely immune to the synchronous impact of macro risks.

Entering the second quarter, these five markets began to significantly diverge: Korean Stock Exchange surged from 175.5 at the end of April to 295.8 at the end of June, showing the steepest and most sustained uptrend; Japanese Stock Exchange, A-share, and Taiwan Stock Exchange displayed relatively mild and steady climbs; whereas the US Stock Exchange followed a completely different trajectory from the other four markets—hitting a peak at 124.5 by the end of May, then falling to 112.9 by the end of June, being the only market among the five to experience a significant retreat in the closing stage of H1. The "madness" in the US Stock Exchange AI sector was more reflected in extreme performance in individual hardware segments such as storage, PCB, rather than a broad market rally. Representative AI leading stocks such as NVIDIA, Broadcom underperformed the market in H1, directly dragging down the overall index performance of the US Stock Exchange.

· Sector Representative Stock Mapping (Volume Priority Principle)

Weighted Index of AI-Related Sector in the First Half of 2026 (Base Year: 2026 = 100)

Breaking out of regional boundaries and redefining by industrial chain segment, the performance of the eleven H1 sectors presents a more distinct hierarchy: PCB/Semiconductor Packaging Substrate (+282.8%) and Storage Chip/Memory (+243.9%) are undoubtedly the strongest poles, with a particularly steep rise in the second quarter; Optical Communication/Optical Module/CPO (+123.3%), Computing Power Infrastructure/Cloud Services (+98.7%), Wafer Foundry/Manufacturing (+88.9%), Data Center Power and Energy (+70.1%) form the middle echelon; while AI Chip Design/Logic Chip (+19.9%) and Semiconductor Equipment and Materials (+5.7%), which are usually seen as the "AI core" sectors, saw growth rates significantly lagging behind the upstream bottleneck segments.

Of particular note are the top three sectors—Humanoid Robot/Physical AI (-0.2%), Hyperscale Cloud Providers (-5.9%), AI Large Model/Software Application Layer (-33.2%), the only three categories in H1 with collective negative growth, with the software application layer experiencing the deepest decline. Looking at the monthly trend line, the common feature of these three sectors is to narrow-range oscillate or even continue to decline around the 100 baseline throughout the entire period, never truly participating in the current upward trend—even though hyperscale cloud providers are the absolute main contributors to this round of AI capital expenditure (Microsoft, Amazon, Google, Meta), and humanoid robots are the hottest frontier topic in market discussions, the stock performance of both has failed to live up to the narrative hype.

The core logic revealed by this "pyramid" structure is: the pricing power of this global AI market cycle lies in the scarce links of the hardware supply chain, rather than in the contributors to capital expenditure or the ultimate application realization—the cloud providers that spend the most money, the software application layer and robotics raceway with the most lively stories are the three weakest-performing categories in H1; the sectors that truly squeezed out every bit of the price surge are in the midstream and upstream sectors such as storage and packaging substrates, which seem unremarkable but have the most limited short-term production capacity expansion capabilities.

Super IPO Shockwave: Capital Withdrawal and Profitability Question

The most far-reaching event of this quarter for the AI sector and even the entire capital market was the nearly simultaneous launch of the super IPO processes by the three giants SpaceX, OpenAI, and Anthropic. The combined target valuation of these three companies is approximately $36 trillion to $38 trillion, equivalent to the GDP of the sixth largest economy in the world, India, marking the largest-ever wave of technology company IPOs in capital market history.

· The Significant Scale of Funds Extracted from the Market

Three companies have a combined target fundraising scale of over $200 billion, while the entire U.S. IPO market only raised $450 billion throughout the entire year of 2025. This means that the three companies are attempting to extract from the public market an amount of funds equivalent to over four times the total U.S. IPO volume from last year in just a few months.

Although some analysis suggests that there is approximately $8 trillion in existing funds available in U.S. money market funds for absorption (SpaceX's $75 billion fundraising target accounting for only about 1% of this), and institutional funds have previously gained AI exposure more indirectly through “proxy targets” such as NVIDIA, Microsoft, and Google, the listing of pure AI assets is expected to unleash incremental demand. However, the rebalancing effect of funds is also worth noting—when hundreds of billions of dollars flow into new stocks, institutional investment portfolios will inevitably undergo reallocation, and funds are likely to passively flow out of existing “Tech Magnificent 7” stock holdings. Even investors who have never participated in IPO subscriptions may feel the headwinds of this “bloodletting” effect on their own positions.

· The Imminent Public Market Examination of Massive Losses

All three companies are currently heavily reliant on first-market financing “blood transfusions,” with a general cost structure that is deeply in the red:

1. OpenAI is the most fragile link among the three: With $5.7 billion in revenue in Q1 2026, a 300% year-on-year surge, it still incurred an operating loss of $9.3 billion—equivalent to paying out around $1.60 for every $1 earned; HSBC predicts its cash burn in the full year of 2026 to reach $17 billion, with a daily cash burn exceeding $46 million. More critically, Deutsche Bank estimates that OpenAI's cumulative negative free cash flow from 2024 to 2029 could reach $143 billion, with the earliest possible profitability not until 2029.

2. SpaceX's AI division is also a cash-burning force: Despite the Starlink business achieving stable profitability, the artificial intelligence segment, post-xAI merger, saw revenues of $818 million in Q1 2026, but operating losses soared to $2.469 billion, with an adjusted EBITDA of -$609 million; the company recorded a total loss of $4.94 billion in 2025, with a net loss of $4.3 billion in Q1 2026, accumulating losses reaching $41.3 billion.

3. Anthropic is currently the only exception: The company is expected to achieve approximately $559 million in quarterly operating profit in the second quarter of 2026, becoming the first of the three enterprises to be a profitable cutting-edge AI lab. However, its latest financing round valued the company at $965 billion. Even with the most optimistic annualized revenue estimate, the price-to-sales ratio still exceeds 24 times, higher than NVIDIA's historical peak of 28 times, which is the upper limit of a reasonable price-to-sales ratio. The valuation itself has already priced in quite optimistic growth expectations.

Initial Public Offering (IPO) prospectuses typically require disclosure of much more detailed and truthful financial data than the private placement stage. This means that the previously high valuations supported by "stories" and non-public data are about to undergo their first positive inspection by public market investors. Especially for OpenAI, its registration documents will publicly disclose real revenue and profit margin data for the first time. If the disclosed profit margin is significantly lower than investors' expectations based on previous private placement rounds, its target valuation may face significant downward pressure. This could trigger a series of markdowns involving venture capital firms holding a large amount of equity in related companies, and this impact could also transmit to the secondary market through "proxy shares" like Microsoft (holding approximately 27% of OpenAI) and SoftBank.

· Potential Impact on Market Sentiment and AI Stock Performance

From the recent performance of AI-related companies that have gone public, the vulnerability of this "asset-heavy, high burn rate" model has begun to surface. Taking CoreWeave, an AI compute infrastructure company, as an example, its free cash flow in the first quarter of 2026 was -$4.71 billion, with a single-quarter capital expenditure as high as $7.7 billion, and interest expenses doubling year-on-year. The total debt has soared from $2 billion in 2023 to $35 billion. Even with a 111.7% year-on-year revenue growth in the quarter, it still cannot dispel the market's doubts about the sustainability of its "burn rate for growth." Coupled with short reports and news of executive sell-offs, the stock price once fell more than 40% from its 52-week high. The late June "Black Tuesday" particularly highlighted this vulnerability—the market's concern about whether the investment in AI infrastructure can provide sustainable cash returns, combined with the large-scale bond financing of SpaceX before its IPO, collectively triggered a simultaneous global retreat in AI trading.

The development of AI’s large models continues to iterate and expand rapidly, with domestic manufacturers closely following innovation on the technological front. However, at the financial level, the market has shifted from the earliest and most prominent AI trading theme of "AI chip leader + software application layer + cloud providers" to the relatively newer theme of "storage/PCB-like AI hardware supply chain bottleneck stages." The stocks of the mega-scale cloud providers (Microsoft/Amazon/Google/Meta), who are the biggest contributors to AI capital expenditure, have not risen despite their heavy investments, indicating lingering market doubts about whether "burning cash can lead to returns." The three mega IPOs are collectively targeting raising over $200 billion, and the issue of huge losses is about to be directly tested by the public market. The stock market's AI sector is transitioning from a high-growth period to an uncertain and pressured phase. This also means that the AI stock market no longer has such a direct impact and appeal on the Crypto market, as seen in the analysis of the Real World Asset (RWA) market where signals of relevance are diminishing.

RWA Becomes the Largest Growth Engine in the Crypto Market

Real World Asset (RWA) Tokenization has emerged as the most institutionalized and fastest-growing track in the crypto market this quarter, serving as the most direct bridge between traditional finance and the crypto world. This section revolves around three main themes: the current scale and growth status of the overall RWA market; policy changes and background of the RWA market; the fastest-growing segment within the market—pattern classification of tokenized stocks; with Binance as the main sample, observing the data growth of RWA-related product lines.

Current State of the RWA Market

2026 marks the year that RWA transitioned from "concept validation" to "institutional infrastructure." According to RWA.xyz statistics, the on-chain RWA total size, excluding stablecoins, has rapidly increased from around $2.8 billion in early 2025 to over $33 billion by mid-2026, growing more than 11 times in 18 months. This growth far exceeds the overall DeFi market performance during the same period—while the total value locked (TVL) in DeFi declined from around $115 billion in early 2025 to $70 billion by mid-2026. RWA is one of the few asset categories in the entire crypto market that has seen net inflows during the bear market, demonstrating a clear structural shift in market funds from speculative protocols to institutional-grade on-chain infrastructure.

Changes in the On-chain RWA Market Size in the First Half of 2026 (rwa.xyz)

Changes in DeFi TVL Size in the First Half of 2026 (DefiLlama)

Looking at the asset structure of RWA

• US Treasury Bonds: Always the absolute mainstay, occupying the largest share of the RWA market value, this is the most mature, "boring," yet stable part of the RWA market. The attractiveness of Treasury bond tokenization is built on the logic that as long as the federal funds rate remains above 3%, the proposition of "putting idle stablecoins into tokenized Treasury bonds to earn 3.5%-4.5% APY annually" continues to hold, which is why tokenized Treasury bonds can securely maintain their position as the "ballast" throughout the entire RWA market—it essentially benefits from the high-interest-rate cycle of the Federal Reserve.

• Commodities: The second largest category, mainly gold tokenization, directly impacted by the fluctuation of gold prices.

• Asset-Backed Lending: Medium-scale, with significant increase in institutional participation in the last two years.

• Equities: While it may not seem significant from the image alone, it was actually the fastest-growing category this quarter.

Changes in the On-Chain RWA Equities Market Size in the First Half of 2026 (rwa.xyz)

RWA.xyz shows that tokenized equities increased from $291 million in 2025 to $1.816 billion in the first half of 2026. Due to statistical dimensions and restrictions of the rwa.xyz data collection framework, it did not actually cover the entire tokenized equities market data. Further dimensional data analysis supplements will be provided later in this article.

In the first half of 2026, the on-chain RWA market size expanded from around $21.6 billion at the beginning of the year to $33 billion, with the number of holding addresses growing at a rate far exceeding the asset's value itself. These data sets outline a market that is transitioning from "concept validation" to "institutional infrastructure" at an accelerated pace. However, the scale numbers alone only answer "what happened," not "why did it happen at this time." In fact, this round of expansion was not an isolated market behavior but highly synchronized with the global regulatory environment at that time—from the U.S. SEC's stance on tokenized securities and FINRA's issuance of the first custody license to Securitize, to the GENIUS Act, Hong Kong's Stablecoin Regulations, the EU's MiCA, and Mainland China's Document 42 almost simultaneously coming into effect. Behind each surge in the RWA market size in this period, there is almost always a corresponding policy relaxation footnote. To understand why the RWA market in this half-year expanded against the overall DeFi contraction backdrop, the focus must shift from the "market itself" to the "policy background." And behind this lies two regulatory tracks that are both independent and mutually supportive, progressing simultaneously: one overseeing whether "assets" can legally trade on-chain (securities regulatory track) and one governing what kind of "currency" is used for settlement in on-chain transactions (stablecoin regulatory track).

RWA Macro Policy Background: Global Stablecoin Regulation "Three Kingdoms," Concentrated Implementation in the First Half of 2026

In regulation, RWA is divided into the stablecoin regulatory track and the securities regulatory track, with the stablecoin regulatory track as follows:

The key legislation in this stablecoin regulatory track (GENIUS Act) was actually completed in 2025, with the first half of 2026 focusing more on the implementation period and global ripple effects: Using the GENIUS Act as a benchmark, Hong Kong, the European Union, and mainland China almost simultaneously introduced or strengthened their stablecoin regulatory frameworks from the end of the first quarter to the beginning of the second quarter in 2026, forming a de facto "regulatory trilogy" — the United States followed the federal framework + OCC licensing route, Hong Kong adopted a centralized licensing system by the HKMA, the EU pursued the dual licensing route under MiCA and PSD2, and mainland China implemented a strict "one-size-fits-all" approach domestically and provided access for foreign entities. This simultaneous global tightening of regulations did bring about short-term pressure — a significant retracement occurred in the stablecoin market in May and June, the most pronounced since 2023 (approximately $10 billion, 3%), mainly driven by the contraction of the two major players, USDT and USDC. The market interpreted this as some funds choosing to wait and see during a period when regulatory compliance costs were rising. However, the extent of the contraction was much smaller than the bear market run on the market in 2022 (which saw a contraction of over 26%), indicating that this time was more like "institutional labor pains" rather than a crisis of trust. What is even more noteworthy is the institutionalized actions that occurred simultaneously with the contraction — over 140 traditional giants jointly launched the Open USD Alliance, Swift formed a blockchain consortium with 17 banks, and Circle obtained an OCC trust charter. These all took place during the same period of decline in stablecoin volume, indicating that the regulatory tightening temporarily squeezed out speculative funds, but at the same time, it also accelerated the formal entry of traditional financial institutions in a more compliant manner. This is consistent with the logic behind FINRA issuing a license to Securitize on the "RWA Securitization" track: Both tracks, despite their different paths, are leading the industry from the "gray area" to "licensed operation."

The track of securities regulation is as follows:

Regarding securities regulation, it needs to define whether traditional securities such as stocks and fund shares can be legally tokenized, who is eligible for custody, and where they can be compliantly listed for trading. The entire timeline shows a rhythm of "taking two steps forward, one step back": The SEC's stance statement in January opened the door, the rule approvals of two major exchanges in March and the ADGM's preemptive landing further confirmed the direction, and on May 4th, FINRA formally licensed Securitize, marking the highest-level substantive endorsement by regulators this quarter — no longer just "posturing," but an actual licensed U.S. entity obtained the qualification to custody tokenized securities. However, in May of the same year, the SEC delayed the formal trading framework for tokenized U.S. stocks on a larger scale, indicating that regulatory agencies are still cautious about "how open the system should be" and are not blindly optimistic. The CFTC's actions on May 29th represent another path — the traditional, regulated derivatives market actively embracing 24/7 trading and crypto assets (Kalshi obtaining Bitcoin perpetual contract qualifications), which is the opposite of "crypto exchanges selling stocks." Although the SpaceX incident in June was not initiated by regulatory agencies, it exposed the structural flaws of the current tokenized stock model in the most direct way — while regulation provided a "path," it does not mean that the underlying asset supply and custody chain have already undergone stress tests. This is also why the DTCC chose to officially launch the pilot program in July, rather than hastily starting in January when the regulatory stance was just clarified: Traditional financial institutions have a significantly more conservative approach to the maturity of infrastructure compared to crypto exchanges.

However, the entire path of RWA became increasingly clear during this process. This has provided better policy support for the subsequent development of tokenized stocks.

Core Increment in the RWA Market: Tokenized Stocks

Due to the hot stock market in the past two years and the downturn in the crypto market, RWAs have become the sought-after source of innovation for crypto trading platforms. Tokenized stocks are undoubtedly the "core increment" of 2026 — although their stock size still accounts for the smallest proportion in the entire RWA market (about 2.5%), it is the fastest-growing segment of the year with the most influx of new players. As mentioned earlier, the data on rwa.xyz only shows one of the solutions for tokenized stocks. Strictly speaking, it can be divided into at least four different rights structures with completely different risk attributes. Investors cannot judge their attributes based solely on the product name:

· Real Stocks: Trading platforms rely on Alpaca and other brokerage infrastructure for real holdings

This is the model that is closest to the traditional securities account experience. The trading platform itself does not issue tokens but allows users to directly purchase and hold real stocks through licensed brokerage channels, with on-chain/apps only serving as a trading entry point. Binance is the latest representative of this model: in June 2026, it added an independent traditional US stock business, with trading execution handled by licensed Abu Dhabi brokerage Nest Trading, and Alpaca responsible for stock custody and dividend processing. In the first month of launch, it covered over 7,000 US stocks and ETFs, with spot transaction volume exceeding $1 billion within 30 days. On the day SpaceX went public, platforms such as Backpack Securities and Sunrise DeFi launched SPCX tokens following the same model — each token corresponds one-to-one with real SpaceX stocks held in a US brokerage.

· Tokenized Stocks: Synthetic packaging models such as xStock/Ondo/Securitize

This was the earliest solution for tokenized stocks and is currently the largest and most densely populated model. The key feature is that the issuer pledges full collateralization of real stocks at a 1:1 ratio and issues tokens representing the debt securities on-chain:

1. xStock (Backed Finance / Kraken): Launched in June 2025, issued by Swiss-registered asset issuer Backed Assets (JE) Limited, all products are fully backed by real stocks. Within 8 months of launch, the total market turnover exceeded $25 billion, making it the largest tokenized stock platform by trading volume, covering over 60 stocks and ETFs. Tesla, one of its targets, accounts for over a quarter of its custodied market value.

2. Ondo Finance (Global Markets): Launched in September 2025, covering over 260 U.S. stocks and ETFs (including SPY, QQQ, NVDA, TSLA, etc.). On May 11, 2026, the Total Value Locked (TVL) surpassed $1 billion, making it the first tokenized stock platform to achieve this milestone, capturing over 70% of the market share in the space with a total trading volume of approximately $18 billion. In June 2026, it further introduced a separate custody product tailored for U.S. compliant users, bringing assets like IVV and Micron under a more regulatory-friendly access model.

3. Securitize: The largest tokenization platform by assets under management (around $3.5 billion), notable for representing the BlackRock BUIDL Fund. Distinct from xStock and Ondo's "wrapping of third-party stocks," Securitize directly tokenizes its newly listed common stock on-chain. In July 2026, its tokenized holdings briefly peaked at $270.6 million, leading the overall tokenized stock charts, with a monthly on-chain transfer volume reaching $8.47 billion (a MoM increase of +92.77%).

· On-Chain RWA Perpetual Contracts: Permissionless Synthetic Derivatives

Represented by HyperliquidX's HIP-3 mechanism, this is a fully permissionless synthetic derivative market where any asset (commodities, stocks, etc.) can have a corresponding perpetual contract market within hours, as long as there is a reliable oracle price feed. Unlike the "issuer and custodian risk" faced by xStock and Ondo, on-chain RWA perpetual contracts encounter "oracle and liquidation risks" — price information must be accurate, and margins must be sufficient, or the position will be liquidated permanently. trade.xyz is the key player in this space.

· CEX RWA Perpetual Contracts: Traditional Asset Perpetual Contracts on Centralized Exchanges

This is the fastest-growing of the four models currently and the most significant new variable this quarter — centralized exchanges directly offer perpetual contracts tracking traditional asset prices such as stocks, commodities, and forex, settled in USDT, with up to 20x leverage and 24/7 trading. According to CoinDesk Research, the volume of RWA-related trades processed by cryptocurrency exchanges has nearly reached $1 trillion since 2026, with approximately 60.9% coming from Binance alone. The overall trading volume of RWA perpetual contracts surged from $12.37 billion in Q4 2025 to $203 billion in Q2 2026, a 20x QoQ increase mainly driven by commodity trading on Binance, Hyperliquid, and Pyth Network. This model will be detailed in conjunction with Binance's specific product lineup.

Among the current four modes of stock tokenization, the volume of RWA perpetual contracts is much larger than the other three types.

Seven RWA Product Lines Under the Binance Ecosystem

To understand the true impact of RWA-related products in the overall business of a crypto trading platform, the most direct way is to compare them within the product matrix of the same exchange. Binance is currently the most aggressive in RWA/TradFi expansion, with relatively full data disclosure among trading platforms, and also the globally most liquid crypto exchange. Therefore, using Binance as a sample for a single-framework comparison is quite representative.

Even though RWA perpetual contracts (TradFi) have experienced explosive growth from zero to an average monthly volume of hundreds of billion dollars over the past six months, their absolute scale is still one to two orders of magnitude lower compared to Binance's crypto-native business (spot, USDT-margined/coin-margined perpetual contracts). The monthly trading volume of Binance's cryptocurrency USDT-margined perpetual contracts has consistently remained at the level of hundreds of billions to trillions of dollars, which is 4-5 times higher than the peak of TradFi perpetual contracts. The two newest product lines, "Real Stocks" and "bStocks," currently have a volume even one to two orders of magnitude smaller than RWA perpetual contracts and are still in the very early stage of business incubation.

However, the comparison of growth rates tells a completely different story. Binance's TradFi perpetual contracts were launched on January 28, 2026, with a trading volume of only a few hundred million dollars in the first week of launch. Within two months, the cumulative trading volume exceeded $153 billion, and the market share in the first quarter reached 62.7%. By entering May, the weekly trading volume once reached $603 billion, accounting for 10.3% of the platform's total perpetual contract trading volume. By June, Binance had captured an 80% share of the CEX stock perpetual contract market, with the weekly trading volume of CEX stock derivatives hitting a historic record of $11.6 billion in that month. A corresponding surge was seen in Pre-IPO perpetual contracts — the SpaceX Pre-IPO contract captured a 65% market share within a week of launch, with a cumulative trading volume of $400 million; the OpenAI Pre-IPO contract went live for two days, with 85% of the total market trading volume occurring on Binance. A business line that was zero just six months ago now accounts for one-tenth of the overall perpetual contract trading volume on the platform, a growth rate unprecedented in the product incubation history of any traditional financial institution.

To more intuitively demonstrate the data increment of RWA Perp, we compare the data of the largest CEX and DEX in the crypto industry together (alpha and bStock/Stocks currently do not have API exports).

Binance and Hyperliquid's Crypto and RWA H1 2026 Monthly Transaction Volume Statistics (Binance and Hyperliquid API)

Binance and Hyperliquid's Crypto and RWA H1 2026 Monthly Transaction Volume Statistics

From the above intuitive comparison, we have drawn the following conclusions:

1. The impact of the US stock market on Crypto spot trading is the most significant, with a whopping 40% drop in May compared to January.

2. Crypto U perpetual contracts, in contrast to spot trading, experienced a smaller impact from the US stock market, with a year-on-year drop of 18%. This trend signifies the gamification of Crypto among existing players.

3. The data increment on RWA perpetual contracts for platforms like trade.xyz, representing RWA DEX, and Binance, representing RWA CEX, is quite remarkable. Trade.xyz saw a 4x growth in the first half of this year, while Binance exhibited exponential growth. Leveraging its user base, Binance's current trading volume has surpassed that of trade.xyz. Furthermore, Binance's RWA data was generated with only 90 trading pairs, while in the same period, Binance had 588 U perpetual contract trading pairs in June.

4. The peak impact on Crypto has already passed, and from the data, it is evident that the Crypto market is slowly recovering.

Tokenized Stock Weekly Transfer Volume (rwa.xyz)

Simultaneously, on-chain tokens also reflect similar growth in data. Starting from 408M at the beginning of 2026 to 3982M in mid-June 2026. Combining this with the data from the above graph, we boldly propose a hypothesis. Throughout 2026, Crypto has been continuously impacted by the US stock market, whether in terms of the underlying asset's value and market compliance or the speculative nature of prices and gains. Tokenized assets have lagged behind AI concept stocks. However, this impact reached a peak in the data for June 2026. We believe that if the Crypto market does not introduce new bearish news, in terms of the impact from the US stock market, we are at a trough.

Our Assessment of Tokenized Stocks:

1. For tokenized stocks, our prediction is as follows: Due to the lack of a first-mover advantage, older solutions like xStock and Ondo may be squeezed out by Binance's bstock solution in subsequent competition. Additionally, due to liquidity issues, tokenized stocks may develop more slowly compared to other solutions.

2. Real-world asset (RWA) stock solutions, with their first-mover advantage, will gradually replace tokenized stock solutions and continue to grow steadily.

3. RWA Perp solutions, whether on centralized exchanges (CEX) or decentralized exchanges (DEX), trading platforms will not give up on the most growth-oriented segment and will continue to inject resources into it. At the current growth rate, it should quickly catch up to the market size of crypto perpetual futures.

Summary

In the first half of 2026, the cryptocurrency market can be summarized in one sentence: multiple external impacts and internal shift of focus.

On a macro level, the change in leadership at the Federal Reserve has not only brought about a shift from a rate-cutting to a rate-hiking stance but also a restructuring of the policy communication framework, significantly reducing the predictability of the interest rate path. The U.S.-Iran war and the closure of the Strait of Hormuz have led to a supply shock-type inflation that cannot be resolved by rate hikes but is forcing the Federal Reserve to avoid a dovish stance. The combination of these factors has led to extreme volatility in the pricing environment for cryptocurrency assets throughout the first half of the year.

On the AI front, technological iterations are still happening at a rapid pace, but the narrative has shifted from "AI chip leaders + software application layer + cloud providers" to "storage/PCB and other hardware supply chain bottlenecks." The share prices of the mega-scale cloud providers, who invest the most, have not risen, and the market is starting to question whether "burning money can yield returns." The combined IPO funding of the big three tech giants exceeds $200 billion, and their massive losses are about to face public market scrutiny. The AI stock market is transitioning from a period of rapid growth to a period of pressure, meaning that the strongest moment of AI's impact on the cryptocurrency market may have already passed.

On the cryptocurrency front, RWAs were the only asset class to see a net inflow this quarter: On-chain RWA has expanded from $21.6 billion at the beginning of the year to $33 billion, while DeFi TVL has slid from $115 billion to $70 billion during the same period. Among the four tokenized stock models, RWA perpetual contracts' trading volume is much larger than the other three categories—Binance TradFi Perp accounted for one-tenth of the platform's perpetual trading volume in just half a year, and Hyperliquid's TradFi trading volume even represents half of the platform's volume. The crypto industry couldn't keep the funds' demand for coins, but it caught the demand for the "traditional asset trading conduit."

Our basic assessment is that, in terms of the impact of the US stock market on the crypto market, the data from June 2026 is likely the bottom; if the market itself does not create new bearish news, the recovery may already be slowly underway.

Original Article Link

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia