Crypto Morning Brief: Moonshot AI Plans to Complete IPO in Hong Kong Within Six Months, China's Daily Average Token Call Volume Increases Over 1000-Fold
TechFlow Selected TechFlow Selected
Crypto Morning Brief: Moonshot AI Plans to Complete IPO in Hong Kong Within Six Months, China's Daily Average Token Call Volume Increases Over 1000-Fold
Approximately 50% of Bitcoin supply has changed hands above $59,000, and the market bottom range may be forming.
Navigating Web3 tides with focused insightsAuthor: TechFlow
Yesterday's Market Dynamics
Kevin Kelly: If Token Costs Become Crucial in the Future, Chinese Open-Source Models Will Have a Significant Advantage
On July 18, at the 2026 World Artificial Intelligence Conference, famous futurist and "Father of Silicon Valley Spirit" Kevin Kelly stated in an interview with media that if everyone starts caring about Token (token) costs someday, the existence of open-source models will be an advantage for Chinese AI.
Kevin Kelly mentioned that Token (token) consumption costs are becoming increasingly important, though currently everyone seems unconcerned about costs, "But I think when we are consuming such massive Tokens all the time, everyone will start caring (about costs). If you can provide costs that are only one-tenth of Anthropic's, that will disrupt the situation."
However, Kevin Kelly also reminded that the open-source model requires sufficient funding to maintain operations, as it is not as profitable as the closed-source model, "Building these large models requires massive funding."
Moonshot Allegedly Plans to Complete IPO in Hong Kong Within Six Months, Latest Valuation May Exceed $30 Billion
On July 19, according to Bloomberg citing informed sources, Chinese AI company Moonshot has distributed shareholder resolutions to investors, seeking support for its listing in Hong Kong, with the IPO potentially happening as early as within the next six months.
Informed sources stated that Moonshot is currently completing a round of financing, and the company's valuation may exceed $30 billion after financing.
After achieving $300 million in ARR (ARR, an important metric measuring future sales capability) in June this year and seeing multiple-fold growth following the release of the K3 model, the company believes the timing is ripe.
It is reported that Moonshot's annual recurring revenue tripled within 3 months previously. The new model Kimi K3 surpasses competitors other than Claude Fable 5 and GPT-5.6 in comprehensive capabilities. Kimi K3's pricing is roughly equivalent to Anthropic's Sonnet.
Claude Adjusts Subscription Strategy Due to Compute Pressure, Fable 5 Included in High-Tier Plans with 50% Quota Limit
On July 18, Claude announced that starting July 20, Fable 5 will be officially included in the Max and Team Premium subscription plans, and users can obtain usage rights equivalent to 50% of the full quota.
Pro and Team Standard users can still access Fable 5 by using credits and will receive a one-time $100 quota compensation.
Claude stated that since the scale of demand after Fable 5's launch was difficult to predict accurately, a phased opening strategy was adopted previously, and the access scope for subscription users was expanded multiple times as compute resources increased. This process brought uncertainty to some users, so this adjustment aims to let users understand the model usage rights included in the subscription plan more clearly.
In the future, investment will continue to expand computing capacity, and Fable 5's access policy will be continuously updated based on capacity conditions.
Analysis: Approximately 50% of Bitcoin Supply Has Changed Hands Above $59,000, Market Bottom Range May Be Forming
On July 19, CryptoQuant analyst Darkfost stated that data shows Bitcoin is currently building important support levels in the $59,000 to $70,000 range, which has also become one of the most fiercely defended price ranges in Bitcoin's history.
Currently, about 50% of the total Bitcoin supply has changed hands above $59,000. If the millions of Bitcoins believed to be permanently lost are excluded, this ratio will increase further.
This round of turnover is mainly driven by short-term holders. The behavior of market participants is diverging; some investors choose to panic sell, while others continue to increase their holdings.
Although multiple indicators have entered extremely pessimistic or oversold regions, and the formation of the $59,000 to $70,000 range has certain rationality, this does not mean the market has confirmed a bottom. More accurately, the Bitcoin bottom structure is currently still in the process of construction.
Musk: Institutions Maintaining Large Short Positions on SpaceX Long-Term Have Extremely Low Survival Probability
On July 19, Musk posted on the X platform responding to "SpaceX stock price fell 30% from peak after IPO, shorts profited about $8.7 billion," stating that institutions maintaining significant short positions on SpaceX long-term have a very low probability of continued survival.
Serenity: SK Hynix ADR Premium Over Korean Stocks Exceeds 25%, Conversion Opening on July 29 May Trigger Arbitrage
On July 19, "White Hair Stock God" Serenity posted stating that currently SK Hynix American Depositary Receipts (ADR) have a premium of over 25% compared to Korean local stocks, and the conversion mechanism between ADR and Korean stocks will officially open on July 29. This may create conditions for market arbitrage trading and drive the narrowing of the price spread between the two locations.
Serenity pointed out that currently only about 2.5% of SK Hynix shares belong to circulated US ADRs, and there is conversion space for about 22.5% of shares in the future.
As the conversion channel opens, arbitrage funds may profit by buying Korean local stocks and converting them to ADRs, thereby driving up Korean stock prices or exerting pressure on US stock ADR prices.
Changxin Technology Winning Numbers Released, Totaling Approximately 7.7022 Million
On July 19, according to Changxin Technology's announcement, the company's initial public offering and listing on the STAR Market offline preliminary allocation results and online winning results have been published.
The issue price is 8.66 yuan/share, the initial number of shares issued is approximately 6.688 billion shares, and there are a total of 7,702,207 winning numbers.
Each winning number can only subscribe for 500 shares of Changxin Technology A-share stock.
CAICT: China's Daily Average Token Call Volume Grows Over 1000 Times
On July 18, Wei Liang, Vice President of the China Academy of Information and Communications Technology, stated that China's daily average Token call volume has soared from about 100 billion at the beginning of 2024 to 140 trillion by the end of March 2026, growing over 1000 times.
Agents are spawning a completely new economic form, namely the "Token Economy".
Token is called "token" in Chinese, and is an important unit for large models to process information and measure model call costs.
Its growth rate is the most intuitive barometer of industry development.
In the agent era, a user command may trigger multiple rounds of model calls and resource consumption, directly driving an explosion in artificial intelligence compute demand.
With the scaled application of agents, a series of new models surrounding Token measurement, scheduling, pricing, and trading are forming.
JPMorgan: Strategy Cash Reserves Increase to $3 Billion, Bitcoin Futures Inflow Releases Positive Signal
On July 18, according to The Block, JPMorgan analysts stated that Strategy's recent increase in USD reserves, and the Bitcoin futures market recording capital inflows, are "encouraging signs" in Bitcoin's outlook, although spot Bitcoin ETF capital flows remain relatively volatile recently.
Analysts pointed out that spot Bitcoin ETFs recorded capital inflows last week, but turned to outflows this week.
In comparison, leveraged ETFs linked to Strategy have maintained relatively stable positive inflows for 7 consecutive weeks.
JPMorgan believes that this part of buying mainly comes from retail investors, which may provide support for Strategy's stock price and prevent its common stock from falling below the net asset value of the Bitcoin held by the company.
Strategy recently increased USD reserves from $2.55 billion to $3 billion, equivalent to about 20 months of preferred stock dividend expenditure.
JPMorgan previously stated that if the company can reserve enough cash to pay 2 to 3 years of preferred stock dividends, it will help alleviate market concerns about potentially selling Bitcoin to pay dividends in the future.
Analysts emphasized that it is currently difficult to judge whether Strategy's increase in cash reserves has directly improved Bitcoin investor sentiment.
However, while spot Bitcoin ETFs saw capital outflows, CME Bitcoin futures and perpetual futures still recorded positive inflows this week, showing that futures market demand led by institutional investors has improved.
Strategy President and CEO Phong Le previously stated that the company still plans to buy Bitcoin long-term and hopes to continue being the largest Bitcoin buyer in the market in the foreseeable future.
The company also plans to issue further after STRC preferred stock rebounds to a par value of $100 per share, and the proceeds may be used to purchase Bitcoin and increase USD reserves.
Today's Market Conditions
Recommended Reading
Podcast Notes | SemiAnalysis Breaks Down Kimi k3: China Finally Has a Frontier Model, AI Labs Selling Tokens May Be More Profitable Than SaaS
https://www.techflowpost.com/article/32636
Moonshot (Moonshot AI) released Kimi K3, surpassing Google and Meta on multiple comprehensive benchmarks, becoming the world's third-best model, second only to Anthropic's Fable and OpenAI's Soul 5.6.
Two analysts from SemiAnalysis, Jordan and Max, broke down the implications of this release in an emergency program: A model with 2.8T parameters is served at the same pricing as Sonnet (3/15 per million tokens). If it is comparable in scale to frontier closed-source models, then Anthropic's profit margin of 10/50 could be at a mindboggling level.
A sharper judgment comes from Jordan: The frontier gap is narrowing, which he attributes to the US government restricting Anthropic/OpenAI from releasing the strongest models, artificially giving chasers a time window.
Open source has not truly caught up.
At the same time, Western open source is completely vacant; no American company's open-source model can catch up with China's fifth-best.
Model competition is becoming a competition of harness (using tools) and geopolitics.
Podcast Notes | Frontline Staff at Silicon Valley AI Companies Explain FDE in Detail, the Hottest Job in the AI Industry Right Now
https://www.techflowpost.com/article/32635
Jove currently leads the FDE team at AI call center company Cresta, expanding from a few people when taking over last year to 30 people, with a target of 100 people this year, recruiting globally.
Cresta itself is a 9-year-old enterprise service company, with clients including Marriott and United Airlines, always doing customer service experience optimization, and naturally transitioning to AI Agent deployment after the AI wave came.
The core tension of this dialogue lies in: FDE is being sought after by top AI companies like OpenAI and Anthropic and the VC circle as "the most promising job in the AI era", but when Jove organized events in the Bay Area, he found that engineers from big factories like Meta and Google had not even heard of this job.
Jove's view is very clear: AI Agent FDE is not a traditional outsourced or on-site engineer, but a new species: You need to understand all the pitfalls of AI implementation (hallucinations, RAG latency, VAD debugging, model version switching), be able to win the trust of client CTOs face-to-face, and also bring frontline lessons back to the company to improve products.
Essentially, it is recruiting a group of "startup CTOs" to make AI land.
JPMorgan Research Report Interpretation: AI Panic Narrative Has Not Finished, Mid-August Is the Entry Window
https://www.techflowpost.com/article/32629
The Philadelphia Semiconductor Index (SOX) has fallen cumulatively by about 19% since hitting a historical high on June 22, approaching the 20% bear market threshold.
AI is still falling, JPMorgan says: Don't rush to enter the market to buy the dip yet.
This judgment comes from the "AI Bubble Interest Score" constructed by JPMorgan's quantitative team, which tracks how hot global media is talking about the AI bubble.
The score peaked on June 29 and is now starting to decline.
According to historical patterns, this narrative cycle from hot to cold is about 1.5 months, dropping to the low point as early as mid-August.
The problem is that the absolute value of the score is still at 687, remaining in the highest quintile in history.
Even if it starts cooling down now, as long as it remains in the high-level range, there is still a 53.8% probability of SOX experiencing a drawdown of more than 8% in the next 20 days.
JPMorgan's quantitative model states plainly: The AI alarm has not been lifted yet.
Before mid-August, buying the dip on AI semiconductors is high risk.
Trump's Posts Sold to Wall Street by Millisecond, Counting Truth Social Parent Company's "Influence Business Strategy"
https://www.techflowpost.com/article/32617
Truth Social parent company Trump Media (DJT) announced that starting August 1, it will sell millisecond-level real-time data of Trump's posts to hedge funds.
This is not an isolated event.
From SPAC listing to $2.5 billion Bitcoin treasury, $6 billion nuclear fusion M&A, to today's data licensing, this company with quarterly revenue of less than $1 million has been doing the same thing for two years: systematically productizing Trump's influence.
On the same day, CNN exposed that Trump publicly praised these companies on Truth Social after buying stocks of 21 companies.
Truth Social has finally officially priced its most valuable asset.
According to reports from Fast Company and other media on July 16, Truth Social parent company Trump Media & Technology Group (Nasdaq: DJT, hereinafter referred to as Trump Media) announced the launch of Truth API, providing institutional clients with millisecond-level real-time data push of posts from the top 10 high-impact accounts on Truth Social, officially launching on August 1.
Trump Media Interim CEO Kevin McGurn stated in the announcement, "The market is already fluctuating due to Truth Social posts," and Truth API is part of the company's strategy to "monetize proprietary assets through high-margin, recurring revenue streams."
The company stated that it has signed some clients before launch, and pricing has not been disclosed yet.
Viewed separately, this is just a social media platform opening an enterprise-level API; X and Reddit have both done similar things.
But viewed within Trump Media's monetization path over the past two years, Truth API marks that this company has completed the closed loop from "selling stocks" to "selling influence itself."
Ansem: Why Do I Think Token Buybacks Solve Nothing?
https://www.techflowpost.com/article/32616
Famous Solana trader Ansem posted stating that the buyback mechanism itself does not create value; what determines the token valuation multiple is the "trust premium" between the team and the community.
He used Hyperliquid (annualized revenue about $800 million, FDV about $65 billion) and pump.fun (annualized revenue about $440 million, FDV only about $1.4 billion) as controls, pointing out that both are buying back on a large scale, but the valuation multiples differ by nearly 50 times.
One of the crypto market's most enduring narratives is being challenged: Pouring protocol revenue into buybacks will cause token prices to rise.
On July 16, famous Solana trader Ansem (@blknoiz06) posted a long thread on the X platform, proposing a counter-intuitive assertion: The buyback mechanism itself does not create value; what truly determines the token trading multiple is the "trust premium" between the team and the community.
After this post was published, it quickly gained over 469,000 views, 3,240 likes, and 509 reposts.
Ansem chose two of the strongest revenue-generating protocols in the crypto industry as controls.
Hyperliquid annualized revenue about $800 million, HYPE FDV about $65 billion; pump.fun annualized revenue about $440 million, HYPE FDV about $65 billion; pump.fun annualized revenue about $440 million, HYPE FDV about $65 billion; pump.fun annualized revenue about $440 million, PUMP FDV only about $1.4 billion.
Ansem's conclusion is: The gap is not in revenue scale, but in the trust accumulated by team behavior.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
深潮TechFlow
