AMD secures a tens-of-billions-of-dollars chip order from Anthropic, and will invest an extra $5 billion to acquire an equity stake.

Contract whale "Xian Ding 10 Big Goals" posted a notice outlining its trading plan, stating that its previous BTC long position was entered near $64,980 and exited after triggering a stop-loss when the price dropped to $64,000–$64,300. The stop-loss was set to prevent a further rapid plunge to $63,000 if the $64,000 level broke. The trader will continue monitoring the market; if $64,000 is not effectively broken, it will gradually re-enter long positions near $64,000 next Monday and Tuesday, with a new stop-loss level set at $63,500–$63,000. On July 21, "Xian Ding 10 Big Goals" set a $300 million profit target for this round: in the prior round, it used 150 BTC to open positions targeting $150 million, realizing a maximum profit of $120 million. However, it misjudged the direction during a plunge from $120,000 at the last minute, erasing all profits and ultimately preserving only its principal plus a small gain. For this round, it is using 300 BTC to open positions with a $300 million target, and has so far booked $60 million in profit.

According to Bloomberg, DeepSeek has announced the suspension of its second round of financing, and has verbally notified some potential investors that it will not sign the investment agreement as scheduled. The suspension is partly rooted in founder Liang Wenfeng’s dissatisfaction over the leak of a transcript of his internal remarks made during the company’s first financing round, where he delivered 52 speeches at investor meetings covering DeepSeek’s organizational culture, open-source logic, technical roadmap, and views on the industry’s competitive landscape. DeepSeek noted that it may still resume the transaction process in the future, and has already begun preparations for an initial public offering (IPO), with the earliest possible filing this year. DeepSeek completed its first external financing round in June 2026, raising over 50 billion yuan (approximately $7.4 billion) at a post-money valuation of around $50–59 billion. The round adopted a special structure: most funds flowed into a limited partnership managed by Liang Wenfeng, investors held no voting rights and faced a 5-year lock-up period, with Liang himself contributing roughly 20 billion yuan as the largest single investor. Participants in the round included Tencent, CATL, JD.com, NetEase, and China’s National AI Fund. DeepSeek subsequently launched discussions for its second financing round in mid-July, targeting a valuation of approximately $71 billion.

DeepSeek announced the suspension of its second round of financing, and has verbally notified some potential investors that it will not sign the investment agreement as scheduled. The suspension is partly due to founder Liang Wenfeng's dissatisfaction with online reports about his remarks to investors during the first financing round. DeepSeek may still choose to resume the transaction process in the future, and has begun preparing for its initial public offering (IPO), which could be filed as early as this year.

An Iranian source said that Iran's Fordow nuclear facility is not only heavily fortified but also "empty", adding that any U.S. attack on Fordow is doomed to fail. The source noted that Trump aims to score political points by striking Iran's Fordow site, though it is widely known that attacking an empty facility only serves to spread disinformation and brings no practical benefits. The official also stated that Iran will respond to such an action in multiple ways, "not limited to large-scale military operations". Earlier on the 22nd, the U.S. notified Israel of its plan to escalate strikes against Iran in the coming days, including the first use of heavy bombers since the U.S. and Israel launched military operations against Iran this year. The heavy bombers are expected to target Iran's Fordow underground nuclear facility. Trump has repeatedly threatened over the past week that he could "soon" strike the region. Located near the Natanz nuclear facility in Isfahan Province, central Iran, Fordow is considered one of Iran's most heavily fortified underground nuclear sites. (Xinhua News Agency)

Renowned analyst PlanB has voiced opposition to Bitcoin’s BIP-110 proposal: “I oppose the BIP-110 fork. In my view, supporters of the fork fail to grasp three key points: Bitcoin is a decentralized bearer asset; Bitcoin enthusiasts hate spam more than they do; and why the previous BCH fork failed.” BIP-110 is a temporary soft fork proposal that would add seven consensus limits over its roughly one-year validity period, including restrictions on new script public key lengths, the size of certain push data and witness items, and disabling some Taproot extension paths. The proposal aims to drastically reduce arbitrary non-payment data (such as inscriptions and runes) embedded in transactions, thereby lowering node burdens, curbing spam, and refocusing Bitcoin on its monetary use case. Currently, BIP-110 only has around 1% miner support, far below the 55% threshold required for adoption. Its prospects are dim, making it nearly impossible to become an official rule on Bitcoin’s main chain. The final outcome will be revealed after the mandatory signaling window opens around August 7-8.

Citrini analyst Jukan has responded to recent bearish NAND notes and negative rumors about QLC price negotiations circulating in the market. Accepting a price lower than the initial offer when SanDisk signed a long-term agreement (LTA) with Meta is not surprising. As the most active NAND vendor in pursuing LTAs, SanDisk plans to allocate over 50% of its total shipments to such deals. Based on this strategy, it is naturally willing to accept LTA prices lower than current quarterly contract rates, so one cannot infer that "SanDisk cannot seamlessly resell all orders to higher-bidding North American clients." Regarding the rumor that Chinese module manufacturers were rejected when promoting eSSDs to domestic cloud service providers (CSPs), Jukan explained that Chinese CSPs have direct procurement channels from Yangtze Memory Technologies (YMTC) rather than insufficient demand. As for the claim that hyperscale cloud vendors are pressuring down QLC eSSD prices leading to unsold volumes, he noted that new cloud vendors have enough demand to absorb these volumes. Jukan concluded that negative headlines tend to be amplified when storage stocks underperform, but the sector’s fundamentals have not seen substantial deterioration. He reaffirmed his "bullish stance on storage." Earlier, Jukan had stated that DRAM contract prices still have around 40% upside potential by the end of 2027, and HBM supply remains tight. This clarification on the NAND segment further solidifies his bullish outlook for the entire storage space.