Hong Kong’s monetary regulator warned that tokens using the names HKDAP and HSBC are unauthorized and unrelated to the city’s newly licensed stablecoin issuers, which said no public products have launched.
Hong Kong has awarded its first stablecoin licenses to HSBC and Anchorpoint, with the initial phase centered on HKD-backed tokens that may launch in mid to late 2026.
Hong Kong regulators warned that fake tokens using the names HSBC and HKDAP are circulating before any regulated stablecoin has officially launched.
Anchorpoint Financial Technology, a JV backed by Standard Chartered, HKT and Animoca Brands, secured a stablecoin issuer license from HKMA and plans a phased rollout of 1:1 HKD-backed stablecoin HKDAP starting Q2 2026.
Circle’s stock has fallen from more than $260 in June last year to $62 on July 19, a decline of roughly 76%, as investors reassess how to value stablecoin issuers. The company’s long-term strategy remains central to management’s message, but Wall Street has turned more cautious. Mizuho cut CRCL to “underperform” from “neutral” and lowered its price target to $50 from $85, arguing that even if interest rates stay elevated through 2027, pricing pressure and tougher competition could erode margins. The competitive picture is also shifting. USDC still has scale, with about $73 billion in circulation across 34 blockchains, yet Circle is now facing a fresh challenge from Open USD, a new stablecoin backed by around 140 companies and designed to attract partners by sharing reserve income. Visa has also launched its Visa Stablecoin Platform, initially supporting Open USD while remaining compatible with USDC and Paxos’s USDG. At the same time, regulatory pressure is building elsewhere in the sector. The report says Tether may have about two years to bring USDT in line with the GENIUS Act, while in Hong Kong, Standard Chartered (Hong Kong) and Anchorpoint Financial Technology could announce a Hong Kong dollar-pegged stablecoin, HKDAP, before the end of this month.
PANews’ July 21 daily roundup tracked a broad set of crypto and market developments, led by Donald Trump agreeing to an ethics provision tied to the Clarity Act and Ant International closing an approximately $1.2 billion Series A round. The ethics language is designed to limit the ability of senior U.S. officials, including the president, vice president, and members of Congress, to profit from digital assets while in office. Debate around the clause has centered on Trump’s meme coin and World Liberty Financial, with the bill text expected in the coming days and a Senate vote required before the first week of August. The digest also covered fresh product and platform updates from Binance, Robinhood, Base, Solv Protocol, and others, along with several security incidents. Among them were the reported theft of about 515 million NIGHT tokens from Wanchain’s Cardano bridge treasury and ZachXBT’s claim that TeleSwap may have suffered an exploit worth more than $735,000 that had still not been publicly disclosed five days later. On the market side, U.S. spot Bitcoin ETFs posted $227 million in net inflows on July 20, extending the streak to five straight days. Strategy did not add to its Bitcoin holdings last week but added $225 million to its dollar reserves, while one company executive said the firm’s Bitcoin treasury could support dividend payments for 31 years.
Moonshot AI this week unveiled Kimi K3, a 2.8 trillion-parameter model described in the source as the largest open-weight model to date. The company said full weights are expected to go live before July 27 under a Modified-MIT license, allowing self-hosting. But the hardware bar is far above consumer setups. In Moonshot AI’s deployment guidance, running K3 on your own requires at least 64 accelerators. Even before inference begins, storing the model’s native 4-bit MXFP4 weights takes roughly 1.4TB to 1.5TB. Based on raw weight size alone, that works out to about 19 NVIDIA H100 80GB GPUs, 11 H200s, or 8 B200s, and that does not include KV cache for long-context operation. The article also cites H100 pricing at roughly $25,000 to $33,000 per PCIe card, with SXM versions above $35,000 to $40,000, while an 8-GPU H100 server costs more than $300,000. At Moonshot AI’s stated 64-GPU threshold, hardware procurement alone would exceed $2.4 million. Power use is also heavy: 64 H100 GPUs at about 700W each add up to roughly 45kW, pushing deployment into data-center territory rather than home or hobbyist environments.
SemiAnalysis said Kimi K3 can sharply reduce KV-cache transfer bandwidth through its KDA-based design, but argued that this does not point to a major contraction in the AI networking market. According to the research firm, roughly three-quarters of Kimi K3’s network layers use KDA, which can cut KV-cache transmission bandwidth by as much as 10x versus a full global-attention model. Even so, the model’s overall infrastructure demands remain large. SemiAnalysis said Kimi K3 has 2.8 trillion parameters and still requires about 1.5TB of HBM bandwidth per forward pass even with MXFP4. To deploy the model profitably while maintaining reasonable interaction speed, operators would still need to connect large numbers of chips through high-bandwidth networking systems such as GB300 NVL72 and rely on WideEP for scaling. The firm added that WideEP distributes 896 expert models across multiple GPUs and performs token dispatch and result merging twice per layer in each forward pass, exceeding 120 operations in a single pass. By comparison, KV-cache transfer between prefilling and decoding happens only once per dialogue round, suggesting the bandwidth saved by KDA may be smaller than the additional scaling demand created by large expert-model architectures.
