Bitcoin falls below $63,000

According to official announcements, Spark will shut down its SparkLend lending protocol on Gnosis Chain on September 14. After the service termination, all outstanding loans may face liquidation, and users are advised to repay their debts and withdraw collateral before the deadline.

Beating AI Express: Tencent Hunyuan has open-sourced AuK, a 1.5-billion-parameter speech generation and editing model, which the official dubs the "audio version of Nano Banana". The model integrates functions including voice cloning, word editing, emotion switching, accent removal, speed and pitch adjustment, noise reduction, and multi-person/music separation, all controllable via natural language. AuK can directly modify existing recordings: it only adjusts incorrect words without re-recording the entire segment, alters emotion/timbre/accent while keeping content unchanged, edits lyrics, removes laughter/coughs, and converts normal speech to whispers. Notably, its voice cloning does not require a reference transcript—only a sample audio and new text are needed. In official tests, AuK leads in multiple speech generation and general speech editing benchmarks, scoring 49.73 on SpeechEditBench, while second-place Ming-UniAudio scores 28.70. The fast variant AuK-Flash, after distillation, needs just 4 inference steps and runs ~4.5x faster. However, the paper notes AuK cannot yet reliably understand all arbitrary natural language inputs, still relying on a Prompt Enhancer to identify tasks, organize parameters and rewrite instructions. Code and model weights are now public under the MIT license.

Beating AI News Flash: Moonshot AI is considering a dual listing on the Hong Kong Stock Exchange (HKEX) and the Shanghai Stock Exchange’s STAR Market. The company has confidentially filed an IPO application with HKEX, and is now signaling to investors that it may list on the STAR Market after its Hong Kong debut. The STAR Market plan remains in the exploratory stage and has not been finalized by the firm. For its Hong Kong listing, Moonshot AI aims to raise around $30 billion, with a potential earliest launch in Q1 2027. The company is currently advancing its final pre-IPO financing round, with a valuation of approximately $50 billion. Moonshot AI’s consideration of a STAR Market listing is partly due to recent weak share performance of AI firms listed in Hong Kong, plus the large number of companies waiting to list on HKEX. A dual listing would add an extra financing channel and expand its investor reach. The STAR Market relaxed its fifth set of listing rules specifically for large language model (LLM) companies in June this year, allowing firms that have not yet achieved a certain revenue scale to apply. One key requirement is that the company must have at least one LLM product launched and realize large-scale commercial application.

Renowned trader Killa stated in a social media post that Bitcoin has been consolidating sideways at the bottom for two months before rallying 27%, yet the market remains in disbelief. There is no inherent reason for the gap below to be fully filled, he explained. The gap initially formed due to the liquidation of roughly $6 billion in short positions—this is only the publicly visible liquidation data in the market. Against this backdrop, the gap does not need to be fully filled, and it is highly unlikely that it will be. During the rally at the end of 2022, only partial gap fills were observed at most, and even those were quickly bought back. If a similar situation plays out this time, BTC could test the $70,000 level, though this is not a certainty. Technically, Killa does not believe Bitcoin will return to his entry point for his 2x long position at $62,600, nor even to the average spot price he publicly shared of $65,800. The worst-case re-test scenario would be slightly below $70,000, potentially around $69,000, if some form of market capitulation occurs. No significantly deeper pullback is expected beyond that, and even that scenario would be a stretch. BTC is very likely to easily defend the $73,000–$75,000 range before ultimately rallying further toward $85,000.

According to Bloomberg, prediction market platform Polymarket has appointed its first Chief Financial Officer (CFO), Warren Jenson, a move viewed as a key step to catch up with rival Kalshi as the company pursues a new round of investment and seeks to rebuild growth momentum. Jenson officially joined Polymarket this week, bringing prior senior finance roles at media research firm Nielsen and martech company LiveRamp. The 69-year-old Jenson boasts decades of financial management experience, while Polymarket’s founder and CEO Shayne Coplan is just 28 years old. Bloomberg notes that Polymarket has recently fallen behind in competition with its main rival Kalshi, prompting the firm to seek new funding and revitalize its development momentum. The hiring of its first CFO signals Polymarket is further strengthening its financial management system in preparation for subsequent financing and business expansion. Polymarket has already secured large-scale financing in the past: in early September, 1789 Capital, a firm where Donald Trump Jr. serves as a partner, led its roughly $1 billion funding round, which valued Polymarket at around $21 billion upon deal completion. Notably, Polymarket has continued expanding its business recently, launching a social feature called Squads on September 8 that integrates group chat functionality into its prediction market platform.

According to an official statement from Bitget, the crypto exchange has completed a full upgrade to its U.S. stock market data feed, officially integrating Nasdaq’s official data sources, making it the first crypto trading platform in the industry to adopt this data standard. Compared to its previous data model, Nasdaq’s official data delivers more accurate and timely market data to users, while maintaining better price stability and continuity amid sharp market volatility. To further ensure data quality, after integrating the official data, Bitget has built a multi-layered safeguard mechanism based on official metadata, covering data processing, real-time monitoring, quality comparison, simulation verification, and abnormal data filtering. As a core data service provider long relied on by global financial markets, Nasdaq’s data standards have been validated over years in the traditional finance sector. This upgrade also marks Bitget’s further advancement in building its global asset trading infrastructure. As its product offerings expand from digital assets to global assets including U.S. stocks, the platform will continue to enhance its underlying data, liquidity, and risk management capabilities, while introducing more mature institutional-grade market standards. It aims to deliver a more stable and efficient cross-market trading experience for both retail and institutional clients, and further strengthen the overall competitiveness of its panoramic trading platform (UEX).