Retail investors purchased a whopping $91 billion worth of US stocks in a single week, setting a new nearly 15-year record, with AI and semiconductors leading capital inflows.

Two people familiar with the matter revealed that Coins.ph, the Philippines’ largest cryptocurrency exchange, is facing regulatory penalties from the country’s central bank, reportedly for providing services to restricted offshore gambling and betting firms. One source noted the violations are severe enough to potentially lead to the exchange’s operating license being revoked. Coins.ph has over 18 million users, with a daily spot trading volume of approximately $50 million, and controls around 75% to 80% of Philippine peso (PHP) stablecoin liquidity in the country. Earlier this month, the Philippine central bank allegedly restricted deposits to Coins.ph via its main wallet operator DCPay Philippines. Prior to this, the central bank had ordered regulated institutions to cut off e-wallet app access to gambling websites; some gambling firms subsequently turned to Coins.ph, using cryptocurrencies and stablecoins to make payments to overseas partners. The report also pointed out that multiple compliance department executives have recently left Coins.ph, including its chief compliance officer and head of anti-money laundering monitoring.

Metaplanet CEO Simon Gerovich announced in a post that the company will issue zero-coupon yen-denominated convertible bonds totaling 52 billion yen, maturing in 2031, while simultaneously advancing its share repurchase program. Of the proceeds from the bond issuance, up to 28.3 billion yen will be allocated to share buybacks: up to 25.3 billion yen will be used for over-the-counter (OTC) block trades covering up to 96,934,800 shares, or roughly 7.1% of the company’s outstanding issued shares; the remaining amount will be acquired via open market purchases. Separately, the company will use approximately 23.6 billion yen to partially repay its credit line, unlocking the Bitcoin currently held as collateral for the facility. Gerovich explained that Metaplanet’s current share price trades below the value of its Bitcoin holdings, so the repurchase is designed to create shareholder value by leveraging this discount. The convertible bonds’ initial conversion price is set above both the bond’s issuance price and 1.0 times its market net asset value (mNAV). Though the issuance will increase the company’s total debt, Metaplanet intends to replace some secured, interest-bearing loans with unsecured debt that carries no periodic interest payments. This move is expected to lower interest costs, extend debt tenors, and boost financial flexibility for future Bitcoin acquisitions.

According to Bitget market data, South Korea’s semiconductor stock market is rebounding. The KOSPI Composite Index is currently up 1.52%, with SK Hynix rising 3.47% and Samsung Electronics gaining 3.66%. Prior to this, South Korea’s stock market had seen five consecutive trading days of sell-offs, with semiconductor stocks experiencing particularly sharp declines. Market concerns first stem from China’s growing competitiveness. ChangXin Memory Technologies recently disclosed that it has made key progress on a memory chip architecture that Samsung Electronics and SK Hynix are also developing. While this technological breakthrough is still far from translating into large-scale, commercial competitive production capacity, investors have begun reassessing how the catch-up of Chinese storage manufacturers will impact the future pricing power and valuation premiums of South Korea’s two leading chip firms. Marginal shifts in AI investment logic have also amplified the downturn. Broadcom CEO Hock Tan noted that demand for computing power infrastructure still exceeds the industry’s current supply, and the supply-demand gap may persist until 2029 or even 2030. As hardware bottlenecks such as chips, memory, and power gradually ease, access to capital could become the biggest constraint on AI expansion. Many AI investors do not clearly understand the business models they are betting on, yet set extreme expectations, ultimately making this round of investment "almost like being in a casino".

According to HTX market data, GNS has surged over 118% in the past 24 hours, currently trading at $0.78 with a market cap of $17.33 million. GNS is the native token of decentralized leverage trading platform Gains Network. Earlier on September 4, Binance announced it would add a watchlist tag for Gains Network (GNS).

According to monitoring by EmberCN, a whale or institution trading ETH via Galaxy Digital's OTC desk sold its final 10,000 ETH today, worth roughly $25 million, completing its full liquidation. On-chain data shows the address bought 52,000 ETH at an average price of $2,161 between July and September, and has since late September sold the holdings in tranches at an average price of $2,632, generating a total profit of around $24.49 million over two and a half months.

Insight Beating AI Flash News: Google has officially rolled out Gemini 4 Argon to all global paid users, after the model was previously only available to select cybersecurity partners. The most eye-catching demo features a robot use case: Argon watches a video of a human throwing sandbags at a target, identifies the field location, calculates full-body joint movements, and writes its own physics simulation program. After simulation and real-device migration, Apollo, the humanoid robot from U.S. robotics firm Apptronik, successfully hits the target 7 times in a row on the actual field. Google notes this workflow cuts manual simulation modeling—originally taking weeks—down to just hours. In other demos, Argon built a simulation environment for Boston Dynamics’ Spot robot dog, letting it practice opening trash can lids and tossing in cans. It can also generate a simulation scene from a single photo of a desk, synthesizing over 4,500 training data segments. After fine-tuning the robot action model with this data, the real-device task success rate rose from 56% to 75%, while time per task dropped from 35 seconds to 24 seconds. In the ERQA benchmark, which tests 3D spatial and physical understanding capabilities, Argon scored 87%, outperforming other leading models. Currently, Google AI Pro and Ultra users can access Argon in the Gemini app; access in Antigravity is limited to Ultra users only, and developers can also call it via a paid API. The API launch price is $2 per million input tokens and $10 per million output tokens, on par with Claude Sonnet 5.5 and GPT-6.1 Sol, and just one-fifth the price of Claude Fable 5.1 and GPT-6 Astra. However, this discount is only valid until November 30, after which prices will double, matching Claude Opus 5.5’s rates.