Goldman Sachs Q1 Report: Liquidates XRP, SOL ETF, Contrarian Increase in Crypto Stocks

German solar equipment manufacturer SINGULUS TECHNOLOGIES announced on November 10 that it has signed a contract with SpaceX to supply a complete heterojunction (HJT) solar cell production line. The first order is valued at over €60 million, with delivery scheduled for the first half of 2027. The two sides’ strategic cooperation also plans to expand to multiple production lines, gradually enabling a solar cell production equipment capacity of over 100GW. The facilities will be used for SpaceX’s satellite manufacturing project in Bastrop, Texas, U.S., to supply high-performance solar cells for its Starmind space AI satellites. Musk subsequently reposted SpaceX’s announcement, noting that SpaceX has ordered 100GW annual capacity HJT solar manufacturing equipment from SINGULUS, stating this is to let Starmind create a "sentient sun" — a phrase echoing his long-term vision of building space AI computing infrastructure via solar satellites. SINGULUS also disclosed that since SpaceX launched its strategic partnership in March 2026, it has provided financial support to the firm in March and October, with the October support including a secured subordinated loan.

Beating AI News Flash: Alma, developed by independent developer Yetone, ranked third in both task pass rate and success cost in the latest round of AI Agent cross-evaluation, outperforming well-known tools like Claude Code and OpenCode. The evaluation was initiated by AI Agent infrastructure firm Runta. According to the current ranking snapshot, Alma’s task pass rate is 72.7%, second only to Codex CLI’s 77.9% and Pi’s 72.8%. Claude Code’s pass rate stands at 70.0%. Alma not only boasts a higher pass rate but also delivers tasks at a lower cost: it averages around $0.80 per successful task, compared to $0.94 for Claude Code, a roughly 16% cost reduction. This FrontierHarness v1.1 evaluation centers on a key question: How much can performance differ when using the same large language model (LLM) with distinct Agent execution frameworks? The assessment covers 21 Harness configurations, 5 LLMs, and 62 tasks, with each task repeated 4 times for a total of 26,040 planned runs. All tools operate in a unified cloud environment, with comparisons across task success rate, token cost, and execution speed. Alma is an AI Agent designed for local operation and long-term memory, capable of directly reading and modifying code, executing terminal commands, web searching, and supporting Skills and MCP. It can freely switch underlying models, with no dependency on a single model vendor. Recently, Alma has integrated Herdr: programming Agents like Claude Code, Codex, and Pi can continue running on Herdr, while Alma centralizes them into a single chat interface to view execution processes, respond to queries, and approve operations. Users can even remotely manage multiple Agents on their computer via iPhone. The evaluation is still ongoing, so rankings are subject to change. Only open-source models including GLM, DeepSeek, Kimi, and Muse were tested, with no closed-source models such as GPT and Claude series included.

South Korean crypto trader @solyfw has publicly accused exchange LBank on X of freezing roughly $1 million in his funds, forcing the liquidation of his futures positions. The trader stated the issue arose after he earned profits from high-leverage futures trades, with his account restricted, and LBank has yet to provide a specific explanation. LBank responded that it does not restrict accounts solely due to user profits, adding the account is under review, and the exchange has hired external legal counsel to address "false statements". According to the trader, starting August 31, he generated significant returns trading high-leverage futures on major altcoins including XRP, DOGE, XLM, NEAR, FIL, SUI and other highly liquid mainstream assets. On September 19, while attempting a standard withdrawal, his account was suddenly restricted; the platform requested additional identity verification, which the user completed. On September 29, LBank further limited his futures trading and demanded more detailed clarifications. On October 14, the exchange finally replied that "the relevant abnormal funds will not be released". In its official response the same day, LBank said: "We will not restrict accounts just because someone is making profits. Trading and withdrawals are currently operating normally. We are communicating directly with the user regarding the account review and will share information as much as possible while protecting his privacy." The exchange also stated it has hired external legal counsel to take action against "false and misleading statements" and warned of potential legal measures against related remarks.

According to an official announcement, Huobi HTX will open NVDA/USDT spot and grid trading at 17:00 (UTC+8) on November 11, and will launch SKHY/USDT spot and grid trading at 19:00 the same day. Recently, Huobi HTX announced the listing of 20 stock tokens and ETF-related assets including SNDK, SOXL, SPCX, CRCL, MSTR, SKHY, MU, TSLA, NVDA, etc., and launched a series of promotional activities such as 0 trading fees for executed orders.

According to TradingBeats monitoring, a crypto whale opened a new NVDA long position with 5x leverage during NVIDIA’s early-morning price drop today, totaling around $8.4243 million at an average entry price of $168.49. As of press time, the address has not reduced its position: its holding value stands at approximately $8.6385 million, with an unrealized profit of roughly $214,200, a return rate of about 12.71%, and a liquidation price of $135.

According to sources, NVIDIA CEO Jensen Huang walked into the company's headquarters in Santa Clara, California on Monday and chaired his usual morning meeting with senior executives. Sources said Huang opened the closed-door meeting by apologizing to employees over the recent controversy. He took no questions and quickly shifted the topic to business and the day's main agenda. Huang also came to the office for a meeting on Tuesday. NVIDIA is set to release its new quarterly earnings report in one week. An NVIDIA insider said Huang is trying to create a "business as usual" atmosphere, but internal morale at the company is "low". (CNBC)