News Heat Trend
Project Overview
Cursor is an AI-driven code editor enhancing developer productivity. It offers features like code generation, debugging, and collaboration.
Kalshi Crypto Business Lead: The "Top-Tier VC" Narrative Around SBF Is Inaccurate; The Real Mastermind Is "AI Stock God" Leopold
John Wang, Head of Crypto Business at prediction market platform Kalshi, stated on X that it is widely believed Sam Bankman-Fried (SBF) was a "top-tier venture capitalist" who successfully invested in star projects like Anthropic and Cursor. However, Wang argued this narrative is inaccurate. The real "core figure" driving these investment strategies and early resource allocation was actually "AI stock guru" Leopold Aschenbrenner, not SBF himself.Analysis suggests that this remark has sparked discussion within the crypto and venture capital circles, once again bringing the attribution of SBF's influence on early-stage investments in Silicon Valley and the crypto industry into the spotlight. It is reported that the AI fund Situational Awareness, founded by former OpenAI researcher Leopold Aschenbrenner, has grown to over $20 billion in scale, with quantitative giant Jane Street making a rare capital injection. Situational Awareness has achieved a year-to-date return rate of 270% and cumulative returns exceeding 1,000% since its inception. Equity bets on Anthropic have contributed the most successful returns, accounting for one-fifth of its assets. Beyond public markets, Situational Awareness also co-led an investment round in AI chip company MatX with Jane Street and participated in the latest funding round of AI cloud computing provider Fluidstack.
Forbes: SBF’s venture capital prowess is being reassessed—if he hadn’t been imprisoned, he might have commanded a $100 billion fortune
According to Forbes, SBF’s early investment portfolio is undergoing renewed market scrutiny; had he not been imprisoned due to the FTX collapse, his venture capital returns could theoretically have generated wealth growth reaching approximately $100 billion. Prior to FTX’s collapse, SBF had built an investment portfolio spanning several high-profile companies—including Anthropic, SpaceX, Robinhood, and Cursor—with Cursor valued at $60 billion and Anthropic nearing a $90 billion valuation. Rory O’Driscoll, Partner at Scale Venture Partners, noted that SBF had backed multiple pivotal AI companies even before the ChatGPT era, demonstrating an exceptionally rare investment acumen.
Microsoft had considered acquiring Cursor but ultimately locked in a partnership with SpaceX.
According to Techstartups, Microsoft had explored acquiring AI programming tools company Cursor but ultimately did not proceed with the deal. Subsequently, SpaceX swiftly secured an option to acquire Cursor at a $60 billion valuation. Cursor has now become one of the key players in the AI programming space, benefiting from strong developer demand for automated programming and productivity tools—where OpenAI and Anthropic are fiercely competing. Meanwhile, Microsoft faces another set of pressures, with its stock down 10% this year, underperforming its peers in the hyperscale data center sector.
SBF invested $200,000 early in Cursor; if he had not sold his equity, it would now be worth approximately $3 billion.
SpaceX announced today that it has reached a cooperation agreement with AI-powered programming startup Cursor, which has granted SpaceX the right to acquire Cursor later this year for $6 billion—or pay $1 billion to advance their collaboration. According to reports, Cursor’s developer, Anysphere, closed a $400,000 pre-seed funding round in April 2022, co-led by Alameda Research and Heroic Ventures. Alameda invested $200,000 in Anysphere, acquiring approximately 5% equity; this stake was sold at its original price during FTX’s bankruptcy liquidation. Had it not been sold, the stake would now be worth roughly $3 billion.
AI-powered programming startup Cursor seeks at least $2 billion in funding, with a valuation of $50 billion
According to TechCrunch, AI-powered coding startup Cursor is nearing completion of a new funding round of at least $2 billion, with a post-money valuation of approximately $50 billion—nearly doubling its $29.3 billion valuation from six months ago. The round is co-led by existing investors Thrive Capital and Andreessen Horowitz, with Battery Ventures and strategic investor NVIDIA also expected to participate. The round has already been oversubscribed, though final terms have not yet been finalized. On the performance front, Cursor projects its annualized revenue to exceed $6 billion by the end of 2026—representing at least a threefold increase over the $2 billion annualized revenue it disclosed in February this year. Regarding profitability, the company achieved a slight positive gross margin overall after launching its in-house Composer model in November last year and incorporating lower-cost third-party models (e.g., Kimi from China). Its enterprise business has already reached gross-margin profitability, while its individual developer accounts remain unprofitable.
Coinbase internal tool Mux reveals AI coding paradigm shift: Engineers transition from "code writers" to "multi-agent orchestrators"
Coinbase, a cryptocurrency trading platform, has disclosed in a technical sharing session that its internal multi-agent development tool "Mux" is reshaping software engineering workflows, transitioning the engineer's role from traditional code implementers to task orchestrators for AI agents.With the widespread internal adoption of AI programming tools such as Cursor, Copilot, OpenCode, and Claude Code, code generation efficiency has significantly improved. However, development workflows have long remained stuck in a traditional "single-task, single-branch, sequential execution" mode, creating a new collaboration bottleneck.Mux was born as an internal tool against this backdrop. By assigning each AI agent an independent git worktree, branch, and terminal environment, the system enables parallel multi-task development and conflict-free collaboration, allowing engineers to simultaneously direct multiple agents to handle tasks such as API development, test writing, vulnerability fixes, and code refactoring.Data shows that as of April 2026, Mux has covered over 600 users within Coinbase (including engineers, product managers, and designers), with 335 actively using it and 197 being high-frequency users. It has facilitated over 5,000 PR merges across 461 code repositories and 10 organizations. Engineers using Mux achieved an average of 39.6 PR merges, approximately 3.5 times the baseline of 11.4.Coinbase stated that Mux's success relies on its internal infrastructure capabilities, including an LLM Gateway, secure model access, and a code flow deployment system, enabling deep integration of multi-agent tools into real development workflows. This trend marks a structural shift in the software engineering paradigm: as AI reduces the cost of code generation, the core value of engineers is transitioning from "implementation capability" to "problem definition and agent orchestration capability."
North Korean hacking group “HexagonalRodent” leverages AI to industrialize attacks against Web3 developers, stealing over $12 million in crypto assets in three months
According to a research report released by cybersecurity firm Expel, the company is tracking an advanced persistent threat (APT) group dubbed “HexagonalRodent,” which is highly assessed to be a North Korean (DPRK) state-sponsored actor. This group primarily targets Web3 developers and specializes in stealing high-value digital assets—including cryptocurrencies and NFTs. In the first quarter of 2026 alone, the group compromised 2,726 developer devices and stole access credentials for 26,584 cryptocurrency wallets, with the total value of stolen assets reaching as high as $12 million. The group primarily carries out its attacks via fake job postings—publishing lucrative positions on LinkedIn and Web3 recruitment platforms to lure job seekers into completing “skills assessments” embedded with malicious code. These assessments exploit VSCode’s tasks.json functionality to automatically execute malware when victims open the project folder. The malware used includes BeaverTail, OtterCookie, and InvisibleFerret, all of which possess capabilities such as password theft, remote control, and reverse shell execution. Notably, the group extensively leverages generative AI tools—including ChatGPT and Cursor—to develop malware, build counterfeit corporate websites, and generate AI-forged executive teams. It even registered a shell company in Mexico to enhance the credibility of its operations. Additionally, the group recently carried out its first-ever supply-chain attack, successfully infiltrating a VSCode extension.
Robinhood Launches AI Agent for Stock Trading and Automated Spending Features
: Robinhood has announced the launch of an AI Agent feature, allowing users to delegate stock trading and credit card spending decisions to AI agents. According to the introduction, users can connect AI tools such as Anthropic Claude and Cursor to their designated Robinhood accounts, enabling the AI to automatically execute stock trades, portfolio management, and spending tasks, while users retain control through spending limits, approval mechanisms, and real-time notifications.In payment scenarios, the AI Agent can use the Robinhood Gold virtual credit card to automatically search for deals, book travel, and purchase tickets. Robinhood stated that this feature currently only supports stock trading, with plans to expand to options and cryptocurrency trading in the future.
Base launches AI tool Base MCP, enabling users to manage crypto wallets and interact with DeFi applications via ChatGPT.
According to CoinDesk, Base—the Ethereum Layer 2 network operated by Coinbase—has launched the AI-integration tool Base MCP, which connects users’ Base accounts to AI clients such as ChatGPT, Claude, and Cursor, enabling them to perform transactions—including fund transfers, token swaps, balance checks, transaction history reviews, and interactions with DeFi applications—using natural language.
Base Launches MCP Gateway, Enabling On-Chain Application Operations via AI Interfaces like ChatGPT and Claude
: Base, the Ethereum scaling network incubated by Coinbase, has launched Base MCP, aiming to connect AI interfaces with Base wallet infrastructure.Through this integration, users can use natural language commands in AI tools that support the MCP open standard to complete token swaps, fund transfers, and interact with Base ecosystem applications. Supported AI interfaces include Claude, ChatGPT, Cursor, and others.Base MCP will serve as a secure gateway between user Base accounts and AI interfaces. At its initial launch, it will connect with applications such as Morpho, Bankr, Moonwell, Avantis, Aerodrome, Virtuals, and Uniswap, covering scenarios including lending, swaps, perpetual contracts, new tokens, and AI Agent issuance.MCP stands for Model Context Protocol, originally proposed by Anthropic, and is an open standard for connecting AI models with external tools and data sources.
Musk: Grok Base Model V9-Medium (1.5T) training completed with good evaluation results
Musk said Grok Base Model V9-Medium (1.5T) training has been completed with good evaluation results. A large amount of Cursor data has been added during supplementary training, and more will be added subsequently. Fine-tuning is currently underway, and reinforcement learning is expected to begin in a few days. The official release is anticipated in 2 to 3 weeks.
Musk: Grok V9 and V8 Have a Massive Gap; V9 Training Version Already Shows Superior Performance
Elon Musk posted on X, stating that the latest completed training run of Grok V9 (1.5T parameters) has "performed very well," and this result has not yet incorporated the supplementary training portion from Cursor data. The base model currently under internal development is V9, with approximately 1.5 trillion parameters. Compared to V8, it features significant improvements in data cleaning, training methods, model scale, and has been optimized for the Blackwell architecture to enhance computational efficiency.Musk emphasized that, in contrast, the current public-facing version v4.2, built on the V8 base model with approximately 0.5T parameters and running on the Hopper architecture, still has certain limitations in training data quality and coverage. The performance gap between Grok V8 and V9 is massive, with the new-generation model achieving a leapfrog upgrade in overall capabilities.
Coinbase internal tool Mux reveals AI coding paradigm shift: Engineers transition from "code writers" to "multi-agent orchestrators"
Coinbase, a cryptocurrency trading platform, has disclosed in a technical sharing session that its internal multi-agent development tool "Mux" is reshaping software engineering workflows, transitioning the engineer's role from traditional code implementers to task orchestrators for AI agents.With the widespread internal adoption of AI programming tools such as Cursor, Copilot, OpenCode, and Claude Code, code generation efficiency has significantly improved. However, development workflows have long remained stuck in a traditional "single-task, single-branch, sequential execution" mode, creating a new collaboration bottleneck.Mux was born as an internal tool against this backdrop. By assigning each AI agent an independent git worktree, branch, and terminal environment, the system enables parallel multi-task development and conflict-free collaboration, allowing engineers to simultaneously direct multiple agents to handle tasks such as API development, test writing, vulnerability fixes, and code refactoring.Data shows that as of April 2026, Mux has covered over 600 users within Coinbase (including engineers, product managers, and designers), with 335 actively using it and 197 being high-frequency users. It has facilitated over 5,000 PR merges across 461 code repositories and 10 organizations. Engineers using Mux achieved an average of 39.6 PR merges, approximately 3.5 times the baseline of 11.4.Coinbase stated that Mux's success relies on its internal infrastructure capabilities, including an LLM Gateway, secure model access, and a code flow deployment system, enabling deep integration of multi-agent tools into real development workflows. This trend marks a structural shift in the software engineering paradigm: as AI reduces the cost of code generation, the core value of engineers is transitioning from "implementation capability" to "problem definition and agent orchestration capability."
Related news
Kalshi Crypto Business Lead: The "Top-Tier VC" Narrative Around SBF Is Inaccurate; The Real Mastermind Is "AI Stock God" Leopold
John Wang, Head of Crypto Business at prediction market platform Kalshi, stated on X that it is widely believed Sam Bankman-Fried (SBF) was a "top-tier venture capitalist" who successfully invested in star projects like Anthropic and Cursor. However, Wang argued this narrative is inaccurate. The real "core figure" driving these investment strategies and early resource allocation was actually "AI stock guru" Leopold Aschenbrenner, not SBF himself.Analysis suggests that this remark has sparked discussion within the crypto and venture capital circles, once again bringing the attribution of SBF's influence on early-stage investments in Silicon Valley and the crypto industry into the spotlight. It is reported that the AI fund Situational Awareness, founded by former OpenAI researcher Leopold Aschenbrenner, has grown to over $20 billion in scale, with quantitative giant Jane Street making a rare capital injection. Situational Awareness has achieved a year-to-date return rate of 270% and cumulative returns exceeding 1,000% since its inception. Equity bets on Anthropic have contributed the most successful returns, accounting for one-fifth of its assets. Beyond public markets, Situational Awareness also co-led an investment round in AI chip company MatX with Jane Street and participated in the latest funding round of AI cloud computing provider Fluidstack.
Robinhood Launches AI Agent for Stock Trading and Automated Spending Features
: Robinhood has announced the launch of an AI Agent feature, allowing users to delegate stock trading and credit card spending decisions to AI agents. According to the introduction, users can connect AI tools such as Anthropic Claude and Cursor to their designated Robinhood accounts, enabling the AI to automatically execute stock trades, portfolio management, and spending tasks, while users retain control through spending limits, approval mechanisms, and real-time notifications.In payment scenarios, the AI Agent can use the Robinhood Gold virtual credit card to automatically search for deals, book travel, and purchase tickets. Robinhood stated that this feature currently only supports stock trading, with plans to expand to options and cryptocurrency trading in the future.
Base launches AI tool Base MCP, enabling users to manage crypto wallets and interact with DeFi applications via ChatGPT.
According to CoinDesk, Base—the Ethereum Layer 2 network operated by Coinbase—has launched the AI-integration tool Base MCP, which connects users’ Base accounts to AI clients such as ChatGPT, Claude, and Cursor, enabling them to perform transactions—including fund transfers, token swaps, balance checks, transaction history reviews, and interactions with DeFi applications—using natural language.
Base Launches MCP Gateway, Enabling On-Chain Application Operations via AI Interfaces like ChatGPT and Claude
: Base, the Ethereum scaling network incubated by Coinbase, has launched Base MCP, aiming to connect AI interfaces with Base wallet infrastructure.Through this integration, users can use natural language commands in AI tools that support the MCP open standard to complete token swaps, fund transfers, and interact with Base ecosystem applications. Supported AI interfaces include Claude, ChatGPT, Cursor, and others.Base MCP will serve as a secure gateway between user Base accounts and AI interfaces. At its initial launch, it will connect with applications such as Morpho, Bankr, Moonwell, Avantis, Aerodrome, Virtuals, and Uniswap, covering scenarios including lending, swaps, perpetual contracts, new tokens, and AI Agent issuance.MCP stands for Model Context Protocol, originally proposed by Anthropic, and is an open standard for connecting AI models with external tools and data sources.
Musk: Grok Base Model V9-Medium (1.5T) training completed with good evaluation results
Musk said Grok Base Model V9-Medium (1.5T) training has been completed with good evaluation results. A large amount of Cursor data has been added during supplementary training, and more will be added subsequently. Fine-tuning is currently underway, and reinforcement learning is expected to begin in a few days. The official release is anticipated in 2 to 3 weeks.
Musk: Grok V9 and V8 Have a Massive Gap; V9 Training Version Already Shows Superior Performance
Elon Musk posted on X, stating that the latest completed training run of Grok V9 (1.5T parameters) has "performed very well," and this result has not yet incorporated the supplementary training portion from Cursor data. The base model currently under internal development is V9, with approximately 1.5 trillion parameters. Compared to V8, it features significant improvements in data cleaning, training methods, model scale, and has been optimized for the Blackwell architecture to enhance computational efficiency.Musk emphasized that, in contrast, the current public-facing version v4.2, built on the V8 base model with approximately 0.5T parameters and running on the Hopper architecture, still has certain limitations in training data quality and coverage. The performance gap between Grok V8 and V9 is massive, with the new-generation model achieving a leapfrog upgrade in overall capabilities.