Thursday, July 9, 2026

Micron shares rise 7% after announcing billions more in U.S. chipmaking investments

Micron Technology's stock rose 7% following the announcement of a new $3 billion investment aimed at enhancing the U.S. semiconductor supply chain. This investment includes a significant $500 million allocation to Taiwanese company GlobalWafers for expanding its wafer development and manufacturing operations in Texas, along with a 10-year supply agreement for raw silicon wafers. Additionally, Micron plans to raise its total U.S. investment to $250 billion by 2035, reflecting a $50 billion increase as demand for memory chips, particularly driven by artificial intelligence, continues to grow.

Meet SK Hynix, the trillion-dollar South Korean chipmaker debuting on U.S. markets

SK Hynix, South Korea's second most valuable company, is set to debut on the Nasdaq with a market capitalization of approximately $1 trillion, following a stock price increase of over sevenfold in the past year. The listing coincides with the company's ambitious plans to expand in the U.S., including a $4 billion production facility in Indiana, which is expected to be completed by 2028. As a leading manufacturer of high-performance memory chips, SK Hynix is positioned to benefit from the surging demand driven by artificial intelligence, particularly through its partnership with Nvidia, the largest buyer of high-bandwidth memory.

OpenAI's newest AI model is 54% more token efficient on agentic coding, Altman tells CNBC

OpenAI's CEO Sam Altman revealed that the new GPT-5.6 Sol model is 54% more token efficient on agentic coding tasks compared to previous versions. This model, along with Terra and Luna, is being rolled out broadly after an initial limited launch to a select group of trusted partners, as requested by the U.S. government. Altman noted that OpenAI collaborated closely with senior officials, including Commerce Secretary Howard Lutnick, to ensure safety and compliance prior to the deployment. He emphasized the importance of demonstrating safety claims to gain broad access to powerful AI models.

New AI Model Puts SpaceX Back In Race Versus Anthropic, OpenAI, Google

SpaceX's xAI has launched its latest artificial intelligence model, Grok 4.5, which analysts claim has closed the performance gap with industry leaders Alphabet, Anthropic, and OpenAI. This development comes amid skepticism regarding xAI's revenue potential compared to other sectors of SpaceX. The introduction of Grok 4.5 signifies a strategic move by Elon Musk's company to enhance its competitive standing in the rapidly evolving AI landscape.

Anthropic’s new Claude feature is quietly selling you on AI

Anthropic has introduced a new feature called Reflect for its AI tool Claude, aimed at enhancing user engagement amid growing AI skepticism. Reflect provides users with an analytics dashboard that tracks their interactions with Claude, offering insights into usage patterns and topics discussed. This feature not only positions Claude as an integral productivity tool but also encourages users to reflect on their AI habits through prompts like, 'What’s one thing you want to keep doing yourself?' Currently in beta for Free, Pro, and Max users with memory enabled, Reflect is designed to deepen user integration with Claude, potentially reducing churn and discouraging shifts to competing AI tools.

Anthropic, OpenAI, and SpaceX are bigger than the last 25 years of tech exits

The recent NCVA-Pitchbook Venture Monitor report highlights the unprecedented scale of upcoming IPOs from SpaceX, Anthropic, and OpenAI, predicting their combined value will surpass all U.S. VC-backed exits since 2000. SpaceX has already launched at a staggering $1.77 trillion valuation, while Anthropic and OpenAI are also expected to reach trillion-dollar valuations. In stark contrast, the U.S. Securities and Exchange Commission reported just $70 billion in IPO proceeds last year, underscoring the magnitude of these impending exits and the shift in the tech landscape.

Trump can halt trade with Spain using law behind scrapped tariffs: Greer

President Donald Trump has the authority to halt trade with Spain under the International Emergency Economic Powers Act (IEEPA), as confirmed by U.S. Trade Representative Jamieson Greer. This comes after Trump expressed frustration over Spain's reluctance to increase NATO defense spending, declaring he was done dealing with the country. However, following a positive meeting with Spanish Prime Minister Pedro Sánchez, Greer indicated that Trump might reconsider this drastic step. The potential trade severance could impact key Spanish exports to the U.S., including refined petroleum and pharmaceuticals.

Tanker traffic through Strait of Hormuz slows after Iranian attacks trigger renewed fighting with U.S.

Tanker traffic through the Strait of Hormuz has sharply declined following Iranian attacks on three commercial vessels, with only 13 tankers transiting the strait on Wednesday compared to an average of 33 per day over the previous week, according to trade intelligence firm Kpler. The situation has heightened tensions, leading to renewed fighting between the U.S. and Iran, including two rounds of U.S. airstrikes in retaliation. Despite these developments, the oil market is not anticipating a complete closure of the strait, as analysts suggest a new normal of intermittent conflict and relative calm that allows for tanker transit.

Meta says its new AI model is ready to compete on coding

Meta has introduced Muse Spark 1.1, an advanced AI model designed for coding, now accessible to US developers through a public API preview. This model represents a significant upgrade from its predecessor, featuring enhanced capabilities such as complex bug detection and fixing, support for end-to-end workflows, and native multimodal perception across various media types. The launch comes shortly after the controversial Muse Image model and is part of Meta's strategy to compete with AI leaders like OpenAI and Google, following substantial investments in AI development.

Character.AI wants a piece of the microdrama pie

Character.AI has expanded its offerings with the launch of c.ai Series, a new platform for interactive, AI-generated microdramas. This initiative features three series: Last Summer, which explores secret admirers; The Nighttime Game, centered on a deadly card game; and Eden Fall, which follows elite MMO players in a virtual reality setting. Each series consists of 10 episodes, each under two minutes long, with the first eight available for free and the final two behind a paywall. Character.AI aims to tap into the projected $26 billion microdrama industry while allowing viewers to interact with characters post-episode.

FL Studio 2026 turns its AI chatbot into your assistant engineer

FL Studio 2026 has upgraded its AI chatbot, Gopher, transforming it from a simple instruction manual into an interactive assistant capable of executing music production tasks. Users can now ask Gopher to perform specific actions, such as laying down beats or adding effects, which it executes accurately. However, Gopher has limitations; it cannot create automation, insert notes, or select specific presets. Alongside Gopher, FL Studio 2026 features a completely rebuilt Flex virtual instrument, offering a more efficient sound engine and enhanced preset browsing. The update also includes automatic cloud backups for FL Cloud subscribers and an audio logger that captures the last 60 seconds of output.

Popular open source AI developer tool Ollama raises $65M, grows to nearly 9M users

Ollama, an open-source AI developer tool, has raised $65 million in Series B funding led by Theory Ventures, bringing its total funding to $88 million since its launch in 2023. The tool enables developers to run open-weight AI models on their PCs quickly and has gained popularity, now serving over 8.9 million developers each month, including 85% of Fortune 500 companies. Ollama offers subscription tiers for accessing larger models, with prices ranging from free to $100/month, while tracking usage based on GPU time instead of token limits.

OpenAI finds roughly 30 percent of popular AI coding test is broken

OpenAI has retracted its endorsement of the SWE-Bench Pro coding test after discovering that approximately 30% of its tasks are flawed. The issues arise from tasks derived from real software projects, which can be overly strict, vague, or misleading for AI models. OpenAI's review utilized an automated screening tool that flagged 286 tasks, with human evaluators confirming 200 as flawed. This misrepresentation can significantly impact AI model assessments and safety evaluations under OpenAI's Preparedness Framework. The company is now advocating for the development of new, reliable benchmarks that are meaningful and difficult to manipulate.

The Market Prevails

Argus has updated its Focus List, introducing three new stock recommendations: Delta Air Lines Inc (DAL), Nucor Corp (NUE), and Charles Schwab Corp. This list features 30 top stocks identified by Argus analysts and is regularly revised to provide investors with the best ideas. Additionally, the monthly investment strategy column offers insights into economic forecasts and asset allocation models, enhancing the value of the portfolio selector for investors.

A Majority of European Lawmakers Voted Against Letting Big Tech Read Our Messages. They’re Going to Anyway

The European Parliament has voted to extend legislation permitting tech companies to voluntarily scan users' private messages for child sexual abuse material, despite a majority of lawmakers opposing the proposal. This ruling reinstates permissions for companies like Meta, Google, and Microsoft to scan private communications, while end-to-end encrypted chats on platforms such as WhatsApp and Signal remain exempt. The European People's Party, the largest political group in the Parliament, argued that such scanning is necessary to protect children, especially after the expiration of a prior law in April. Although more Members of the European Parliament voted against the regulation than in favor, they fell short of the required majority by 47 votes, allowing the scanning to continue until 2028 or until new legislation is enacted.

Apple TV shares first trailer for ‘There’s No Place Like Home, Snoopy’

Apple TV has unveiled the trailer for ‘Snoopy Presents: There’s No Place Like Home, Snoopy,’ set to debut on July 31, 2026. In this new special, Snoopy embarks on a quest to find his beloved doghouse after it is mistakenly sold at a yard sale. Accompanied by Charlie Brown, the duo explores the meaning of home. The special features a voice cast that includes Riley Vargas and Terry McGurrin, along with an original song by Allen Stone titled ‘Home, Where Your Heart Found Me.’ This release continues Apple’s collaboration with Peanuts, which has established Apple TV as the exclusive streaming platform for Peanuts content.

Conspiracies and regrets abound in Dune: Part Three trailer

The newly released trailer for Dune: Part Three, directed by Denis Villeneuve, offers a glimpse into the complex narrative that unfolds 17 years after the events of Dune: Part Two. Paul Atreides, portrayed by Timothée Chalamet, faces the repercussions of his ascent to power as Emperor, including a strained relationship with Chani, played by Zendaya. The trailer hints at conspiracies within his court, including threats from Princess Irulan and the villain Scytale, as Paul grapples with the moral implications of his actions that led to galactic warfare. The film is set to premiere in theaters on December 18, 2026.

Databricks makes Chinese open-source model GLM 5.2 its default coding engine after it matched Opus at lower cost

Databricks has selected the Chinese open-source model GLM 5.2 as its default coding engine after internal benchmarks demonstrated that it performs statistically on par with Anthropic's Opus 4.8 while being significantly cheaper at $1.28 per task compared to Opus's $1.94. This decision follows a comprehensive analysis using Databricks' own multi-million-line codebase, which revealed that many models fall into three distinct performance classes. Developer feedback from internal pilots supports the transition to GLM 5.2, which the company plans to utilize as a daily workhorse for its developers.

Claude AI may ask you if you’re using it too much

Anthropic is testing a beta feature for Claude AI that encourages users to reflect on their chatbot usage, addressing concerns about over-dependence on AI. This feature provides a summary of user interactions over varying time frames, such as one to twelve months, detailing usage patterns and tasks. Users will receive periodic prompts to consider their reliance on the AI, alongside tools to manage their interaction time, including quiet hours and break reminders. Anthropic's initiative aligns with recent discussions about the potential negative impacts of excessive AI use.

Meta fixed its smart glasses privacy light problem, just in time for always-on model to reportedly not use it

Meta has addressed a significant privacy concern with its Ray-Ban smart glasses by rolling out a mandatory update that disables the camera when the privacy light is broken. This update comes as the company prepares to launch a new model, dubbed 'super sensing,' which will feature constantly active microphones and cameras. Notably, this new model is expected to not activate the privacy light when in use, raising serious privacy implications for bystanders unaware they may be recorded. Additionally, Meta plans to utilize the data collected from these glasses to enhance AI models, with the recordings not being accessible to users.

Report: Apple interested in startup that runs giant AI models on iPhone without servers

Apple is reportedly exploring a partnership with PrismML, a startup that specializes in enabling large AI models to run directly on iPhones without relying on server infrastructure. PrismML has successfully adapted the Qwen 3.6 model, developed by Alibaba, to function on the iPhone 17 Pro, boasting 27 billion parameters—a significant increase compared to the few billion parameters typically active in mobile models. The startup plans to release its open-source model on July 14, 2026, which is capable of performing complex tasks such as software development. Apple has already engaged in discussions with PrismML regarding potential applications of this technology.

NHS AI blood test could reduce invasive womb cancer checks

The NHS is preparing to implement an AI-powered blood test developed by PinPoint Data Science to assess the risk of womb cancer in women referred for checks, potentially reducing the need for invasive procedures. Annually, about 90,000 postmenopausal women in England are referred for such checks, with around 10,000 diagnosed with womb cancer. The test, which costs approximately £30, analyzes around 30 blood markers and has demonstrated a 99.1% accuracy in identifying cancers as elevated or high risk. By using this test, the NHS aims to spare about 18,000 women from undergoing uncomfortable transvaginal ultrasound scans each year.

OpenAI Fixes 18-Year-Old GNU libunwind Bug by Treating Crash Debugging Like Epidemiology

OpenAI engineers resolved an 18-year-old bug in GNU libunwind that caused mysterious crashes in Rockset, the C++ data infrastructure service used by ChatGPT. The issue stemmed from a race condition in the _Ux86_64_setcontext function, which led to incorrect memory addresses during exception unwinding. By employing an innovative approach called epidemiological debugging, the team analyzed a year's worth of core dumps, identifying two unrelated bugs and isolating the root cause. After upstreaming the fix to GNU libunwind, the team confirmed that the issue was exacerbated by the high frequency of signal deliveries in Rockset, which turned a theoretical race condition into a production issue.

Payloads used to dictate the terms of launch. That's finally changing.

SpaceX's Starship, with a payload capacity exceeding 100 metric tons to low-Earth orbit, is transforming the space launch industry. As it remains in the experimental phase, both NASA and the US military are exploring its potential for lunar missions and cargo transport. The shift from traditional rocket designs, which catered to satellite needs, is evident as satellite manufacturers adapt to Starship's capabilities. Companies like Muon Space are developing new satellite architectures optimized for Starship's unique deployment methods, indicating a significant change in how payloads will be launched in the future.