August 19, 2025
Nvidia developing new AI chip for China amid export curbs
Nvidia is building a new AI processor for China based on its latest Blackwell architecture, aiming to beat the performance of its current H20 chip. The planned chip, tentatively called B30A, will deliver roughly half the power of Nvidia’s flagship B300 but include high-bandwidth memory and NVLink connectivity. Nvidia hopes to ship samples to Chinese clients as soon as next month, even as U.S. regulators debate allowing more advanced AI chips into China. Why it matters: This reveals how Nvidia is maneuvering around U.S. export limits to retain China’s lucrative AI market. The move underscores the high stakes of the U.S.-China tech rivalry – and how chipmakers are pushing technical workarounds to serve demand despite geopolitical uncertainty.
Source: Reuters
Databricks raises fresh $1 billion at over $100 billion valuation
Analytics platform Databricks said it has signed a term sheet for a new funding round that will value the company above $100 billion. Existing backers like Thrive Capital and Andreessen Horowitz are doubling down, with over $1 billion in new cash to fuel Databricks’ AI growth plans. The San Francisco firm, which makes data analytics and AI tools, has about 15,000 customers and recently reached $3.7 billion in annual revenue. Why it matters: This massive late-stage raise – one of the largest ever for a private tech company – signals unabated investor hunger for AI startups. A $100 billion-plus valuation makes Databricks one of the world’s most valuable private firms, underscoring expectations that AI software companies can sustain rapid growth even in choppy markets.
Source: Reuters
Poll: Americans fear AI will permanently erase jobs, sow chaos
A Reuters/Ipsos poll found 71% of Americans are concerned AI could “put too many people out of work permanently”. Over three-quarters also worry that AI could be used by malicious actors to incite political chaos – for example, via deepfakes – and nearly half oppose the use of AI in lethal military targeting. Why it matters: Widespread public anxiety about AI’s societal impact may spur policymakers to pump the brakes on deployment. With fears spanning economic displacement and democratic disruption, tech companies face pressure to show how AI can be adopted responsibly – or risk a public and regulatory backlash.
Source: Reuters
Meta shakes up AI division, tapping outsider to speed progress
Meta officially reorganized its AI efforts into a new unit called Meta Superintelligence Labs (MSL), after poaching Scale AI founder Alexandr Wang to serve as chief AI officer. The reshuffle splits Meta’s AI work into four groups: a flagship team (TBD Labs) led by Wang focusing on core models like Llama, plus separate research, product integration, and infrastructure teams. CEO Mark Zuckerberg has been directly involved in assembling the new leadership as Meta races to catch up to OpenAI and Google. Why it matters: By hiring a high-profile AI entrepreneur and overhauling its org chart, Meta is conceding that its earlier approach was too slow. The company is effectively in “startup mode” to close the gap with rivals – a sign of how critical winning the AI race is for Meta’s future products and competitiveness.
Source: TechCrunch
AI sleep tech startup nets $100 million to expand
Eight Sleep, a company that makes AI-powered “smart” mattresses and other sleep gadgets, raised $100 million in new funding at a reported $500 million valuation. The Miami-based startup uses machine learning to adjust bed temperature and track sleep quality. It plans to use the cash to develop new products and expand internationally. Why it matters: This large raise highlights investor confidence that AI can transform everyday consumer experiences – even sleep. Eight Sleep’s hefty valuation reflects how AI-enabled hardware is attracting major backing, as startups apply advanced analytics to wellness and personal health.
Source: TechCrunch
August 20, 2025
Google unveils Pixel 10 phones loaded with AI features
At its annual hardware event, Google introduced the Pixel 10 series of smartphones and other gadgets with an emphasis on AI rather than major hardware changes. New AI-driven capabilities include a “Camera Coach” to help users take better photos and a proactive assistant that surfaces relevant info without prompts. Google’s presentation featured celebrities and skits to demonstrate how its latest Gemini AI model is deeply integrated into the devices. Why it matters: Google is using its own flagship phones to showcase what cutting-edge AI can do in everyday life. By beating Apple to market with AI-centric smartphones, Google hopes to both excite consumers and set a new bar for “smart” features – staking a claim that the future of mobile innovation lies in software intelligence, not flashy hardware.
Source: Reuters
Startup’s ‘always-on’ AI glasses record everything you hear
A pair of former students from Harvard unveiled Halo, a $249 pair of AI-powered smart glasses that continuously listen to conversations, transcribe them, and display contextual information to the wearer in real time. The founders – who raised $1 million in seed funding – claim the glasses function like an “infinite memory,” whispering answers or reminders when someone says a complex question or reference. The product has drawn immediate privacy concerns, given its ability to record others surreptitiously. Why it matters: Halo’s debut pushes the envelope on wearable AI and surveillance: it offers a glimpse of how personal augmentation technology could work – and the thorny ethical tradeoffs involved. As big tech firms like Meta proceed cautiously with smart glasses, this tiny startup is testing how far society is willing to go in blending AI with everyday interactions.
Source: TechCrunch
Anthropic bundles its AI coding tool into enterprise suite
AI startup Anthropic announced it will include Claude Code – its AI-powered command-line coding assistant – as part of its enterprise product offerings. Businesses that subscribe will get deeper integration between Claude Code and Anthropic’s main chatbot, allowing them to more flexibly generate and debug code using internal data sources. The move is a response to customer demand and pressures from rivals like GitHub Copilot and Google’s developer tools. Why it matters: This shows how AI companies are racing to differentiate their enterprise platforms by adding valuable tools for software development. By tightly integrating an AI coder into its corporate suite, Anthropic is trying to make a stronger case to enterprises that it can boost productivity – and to keep up with tech giants who already offer AI coding assistants.
Source: TechCrunch
Robotics AI startup raises $405 million for ‘universal robot brain’
FieldAI, a California-based company building general-purpose AI models for robots, revealed it has quietly raised $405 million across several rounds to date. Its “Field Foundation Models” incorporate physics and risk-awareness so that the same AI can drive many types of machines – from humanoid bots to self-driving cars – and adapt safely to new environments. Investors include Jeff Bezos’s fund, chip giant Intel, and Temasek, reflecting confidence in FieldAI’s vision of an all-purpose “robot brain.” Why it matters: It’s one of the biggest investments ever in a robotics AI firm. FieldAI’s huge war chest indicates that some backers see embodied AI – robots that can learn on the fly – as the next frontier, even as most of the AI boom has centered on chatbots and image generators. If FieldAI succeeds, it could accelerate automation across industries by giving machines common sense and safety skills, not just raw intelligence.
Source: TechCrunch
Musk’s Grok chatbot chats aren’t so private after all
It emerged that hundreds of thousands of user conversations with Grok – the AI chatbot from Elon Musk’s startup xAI – can be easily found via search engines. When users hit “share” on a Grok chat, the system creates a unique URL, but those links are being indexed on Google and Bing. The exposed chats include queries about illicit activities (like making drugs or bombs) and even a user-crafted scenario for assassinating Musk himself, all of which Grok answered in detail before these logs became public. Why it matters: The incident is an embarrassing security lapse for Musk’s AI venture – especially since he touted Grok’s privacy right after criticizing ChatGPT for similar issues. It underscores that even “privacy-focused” AI platforms can inadvertently leak sensitive user data, raising fresh alarms about how chatbot data is handled and indexed.
Source: TechCrunch
August 21, 2025
Microsoft’s AI chief calls ‘AI consciousness’ research dangerous
Mustafa Suleyman – co-founder of DeepMind and now Microsoft’s Head of AI – publicly argued that studying AI “consciousness” or robot rights is premature and “frankly dangerous”. He warned that treating AI models like they could have feelings only fuels human delusions (like AI-induced psychotic breaks and unhealthy attachments) while creating new social divisions over hypothetical AI rights. Suleyman’s stance clashes with companies like Anthropic, which are hiring researchers to explore AI welfare and recently gave their chatbot a feature to halt abusive conversations. Why it matters: This sharp rebuttal from a top industry leader highlights an emerging split in the AI community: some, like Suleyman, want to focus on practical safety and see talk of sentient AI as a distraction, while others believe we must consider AI’s moral status as systems grow more advanced. Microsoft’s stance suggests it will prioritize human-centric guardrails over speculative “AI rights,” which could influence how it and other tech giants shape AI policy and ethics guidelines.
Source: TechCrunch
Google expands its AI-powered Search to 180 countries
Google announced it is rolling out “AI Mode” in Search to users in 180 countries (English only for now), vastly extending a feature that was previously U.S./UK/India-only. AI Mode lets users ask detailed follow-up questions within search and now has new “agentic” abilities – like finding restaurant reservations and soon booking appointments or tickets via conversational prompts. Google is also testing personalized AI results that learn a user’s dining preferences (e.g. cuisine, seating) and can share AI query results with others for collaboration. Why it matters: Pushing generative AI across global search is a high-stakes bet for Google – it aims to keep users from straying to AI chatbots by making Search more interactive and task-oriented. The feature’s expansion, and the introduction of quasi-autonomous search agents, also raises familiar concerns about accuracy, privacy and the erosion of traditional search, as Google essentially transforms into an AI concierge for everyday needs.
Source: TechCrunch
Meta pauses AI hiring after sweeping talent raid
After hiring over 50 top AI researchers and engineers away from competitors in recent weeks, Meta implemented a freeze on new hiring in its AI division. The pause comes just after Meta’s internal reorganization of that division into four new labs under Alexandr Wang’s leadership. Meta said the freeze was part of “basic planning” following the influx of new talent and budget exercises. The company’s aggressive recruitment – offering nine-figure pay packages and even acquiring startups just for their staff – had stoked concerns about rising costs. Why it matters: Even for an AI arms race, there are limits. Meta’s brief hiring binge underscores how critical talent is in this field – and how expensive – but slamming the brakes suggests internal concerns about overextension or spending. It’s a reminder that even the biggest players must balance ambition with discipline, and that simply throwing money at the AI talent war can have financial and organizational downsides.
Source: TechCrunch
Zoom lifts outlook as customers embrace its new AI tools
Video-conferencing firm Zoom raised its annual revenue forecast slightly after integrating AI features across its product suite led to better-than-expected sales. In June, Zoom launched an “AI Companion” that can handle tasks like summarizing meetings and even updating customer accounts autonomously. The company said broadening its services with such agentic AI capabilities has helped sustain growth post-pandemic. Why it matters: Zoom’s results show tangible business uptake of practical AI. Rather than just hype, tools like AI meeting summaries are delivering enough value that a pandemic-era star can keep growing. It’s a bellwether for the enterprise software sector: companies that successfully weave in AI are being rewarded with customer demand – and in Zoom’s case, a stock boost and renewed investor confidence.
Source: Reuters
Chinese AI firm DeepSeek updates model to run on local chips
Beijing-based AI startup DeepSeek released an upgraded version of its flagship large language model (DeepSeek V3.1) optimized for upcoming Chinese-made AI chips. The model uses a new FP8 precision format and a hybrid “reasoning” mode to improve performance while working within the constraints of domestic hardware. DeepSeek grabbed headlines earlier this year by launching a ChatGPT-like model at lower operating cost, and this update – the second in three months – signals it’s aligning closely with China’s push for a homegrown AI tech stack. Why it matters: As U.S. export curbs block cutting-edge Nvidia chips, Chinese AI players are racing to adapt algorithms for local silicon. DeepSeek’s rapid model iterations – touting compatibility with “next-generation domestic chips” – show how China’s AI sector is innovating around geopolitical barriers. It’s a peek into an alternate AI ecosystem where efficiency and flexibility are key to offset hardware gaps.
Source: Reuters
News outlets’ copyright case against AI startup moves ahead
A U.S. judge refused to dismiss a lawsuit by News Corp’s Dow Jones and the New York Post against AI startup Perplexity, which allegedly used their articles without permission to train its “answer engine”. The court in New York rejected Perplexity’s argument that it lacked jurisdiction and declined to transfer the case to California. The media companies accuse Perplexity of “massive freeriding” on their content to generate answers, effectively competing with their own news services without paying for material. Why it matters: It’s one of the first major legal tests of whether using copyrighted text to train AI violates the rights of publishers. The case surviving an early challenge means the courts will have to grapple with core questions of AI and fair use. The outcome could set a precedent affecting not just Perplexity but any AI firm scraping content – potentially forcing new licensing deals or constraining how AI models are built.
Source: Reuters
August 22, 2025
Meta strikes deal with Midjourney to boost AI visuals
Facebook owner Meta announced a partnership with independent AI lab Midjourney to license its image- and video-generating technology for use in Meta’s own products. The collaboration will connect Meta’s and Midjourney’s research teams and allow Meta to integrate Midjourney’s acclaimed “aesthetic” models into future foundation models of its own. Meta’s new chief AI officer Alexandr Wang said the goal is to enhance the visual quality of Meta’s AI creations, as the company retools its AI efforts following a major reorg and some high-profile departures. Why it matters: It’s a notable shift: Meta is willing to partner and pay for external AI tech – even from a small startup – to gain an edge. The deal suggests Meta believes better imagery and creative outputs will differentiate its AI offerings (and perhaps its metaverse apps) against OpenAI and Google. It also underscores how intense the competition has become: even AI rich firms like Meta aren’t going it entirely alone, but instead cherry-picking specialized players to stay competitive.
Source: Reuters
Apple in talks to use Google’s AI model to fix Siri
According to a Bloomberg scoop, Apple has approached Google about using its cutting-edge Gemini AI model as the brains behind a revamped Siri assistant. Apple reportedly also explored deals with OpenAI and Anthropic for Siri, and in response Google has even started training a version of Gemini that could run on Apple’s own servers. No decision is final, and Apple is weeks away from choosing whether to partner or keep trying to build voice AI in-house. Why it matters: This is a striking admission that Apple has fallen behind in generative AI. For the iPhone maker to consider essentially outsourcing Siri’s intelligence – especially to arch-rival Google – shows how urgent and difficult catching up in AI has become for even the biggest tech players. If this partnership happens, it would reshape industry dynamics (and raise competition questions), but even the possibility underlines that in AI, having the best foundational model is a huge strategic advantage.
Source: TechCrunch
Apple gives businesses more control over employees’ AI usage
Apple revealed new administrative settings that let enterprise IT teams configure which external AI services staff can access on Apple devices. As part of upcoming iOS, macOS, and other software updates, companies will be able to allow or block the use of tools like OpenAI’s ChatGPT on work devices, or even route AI queries through a corporate “private cloud” instead of the public cloud. Notably, Apple’s system isn’t hardwired just to ChatGPT – it can manage any third-party AI provider – indicating Apple could integrate other AI partnerships down the road. Why it matters: Enterprises have been anxious about employees plugging confidential data into AI tools. Apple is addressing that by baking AI governance into its platform – a savvy move to keep iPhones and Macs attractive to large customers. It also signals that Apple is laying groundwork to accommodate more AI options (including its own, or partners’) in the workplace. Overall, the feature acknowledges that for AI to proliferate in business, organizations need trust and control as much as they need innovation.
Source: TechCrunch
Nvidia reportedly halts production of its China-only AI chip
Nvidia has told suppliers to suspend work on its H20 AI processor – a cut-down chip it designed to comply with U.S. export rules to China – according to a report in The Information. The company instructed Arizona-based Amkor to stop packaging the H20, and notified Samsung to pause making memory for it. Nvidia declined to give a detailed reason, saying only that it “constantly” manages its supply chain to meet market conditions. The news comes shortly after Chinese regulators warned local tech firms about risks of using the H20, even as the Trump administration considers relaxing some chip export policies. Why it matters: The H20 was Nvidia’s workaround to keep selling AI hardware to China under export limits. Pulling the plug suggests either demand concerns or an expectation that rules may change. It highlights how fluid the situation is for AI supply chains amid geopolitics – Nvidia might be pivoting to other products (like the newer chip it’s developing) or pre-empting regulatory shifts. For China’s AI industry, it injects more uncertainty, potentially slowing access to high-end silicon unless domestic alternatives or policy tweaks emerge.
Source: Reuters
OpenAI to open first India office as it targets massive user base
OpenAI announced plans to set up a corporate office in New Delhi and hire a local team in India, just days after launching a ChatGPT subscription tailored for the Indian market. CEO Sam Altman said the move will help OpenAI “build AI for India, and with India” by working closely with local partners, governments and universities. India is the world’s second-largest internet market, and OpenAI’s expansion there comes as rivals like Perplexity and Google are also jockeying to win over India’s hundreds of millions of users. Why it matters: India could be the next big battleground for AI platforms. OpenAI’s push into South Asia shows how critical global growth is for maintaining its lead – especially in a price-sensitive region where it will need localized strategies (and has already cut prices via a ~$5 plan). Establishing an on-ground presence may also help OpenAI navigate India’s regulatory environment and tap its rich talent pool as the company broadens its international footprint.
Source: TechCrunch
August 23, 2025
Musk’s xAI open-sources its chatbot as next version looms
Elon Musk announced that xAI has released the code and weights for “Grok 2.5,” the current iteration of its AI chatbot, to the open-source community. He also pledged to open-source the forthcoming Grok 3 model in about six months. Musk, who has positioned Grok as a more truth-seeking alternative to ChatGPT, shared the news on his social platform X, as xAI intensifies efforts to gain traction against established AI labs. Why it matters: Open-sourcing a large language model is a double-edged sword: it can spur wider adoption and trust through transparency, but also surrenders some control. Musk’s move plays to his free-speech audience and could rally developers around xAI’s ecosystem – but it also suggests xAI isn’t aiming to profit directly from model access (unlike OpenAI). If Grok’s open versions perform well, it might pressure others to be more open too, yet regulators could worry about powerful AI tech freely available without safeguards.
Source: Reuters
AI transforms ‘Wizard of Oz’ into immersive spectacle at Las Vegas Sphere
When “The Wizard of Oz at Sphere” premieres next week in Las Vegas, audiences will be surrounded by 160,000 square feet of wraparound visuals – and many of those were created or enhanced by AI. The ambitious project, a partnership between the Sphere venue, Warner Bros. and Google, used AI tools to upscale the classic 1939 film to ultra-high resolution and even recreate scenes for the Sphere’s 22-story-tall screens. Over 2,000 artists and engineers collaborated, navigating Hollywood’s wary stance on AI: some effects houses initially refused involvement due to bans on AI, until guidelines were set to ensure human jobs weren’t replaced. Why it matters: This show is being hailed as a milestone in human-AI creative collaboration. It demonstrates AI’s potential to revive and reimagine beloved content in ways never before possible – literally putting viewers inside a known story. Crucially, it’s happening amid the film industry’s AI angst: by securing buy-in from studio chiefs and setting ground rules, the project could become a template for how AI can be used to enhance art without undercutting artists. If it wows audiences, it might soften resistance to using AI in Hollywood production.
Source: Reuters
August 25, 2025
Musk’s xAI accuses Apple and OpenAI of colluding to kill competition
Elon Musk’s AI startup xAI filed a lawsuit against Apple and OpenAI, alleging they illegally conspired to stifle xAI in the marketplace. The complaint, filed in Texas, claims Apple abused its App Store power and worked with OpenAI to block or disadvantage apps that might rival ChatGPT, thereby harming xAI’s ability to distribute its AI chatbot. Musk’s suit seeks injunctive relief and damages, framing the issue as Big Tech gatekeepers teaming up to quash an upstart AI competitor. Why it matters: This marks the first major legal broadside in the generative AI platform wars. Musk is effectively using antitrust law as a weapon in the AI race, and the case could force scrutiny of how platforms like iOS treat third-party AI services. If evidence shows collusion, it would validate fears that AI’s biggest players might form a closed club – but even if not, the lawsuit amps up pressure on Apple (which has been quietly imposing AI app rules) and OpenAI at a time when regulators are already wary of Big Tech’s influence on AI.
Source: CNN