AI hardware companies

OpenAI alleges that an employee who left the company for Apple took confidential files relating to an unreleased device before departing. The claim, reported by Reuters, suggests rising tension between the two companies over talent and trade secrets. The story is drawing attention for what it implies about OpenAI's secretive hardware plans and the escalating competition for staff between the AI lab and Apple.

Market analysts argue that the AI investment trade is not ending but rotating: after years of returns concentrated in semiconductor makers like Nvidia, the next phase may favour software and application companies that monetise AI. Investors are being told to watch for earnings evidence that AI spending is translating into revenue across enterprise software, rather than staying with hardware suppliers.

Nvidia is reportedly in talks to acquire Reflection AI, a US startup developing open AI models, according to a Financial Times report. The deal would mark a notable move by the chipmaker into open model development, building on its dominance in AI hardware as competition over frontier model companies intensifies across the industry.

Anthropic is raising money to fund its computing capacity, offering Blackstone a double-digit yield on financing tied to semiconductor purchases. The deal is the latest example of AI companies turning to private credit to bankroll the expensive chips their systems require. It highlights how the race for AI infrastructure is drawing in large asset managers and pushing borrowing costs to unusually high levels.

Humxn Research Labs has announced plans to spend $50 million acquiring data on skilled trades, farming, and manufacturing to train AI-powered robots. The startup is paying for recordings and demonstrations of human workers performing hands-on tasks, betting that real-world labor data is the key bottleneck for robotics. The move highlights a growing scramble among robotics firms for high-quality physical-world training data as generative AI models increasingly move into hardware.

Nvidia has released three new open artificial intelligence models, expanding its portfolio beyond chips into AI software. The models are openly available, allowing developers to inspect, adapt and build on them. Nvidia has been positioning itself as a full-stack AI company, and the release underscores its push to compete with open model efforts from rivals while reinforcing the ecosystem built around its hardware.

Major technology companies are pushing to launch AI-powered gadgets in time for the holiday shopping season, according to CNBC. The scramble reflects the industry's effort to turn generative AI hype into consumer hardware, with devices aiming to catch gift buyers before the year-end rush.

Chinese companies have reportedly purchased more than one million NVIDIA gaming and workstation GPUs, repurposing them as an alternative source of computing power for AI work. The purchases come as domestic Chinese AI chip supply fails to keep up with demand, prompting firms to turn to consumer and professional-grade NVIDIA hardware to fill the gap despite export restrictions.

Amazon has introduced a new AI gadget described as a credit card-sized puck that listens continuously. The device's always-on microphone design is drawing attention and raising privacy questions, as coverage highlights that it records or processes ambient audio all the time. Reaction is focused on how far consumers will accept always-listening AI hardware in their homes and daily lives.

Search interest in Elon Musk's net worth is rising again, with readers tracking the value of his stakes in Tesla and SpaceX. It comes as a new report says SpaceX, along with Broadcom and Oracle, is tapping private credit markets to fund AI hardware purchases, keeping Musk's companies and his estimated fortune in the financial headlines.

ByteDance, the Chinese company behind TikTok, is reportedly in advanced negotiations to purchase custom AI chips from Taiwan's MediaTek, according to a Quartz report. The deal would help ByteDance secure dedicated semiconductor supply for its AI workloads amid tightening export controls on advanced chips from US vendors like Nvidia.

ByteDance, the Chinese owner of TikTok, is reportedly in talks with Taiwan's MediaTek to develop custom AI chips for its data centers. The move would help the company secure AI computing power for its large language model and recommendation workloads amid tightening US export controls on advanced Nvidia chips. A partnership with MediaTek would echo similar custom-silicon efforts by other major cloud and AI players, and highlights how Chinese tech firms are seeking alternatives to restricted US hardware.

Meta has released Muse Spark as an open AI tool, developed with Apple and optimized for Mac hardware. The announcement is drawing attention as the latest step in Meta's push to make its AI technology openly available, with Apple's involvement pointing to tighter integration between AI models and Mac devices. Details about capabilities and availability remain limited.

ByteDance, the Chinese parent company of TikTok, has commissioned MediaTek to design custom chips for its data centers, according to a report. The move would help ByteDance secure its AI and cloud computing infrastructure amid US restrictions on advanced chip exports to China. The news lifted interest in MediaTek, a Taiwanese chip designer, as companies increasingly turn to custom silicon.

Amazon's Bee has introduced two new personal AI devices, with prices starting at $75. The wearable AI products are designed to act as personal assistants, following Amazon's earlier acquisition of the Bee brand. Details on the devices' features and availability are limited so far, with the announcement drawing attention as Big Tech continues pushing consumer-facing AI hardware.

Nvidia and Broadcom are being framed as the two defining stories of the AI chip market, with commentators weighing their contrasting positions. Nvidia remains the dominant supplier of AI accelerators, while Broadcom has drawn attention for its custom silicon work with major cloud customers. The debate centres on which company's approach best captures the next phase of AI hardware demand.

Nvidia says its Vera CPU will serve as a central component of data-center infrastructure for agentic artificial intelligence, with the company pointing to growing demand for fast serial processing as AI agents take on more workloads. The announcement underscores Nvidia's push to expand beyond GPUs into core computing hardware for AI-driven data centers.

Server maker Super Micro Computer has appointed Mark Long as its chief financial officer. The company announced the appointment without releasing further details in the initial announcement. Supermicro, a major supplier of AI server hardware, is a closely watched stock, so leadership changes at the CFO level are drawing attention from investors tracking the company's financials and growth outlook.

Chinese gaming company Kunlun Tech has released an open-source AI game system that can run on a single Nvidia RTX 5090 graphics card. The release means developers and hobbyists can experiment with AI-generated game experiences without expensive server hardware. Details about the system's capabilities and reception in the developer community remain limited.

Tenstorrent, the AI chip startup led by veteran chip designer Jim Keller, has raised $1.5 billion to scale its AI computing systems. The company sells complete machines and licenses processor designs so customers can build their own chips. The funding is drawing attention as one of the most serious attempts yet to challenge Nvidia's dominance in AI hardware.

Amazon has introduced two new AI products under its Bee brand, named Bee Frontier and Bee Eclipse. Bee, an AI wearable company Amazon acquired in 2025, makes devices that record conversations and offer summaries. The announcement of the two new products is drawing attention from technology watchers tracking Amazon's expansion into AI hardware.

Amazon is reported to be stockpiling graphics processing units in warehouses faster than it can deploy them, with its own supply chain team saying there is no clear forecasted path to installing the hardware. The reported over-ordering highlights how fiercely tech companies are racing to secure AI computing capacity, sometimes outpacing data centre readiness and creating costly idle inventory.

A new supercomputer backed by large Swedish corporations has been officially inaugurated. The system is aimed at high-performance computing work, including artificial intelligence development, and is hosted in a new data centre facility in Sweden. The launch marks a significant investment by Swedish industry in domestic computing infrastructure at a time when access to powerful AI hardware is increasingly strategic for companies across Europe.

Greenpeace activists disrupted Nvidia CEO Jensen Huang's keynote in Berlin, demanding that the company "clean up your dirty business". The protest targeted the environmental and human rights impacts of the AI boom, staged during Nvidia's GTC event where the chipmaker's dominance in artificial intelligence hardware is being showcased.

Security company Stealthium has published research on detecting 'Rome attacks', a technique where an attacker secretly modifies a machine learning model's weights directly in GPU memory while it runs. The post outlines hands-on methods for spotting such tampering at runtime. Readers are discussing the implications for AI supply-chain security and the difficulty of trusting models loaded onto shared or untrusted hardware.