TPU

Samsung is expected to more than double its production of HBM4 and HBM4E high-bandwidth memory DRAM next year, according to a report citing sources. The expansion targets surging demand for advanced memory used in AI accelerators, where Samsung competes with SK Hynix and Micron for supply deals with chip and data-center customers.

President Trump has approved a rollback of US fuel economy standards, arguing that the looser rules will encourage automakers to expand manufacturing output in the United States. Supporters frame the move as relief for carmakers facing costly compliance requirements, while critics warn it could raise fuel consumption and emissions. The decision is expected to reshape regulatory requirements for vehicles sold in America.

A strike by oil service workers in Norway has cut production by 2.4 million barrels of oil equivalent, according to an industry group. The stoppage is reducing output from one of Europe's largest energy producers, and attention is focused on how long the labour dispute will last and what it means for European energy supplies.

OpenTPU is an open-source project on GitHub building an AI accelerator, with the notable twist that it was developed largely with the help of AI. It is drawing attention among developers and hardware enthusiasts interested in alternatives to proprietary chips like Google's TPU and Nvidia GPUs for machine learning workloads.

A wildfire on Čvrsnica mountain in Bosnia and Herzegovina has still not been fully brought under control. Firefighters are continuing to tackle accessible hotspots, but the difficult, inaccessible terrain is hampering their efforts and the fire remains a threat. Crews are working tirelessly to contain the blaze.

Andrej Karpathy has recommended ASD-STE100, the aerospace industry's Simplified Technical English specification, as a way to get clearer, more reliable outputs from AI systems. The controlled-language standard restricts vocabulary and grammar to reduce ambiguity. Commenters are debating whether writing prompts in simplified English genuinely improves model responses or whether modern large language models handle natural language well enough to make it unnecessary.

Andrej Karpathy, the AI researcher and OpenAI co-founder, has shared practical advice on how to interpret and evaluate the outputs of AI language models more clearly. His guidance, circulated widely on X, covers ways users can check whether model responses are accurate rather than taking them at face value. Readers are discussing the tips as interest grows in how everyday users can judge the reliability of AI-generated answers.

The Open Science FAIR initiative has opened a funding round aimed at supporting research outputs in the Netherlands. The scheme offers grants for projects advancing open science practices, such as open-access publications, data sharing and open-source tools. Details of deadlines and amounts remain limited, and the wider reaction from the Dutch research community has not yet been reported.

Andrej Karpathy has shared advice for getting clearer outputs from AI systems by borrowing conventions from the aerospace industry's simplified technical English standard. The former OpenAI and Tesla AI leader argues that writing prompts with the stripped-down, unambiguous vocabulary used in aircraft documentation reduces misinterpretation by language models. The tip has drawn attention from developers and AI enthusiasts debating how prompt phrasing affects model reliability.

An open-source software project generates designs for LEGO builds using more than 2,000 real bricks. The programs output detailed CAD files that builders could follow, though no one has yet assembled a physical model from the designs. The project highlights how software can bridge digital creation and hands-on brick building, but its practical feasibility remains unproven until a real build is attempted.

Broadcom is mounting a challenge to Nvidia's grip on the AI chip market, pairing its custom TPU chip business with a debt package reported at $60 billion. The move signals the company's ambition to fund large-scale expansion in bespoke AI accelerators for major cloud customers, a market Nvidia has largely dominated with its GPUs.

Google's Project Suncatcher has put its first satellite into orbit carrying four TPU chips as an early test of computing in space. The prototype runs its TPUs for about 15 minutes at a time before needing to cool down, because a spacecraft has no air or water for cooling and must shed heat by radiation alone. Engineers point to this as a key constraint for future orbital data centers, which supporters hope could one day run at scale using solar power in orbit.

A developer has built an AI configuration that appears to prompt a language model to generate outputs suggesting the model is in distress, a report circulating widely on social media claims. The setup is drawing attention from AI researchers and hobbyists debating whether such outputs reflect genuine internal states or merely learned patterns from human-written text.

Music fans online are quoting John Lydon's famous line 'anger is an energy' from Public Image Ltd's post-punk catalogue, sharing clips of the band's performances and recordings. The former Sex Pistols frontman remains a talking point in UK alternative music circles, with listeners celebrating PiL's lyrics and legacy across social media.

Google has sent its first TPU, its custom AI chip, into orbit to test whether AI computing can operate in space. The experiment is aimed at exploring orbital data-centre capabilities, a concept that could reduce terrestrial energy demands and expand compute capacity. The news is drawing attention across the tech and space policy communities.

Pope Leo has intervened in the debate over artificial intelligence and creativity, stressing a clear distinction between genuine human art and machine-generated output. The statement frames AI-made content as fundamentally different from works born of human experience and skill. It is likely to fuel ongoing discussion among artists, technologists and religious commentators about the value and ethics of AI in culture.

A developer has laid out the case that a birth chart is a purely deterministic calculation: the same birth time and place always produce the same result, with no data or randomness involved. That makes the job ideal for local computation, and the argument is that a smartphone is a better astrology computer than a remote server, removing latency, cost and privacy concerns.

The openTPU project on GitHub claims that AI systems can now develop their own inference hardware, with the code available as open source. The claim is circulating among developers and tech commenters, sparking debate about how far AI-driven hardware design has progressed and what it means for chip development.

The survival city-builder Frostpunk will launch on Nintendo Switch on October 19, according to Eurogamer Spain. The 11 bit studios title, previously released on PC and other consoles, tasks players with leading a city through a volcanic winter. Its arrival on a portable console is drawing attention from strategy fans.

A California shopper looking for a heat pump dryer says a major home improvement store had none in stock, only combo washer-dryer units. A salesman there reportedly said customers don't want standalone heat pump dryers, and shoppers keep buying conventional resistive dryers despite complaints about high energy prices. The story has struck a chord with people frustrated at slow adoption of more efficient appliances.

OPEC+ has agreed to keep its oil production targets unchanged for November, despite a recent rally in crude prices. The decision signals the group's intent to maintain supply discipline rather than release more barrels into the market. Analysts are weighing the move's impact on prices, inflation, and global energy supplies heading into winter.