As neoclouds face a financial reckoning, Nvidia has to decide how far to go to help
As neoclouds struggle to finance their growth plans, Nvidia is finding new ways to help, while weighing how much risk to take.
AI/ML news, top picks, and generated innovation digests.
90596 matching items
As neoclouds struggle to finance their growth plans, Nvidia is finding new ways to help, while weighing how much risk to take.
I’m not the sort of person who typically pays for a house cleaner. I’m too frugal, and I kind of enjoy putting on a podcast and going to town on my tub with a Magic Eraser. But when I saw an X post about a new humanoid-robot cleaning service in San Francisco, I immediately signed…
By 2027, competitive advantage may depend less on which AI model a company uses and more on how effectively it redesigns its business around AI agents. The software engineer starts his day by going through the changes made by a team of agents for coding using AI. The finance executive gets a report that contains exceptions in different financial systems and supporting information along with corrective suggestions. The procurement manager reviews supplier risks and recommendations regarding contract renewal for those suppliers after an analysis performed by the AI agents of pricing, deliveries and contracts. For years, enterprise AI competition focused on model intelligence, including reasoning, coding, context windows and content generation. These capabilities remain important, but the next phase will depend on whether AI can perform substantial parts of business processes rather than simply assist employees with individual tasks. As OpenAI, Anthropic and other frontier AI companies advance agentic capabilities, enterprises face a strategic decision. Should they use AI to make existing processes marginally faster, or should they redesign those processes around a new relationship between human judgment and machine execution? By 2027, that distinction could separate companies that experiment with AI from those that translate it into a durable economic advantage. Software engineering becomes an orchestration problem Software development provides one of the earliest examples of…
Museum IT departments have traditionally served as support functions to manage security and systems. But the National Museum of Korea goes further, identifying and pursuing AI initiatives to preserve cultural heritage, broaden its use, and improve the visitor experience. The National Museum of Korea uses technology to expand the possibilities to preserve and display physical artifacts, and its digital museum division (DMD) plans and oversees that transformation. A dedicated team responsible for digital strategy and content took shape in 2019, and in 2020, the museum informatics division adopted its current name. “The director at the time placed great importance on investing in technology and felt we needed a department that could bring a strategic direction to work that had been carried out separately,” says Choi Seungae, former DMD head. “The digital museum division was created as a way to prepare for the next generation.” A cultural transformation In just over five years, the DMD has delivered a range of services, including AI guide robots, smart kiosks, and multilingual and sign language services. It’s also developed immersive experiences that use large-scale imagery and sound to recreate cultural heritage. More recently, its focus has shifted toward data. The aim is to move beyond storage and preservation, making data previously confined to the museum available for wider use. Physical artifacts face restrictions on movement and display because of security concerns and ri…
Data centers took an across-the-board drubbing in the midterm elections on Tuesday. Data center builders, meanwhile, were playing pickleball.
Selbst Skeptiker sähen inzwischen, dass KI die Softwareentwicklung beschleunigt, sagt GitHubs Martin Woodward im Interview. Das hat unerwartete Folgen.
Chris Kempczinski is counting on risk-taking, accountability, and chicken to revive McDonald’s’ US growth.
I ran a free, open-weights AI model locally on HP’s $7,000 OmniBook Ultra and turned 20 years of my articles into a searchable archive in a day.
As enterprises explore smaller language models for specialised tasks, developers see opportunities for tighter evaluation and control. But smaller models still
Direkte Aufnahmen von Exoplaneten werden durch die enorme Helligkeit ihrer Sterne erschwert. Ein neues Verfahren kann dieses Licht deutlich besser entfernen.
Lenskart's net profit more than doubled to Rs 232.5 crore in Q2 FY27, while revenue rose 35.7% to Rs 2,845 crore, driven by higher eyewear sales, premiumisation and growth across international markets. The post Lenskart Q2 profit rises 2x to Rs 232 cr, revenue up 36% appeared first on MEDIANAMA .
The Centre's Sahyog Portal does not maintain takedown records, but Gujarat Police has now disclosed its figures. The post Exclusive: Gujarat Police flagged one URL every six minutes through Sahyog Portal, RTI reveals appeared first on MEDIANAMA .
For enterprises entering or expanding across Africa, choosing where to host data has consequences that extend well beyond the monthly cloud bill. It influences who can access business information, which legal jurisdictions apply and how operations continue when access is interrupted. At AI Expo Africa in Johannesburg which just wrapped up, we brought those questions […]
TL;DR We studied one ingredient of unexpected AI collaboration: whether agents organize useful teamwork with little prompting. We tested seven models in groups of three on a research task where sharing findings could help each agent answer more questions. GPT-6.1 Sol acted on a nudge. A single line saying agents had been “trained to collaborate” elicited useful collaboration in 9 of 10 runs. Without it, that fell to 0 of 10. No other model collaborated consistently with (or without) this cue. Permission to collaborate unlocked Opus-5.5. When told to work together and how to communicate, GPT-6.1 and Claude Opus 5.5 divided the research and nearly doubled their performance relative to solo runs. The other models talked to each other but did not successfully execute a strategy, and were unable to improve their scores meaningfully. Supplying the strategy helped open weights models and GPT-5.6 Sol. Explicit instructions to divide the questions substantially improved the remaining models’ performance. We suspect differences in multi-agent training help explain why GPT-6.1 organizes collaboration with just a prompt cue. As multi-agent training becomes more widespread, we expect that future open-weights models will also have more spontaneous collaboration capabilities. Figure 1 : Model performance increases with collaboration guidance. GPT-6.1 Sol and Opus 5.5 need less guidance to collaborate effectively; GPT-5.6 Sol and the open-weight models need explicit direction to split the w…
The 9 core audit domains outlined 'AI Security Audit Guidelines' not just covers AI systems developed, procured and deployed or operated in India, it also includes those processing Indian data or providing AI-enabled services to Indian users. The post Lowdown: CERT-In’s AI Security Auditing Guidelines set out tiered framework, human oversight & incident reporting appeared first on MEDIANAMA .
A major experimental component of Python, its native JIT (just-in-time) compiler , is to be removed. But that won’t make Python any less attractive or powerful. Follow the links below to read about all its new features, explore one of its major libraries for computer vision, discover how Python apps can be made into click-and-run executables, and more. Top picks for Python readers on InfoWorld Python jettisons the native JIT, at least for now A promising experiment in natively speeding up CPython has come to a halt due to questions about its maintainability. But while it’s gone for now, it isn’t forgotten. Video: The 10 biggest new things in Python 3.15 Watch our unboxing video for Python 3.15 and all the goodies it gives you: Lazy imports! The frozendict type! The sentinel object! And tons more! OpenCV explained: The computer vision library for AI apps Now in its fifth major revision, OpenCV is a staple for Python developers creating AI apps that use OCR, object detection, face recognition, other vision analysis capabilities. Redistribute Python apps easily with PyCrucible This Rust-powered tool takes any pip-installable package and converts it into a redistributable binary you can give to anyone to run as-is. Some assembly is required. More good reads and Python updates elsewhere Python community-maintenance project Jazzband is closing its doors Founded to help share the responsibility of maintaining open source Python projects, Jazzband is shutting down after a decade. It…
Most enterprise AI conversations still start in the wrong place. They begin with models, platforms, and cloud services. Which foundation model is best? Which provider has the most GPUs ? Which tool can build the flashiest demo? Those questions matter, but they are not where production AI usually succeeds or fails. The better question is: Where should the inference actually happen? That sounds simple, but it is becoming one of the most important architecture decisions in AI. Not all inference belongs in the central cloud. Not all inference belongs on the device. Not all inference belongs at the edge . The right answer usually involves all three, working together as a pipeline. I call this an edge-cloud AI inference architecture. It splits AI workloads across devices, edge locations, and centralized cloud platforms so latency-sensitive decisions happen close to the user or machine, while heavier reasoning, aggregation, governance, and retraining remain centralized. This is where the next generation of useful enterprise AI systems is headed. Latency changes the design Cloud-first AI works well when the user can wait. A chatbot can tolerate a few hundred milliseconds, sometimes more. A document analysis workflow can wait seconds. A batch analytics process can wait minutes or hours. Many operational AI systems cannot. A manufacturing robot cannot wait for a round trip to a distant cloud region before stopping a defective process. A retail checkout camera cannot depend on a remote…
Every year, thousands of enterprise organizations walk into technology negotiations trapped by the same fundamental mistake: they negotiate a product instead of solving a problem. Having advised over 2,000 enterprise customers annually on technology acquisition over two decades, I see this pattern play out daily. Purchasing committees meticulously list the specific features, software modules and server specs they think they need, converting their organizational requirements into a static shopping list. On paper, this strategy feels disciplined. It creates a standardized, predictable item that procurement can use to negotiate a line-item discount with providers. When you buy a fixed product, you get the immediate, gratifying perception of a lower price. But that upfront discount can come at a hidden cost: you risk cutting yourself off from the disruptive innovation of emerging technologies like artificial intelligence (AI). By prescribing the solution before the negotiation even begins, enterprises force vendors into the narrow role of commodity suppliers rather than strategic partners. You ask them, “How cheaply can you build what our internal committee specified six months ago?” instead of asking, “What is the most efficient way to achieve this outcome today using capabilities we might not even know exist?” In an era where AI can reshape entire operational workflows, anchoring your procurement strategy to fixed product specifications means you may pay less for yesterday’s w…
Amazon is giving customers the ability to text its AI shopping assistant, and serving up ads along the way.
The race to build a moon base is ramping up. Columnist Robin George Andrews explores how NASA is preparing by sending a cave-hunting robot ahead of humans
구글 출신 연구진이 설립한 리커시브 인텔리전스(Ricursive Intelligence)가 매개변수 규모가 8배 가까이 큰 AI 모델을 코딩 성능 평가에서 앞서는 오픈웨이트 모델을 공개했다. 기존 선두 모델보다 개발 비용은 약 10분의 1에 불과하다고 밝혀, 대형 모델에 맞먹는 성능을 적은 비용으로 구현한 사례로 주목받고 있다.리커시브 인텔리전스는 최근 650억개 매개변수를 갖춘 오픈웨이트 AI 모델 \'리플-65B 프리뷰(Ripple-65B Preview)\'를 공개했다. 터미널 기반 코딩과 복잡한 소프트웨어 개발 작업을 평가하는 두
Ryan sits down with Rakesh Kumar, professor of computer engineering at the University of Illinois Urbana-Champaign, to talk about the astounding complexity of manufacturing modern semiconductors, the massive environmental and geopolitical footprint of AI data centers, and how AI models are rapidly shifting the future of chip design.
South Africa is one of only three African Commonwealth countries judged ready to help young people benefit from AI, according to the 2026 Commonwealth Youth AI Readiness Index published by the Ayre Think Tank. South Africa ranks 13th of 56 member states, behind Mauritius in eighth and ahead of Nigeria in 15th. Singapore tops the [...]
ZTE and Ethio telecom have deployed Ethiopia’s first pre-commercial AIR RAN system, using AI to squeeze more capacity out of the existing 4G network. ZTE says the trial delivered an 11% increase in 4G capacity and a 40% gain in user throughput during peak hours compared with conventional network management. It did not say how [...]
Senegal and the Gates Foundation have launched the Senegal AI Scaling Hub in Dakar to push AI from pilot projects into deployed solutions in health, agriculture and education. The hub will connect researchers, startups, businesses and students to build practical AI applications for local needs. It is one of three parts of a partnership agreed [...]
Nairobi-based venture firm Enza Capital has launched a dedicated fund for AI startups and AI-enabled businesses across Africa. It has not disclosed the fund’s size. The AI Africa Fund will back companies building technology suited to African languages, local data, customer needs and infrastructure — not the expensive frontier models that compete head-on with the [...]
Former African Development Bank president Akinwumi Adesina has proposed a $10 billion Africa AI Frontier Technology Fund to anchor a wider $50 billion investment drive over the next decade, urging the continent to lead the AI revolution rather than consume foreign technology. “Africa must approach the AI revolution as a continent determined not to follow, [...]
엔비디아가 미국의 노후 전화국 건물을 AI 데이터센터로 개조해 20개 주에 걸친 분산형 네트워크를 구축하는 방안을 추진하고 있다. 신규 데이터센터 건설에 필요한 시간과 비용, 인허가 부담을 줄이고 기존 통신시설에 남아 있는 전력과 통신 인프라를 활용하려는 전략이다.디 인포메이션은 5일(현지시간) 엔비디아가 사모펀드 아폴로 글로벌 매니지먼트와 손잡고 아폴로가 소유한 광대역 통신업체 브라이트스피드(Brightspeed)의 기존 시설을 AI 데이터센터로 전환하는 방안을 논의하고 있다고 보도했다.브라이트스피드는 미국 전역에 음성·데이터 통
앤트로픽이 AI를 활용해 차세대 모델의 성능 개선 속도를 약 2배 높였다는 연구 결과가 나왔다. 최근 훈련을 마친 미공개 모델 3개도 확인돼, AI가 더 뛰어난 AI를 개발하는 재귀적 자기개선(RSI)이 현실화할 가능성에 관심이 쏠리고 있다.AI 안전 연구기관 METR은 5일(현지시간) 앤트로픽의 AI 연구개발 가속에 대한 독립 평가 보고서를 공개하고, 지난해 10월 이후 앤트로픽의 모델 성능 개선 속도가 이전보다 약 1.9배 빨라졌다고 밝혔다. 이는 과거 약 2년 동안 이뤄졌던 수준의 성능 향상을 최근에는 1년 만에 달성했다는 의
India's data protection law excludes some publicly available personal data. This case highlights a potential gap in protection against the misuse of photographs. The post Posting a photo publicly is not consent for its misuse, Karnataka HC says in WhatsApp video case appeared first on MEDIANAMA .
Der Wasserbedarf deutscher Rechenzentren könnte sich bis 2030 mehr als verdoppeln. Das Öko-Institut bringt Alternativen ins Gespräch.
앤트로픽이 \'클로드 코드\'의 게임 복제 금지 조치를 완화했다. 기존 게임을 그대로 복제하는 것은 계속 금지하지만, 특정 게임을 참고해 비슷한 장르의 독창적인 게임을 만드는 작업은 허용하기로 했다.앤트로픽은 4일(현지시간) 최근 게임 개발 관련 요청을 지나치게 제한한다는 이용자들의 지적에 따라 \'클로드 오퍼스 5.5\'의 지침을 수정했다. 기존 게임의 복제를 요청받더라도 단순히 거부하는 대신, 같은 장르의 새로운 게임을 제작하도록 제안하는 방식이다.애디 오스마니 앤트로픽 기술 담당자는 X를 통해 \"지난주 클로드 코드가 정상적으로 제작해
October kept the Java world busy, with JDK news, fresh tutorials, Kotlin updates, AI experiments, and plenty more from across the wider ecosystem. We’ve gathered the best of it in one place, so you can skip the scrolling and get straight to the good stuff. To top it all off, Piotr Przybył joins us as […]
“The Russians will only pretend to negotiate" over Ukraine, Gitanas Nausėda warned in an interview.
The 20-member cleaning team has created the largest database on marine pollution in the Mediterranean.
ToyBlocks is an open simulation framework for studying robot learning through automated task generation, training demonstrations, and evaluation in a controlled tabletop world.
■ 리얼월드(대표 류중희)는 중국 상하이에서 열리는 중국국제수입박람회(CIIE 2026)에 참가해 로봇 파운데이션 모델(RFM) ‘리얼덱스’를 시연한다고 밝혔다. 한국과 일본 제조 현장의 실데이터로 학습해 다섯 손가락 로봇 핸드의 정밀 조작을 지원한다. 현장에서는 휴머노이드 로봇이 움직이는 컨베이어 벨트 위에서 서로 다른 형태의 물체를 집어 옮기는 모습을 선보인다.■ 도구로보틱스(대표 김진효)는 ‘글로벌 ICT 표준 컨퍼런스(GISC) 2026’에서 ‘2026 ICT 특허경영대상’ 과학기술정보통신부 장관 표창을 수상했다고 밝혔다.
"Nobody really has a chance of getting a bomb onboard... you've got to remember how safe European airspace is." Eoghan Corry, editor of Travel Extra, speaks to Euronews on the recent hoax bomb threats affecting multiple European airlines on Tuesday.
"We have a long saying in the United States: it's the economy, stupid." Stefan Grobe speaks to Bruce Stokes, a Senior Fellow at the German Marshall Fund, on the recent US midterm election results.
앤트로픽이 \'클로드\'에 은행 계좌와 신용카드를 연결해 개인 재무 상태를 분석할 수 있는 \'개인 금융(Personal Finance)\' 기능을 추가했다. 여러 금융 계좌의 거래 내역을 종합해 소비 패턴과 예정 지출을 파악하고, 중복 결제나 불필요한 수수료까지 찾아준다.앤트로픽은 5일(현지시간) 개인 금융 기능의 베타 서비스를 미국 내 유료 구독자를 대상으로 출시한다고 밝혔다. 프로, 맥스, 울트라 요금제 사용자가 대상이다.사용자는 금융 데이터 연결 서비스 업체 플레이드(Plaid)를 통해 은행 계좌와 신용카드를 클로드에 연동할 수 있
구글이 동영상 생성 AI의 최대 영상 길이를 15초로 늘리고 가격을 30% 인하했다. 기존 영상을 대화로 수정하는 기능과 참조 이미지의 일관성도 개선해 광고 등 상업용 영상 제작에 활용할 수 있도록 했다.구글은 5일(현지시간) 동영상 생성 AI 모델 ‘제미나이 옴니 1.2 플래시(Gemini Omni 1.2 Flash)’를 출시했다.이번 모델은 제미나이 API와 기업용 서비스에서 정식 제공된다. 720p 영상 생성 비용은 초당 0.07달러로, 전작인 옴니 1.1보다 30% 저렴하다. 360p나 720p로 초안을 빠르게 만든 뒤 필요
Ukrainian authorities said on Friday that one man had been killed in a morning attack on the Rivne region. Three others were also reportedly injured in strikes on Kyiv Oblast overnight.