Apple is collaborating with Broadcom to develop its first server chip for AI processing, aimed at reducing reliance on Nvidia
Image: Shutterstock
Apple has teamed up with Broadcom to develop an AI server chip
Apple is collaborating with Broadcom to develop its first server chip for AI processing, aimed at reducing reliance on Nvidia’s expensive and scarce GPUs, Reuters reports citing news first reported by The Information.
The AI chip, code-named Baltra, is expected to enter mass production by 2026 using Taiwan Semiconductor Manufacturing’s advanced N3P process. This aligns Apple with other tech giants designing their own chips for AI services. Apple plans to use these chips to power AI features in its devices.
Broadcom’s stock rose 5 percent following the announcement, and the company benefits from cloud providers diversifying their supply chains. The custom chip market is forecast to grow to $45 billion by 2028, with Broadcom and Marvell emerging as key players.
AI to drive double-digit growth in chip industry in 2025
The global semiconductor industry is set for double-digit growth in 2025, driven by rising demand for artificial intelligence (AI) and high-performance computing (HPC), according to market researcher IDC. AI will fuel demand for advanced chips, including 2nm technology and high-bandwidth memory (HBM), with memory segments growing by over 24 percent.
Major players like TSMC, Samsung, and Intel are expanding their advanced node capacities, including 2nm chips. The semiconductor supply chain, spanning design, manufacturing, testing, and packaging, will see significant growth, particularly in Asia-Pacific.
Packaging and testing industries in China and Taiwan are benefiting from geopolitical shifts, while AI-driven advancements are boosting demand for cutting-edge packaging technologies. Overall, IDC forecasts a 15 percent growth in the semiconductor market in 2025.
TCS extends contract to manage IT infra for Danish telco Telenor
Tata Consultancy Services (TCS) has won an extension to its contract with Telenor Denmark to manage the telecom provider’s IT infrastructure for the next five years. This follows a successful six-year collaboration, during which TCS modernised Telenor’s infrastructure using a machine-first delivery model, TCS said in a press release.
TCS, India’s biggest IT company, didn’t provide details on the value of the contract. The extended partnership will leverage TCS’s automation technologies to drive cost reductions and improve operational efficiency for Telenor Denmark, which serves over 1.6 million subscribers. The telecom company’s CTO, Louise Haurum, said the collaboration will help keep its IT systems resilient and drive innovation.
TCS will continue to ensure business continuity, using its European delivery centre to enhance Telenor Denmark’s digital assets and user experiences. The partnership solidifies TCS’s position as a leading vendor in the Nordics telecom sector.
Infosys, RheinEnergie partner to support customers’ sustainable energy goals
Infosys has partnered with RheinEnergie to support the German energy services provider’s enterprise customers in their energy transition and sustainability goals. The collaboration will use Infosys’ cloud solutions for the energy sector, part of its Cobalt cloud platform and Topaz AI suite to enhance energy efficiency, cut costs, and drive decarbonisation, the Bengaluru IT company said in a press release.
Together, the companies will develop solutions to optimise energy use, decentralise the energy transition, and improve decision-making on renewable energy investments of customers. The partnership will target industries such as transportation, commercial real estate, and manufacturing. With a focus on cloud, AI, and Industry 4.0, the collaboration aims to help enterprises boost energy efficiency by up to 40 percent.
Harvard to release 1 million books as AI training dataset
Harvard University, with funding from OpenAI and Microsoft, is releasing a large free AI training dataset consisting of nearly a million public-domain books, Wired reports. The project, called the Institutional Data Initiative (IDI), aims to “level the playing field” in the AI industry Wired cites the leaders of the effort as saying.
The dataset contains books scanned as part of the Google Books project that are no longer protected by copyright. And it is around five times the size of the Books3 dataset that was used to train AI models like Meta’s Llama. It includes classics from Shakespeare, Charles Dickens, and Dante alongside obscure Czech math textbooks and Welsh pocket dictionaries, according to Wired.
In collaboration with the Boston Public Library, IDI is also scanning public-domain newspaper articles. The dataset’s distribution is still being finalised. This effort is part of a broader trend of creating public-domain datasets for AI, with other initiatives like Pleias’ Common Corpus and Spawning’s Source.Plus also emerging. Critics, however, argue that these datasets must be used alongside licensed data to truly change the status quo and prevent reliance on scraped copyrighted material, according to Wired.
