Latest news with #Isambard-AI

Miami Herald
22-07-2025
- Business
- Miami Herald
Nvidia move deals a major blow to AMD, Intel, and ARM
Nvidia climbed to become the largest company on the U.S. stock market. But how can it keep growing after hitting a $4 trillion market capitalization? Every company reaches its peak at some point. Are we witnessing Nvidia's, or can it hit $5 trillion? Don't miss the move: Subscribe to TheStreet's free daily newsletter That is a tricky question to answer, but the company thinks it can, and there is a lot more juice to be squeezed out of the artificial intelligence boom. The U.S. government recently reversed its export restrictions on Nvidia cards to China and assured the company that licenses will be granted. This change prompted many analysts to revamp the stock price target. Related: Nvidia-backed stock sinks on unexpected deal "China generated $17 billion in revenue for Nvidia in 2024, roughly 13% of the company's total revenue. In response to this news, we are lifting our NVDA price target to $200 from $185, and reiterating our One rating," writes veteran fund manager Chris Versace for TheStreet Pro. The company wasn't sitting down idly and waiting for this China miracle. It was making moves in Europe. Image source: Nvidia On July 17, Nvidia (NVDA) revealed on its blog that the University of Bristol launched its Isambard-AI supercomputer. Nvidia Grace Hopper Superchips power it and deliver 21 exaflops of AI performance. While this supercomputer's 5,448 GH200 Superchips are impressive, they can't match the Jupiter supercomputer announced in June. Jupiter is also powered by the Grace Hopper platform, with nearly 24,000 GH200 Superchips. With its expected performance of over 90 exaflops, it is the fastest in Europe. Finding new customers for new supercomputers must reach a slowdown at some point. Nvidia knows this, and it's betting on robotics being the next big boom. Related: Unexpected chip deal may reshape the semiconductor market Jetson Thor is Nvidia's upcoming platform for physical AI and humanoid robotics. According to the company, it will be able to deliver up to 2070 FP4 TFLOPS of AI compute. This platform features a 14-core Arm Neoverse-V3AE 64-bit CPU and a 2560-core NVIDIA Blackwell GPU. What is often neglected in the AI GPU hype is that you need a CPU to actually run the software that will then use GPUs for calculations. Nvidia doesn't have its own CPU architecture and has to rely on ARM for this embedded (Thor) platform. The same goes for its Grace Superchips; for the CPU part, they have 72 high-performance Arm v9 cores. Nvidia tried to acquire ARM but was prevented by regulatory problems. The primary reason for the acquisition may have been licensing costs. Nvidia Hardware Engineering VP Frans Sijstermans presented his "Enabling RISC-V Application Processors in NVIDIA Compute Platforms" at RISC-V Summit China, revealing that CUDA is coming to RISC-V. RISC-V is an open source instruction set architecture. This means anyone can make a CPU based on it, and the company does not have to pay for the license, although the company that makes its RISC-V compatible design can license its design. "This port will enable a RISC-V CPU to be the main application processor in a CUDA-based AI system," said RISC-V International. More Nvidia: Fund manager who predicted Nvidia stock rally reboots forecast on ChinaMajor analyst revamps Nvidia stock price target after China surpriseNvidia CEO hits Warren Buffett milestone The timing of this Nvidia move is very interesting. In an attempt to lower the country's dependence on Western-owned technology, China plans to issue guidance to boost the use of open-source RISC-V chips nationwide, reported Reuters in March. Also in March, XuanTie, part of Alibaba's DAMO Academy R&D operation, announced a C930 CPU design, reported The Register. This RISC-V-based design is available to license for system-on-chip makers. The company is marketing its CPU as something that can be used in servers, PCs, and autonomous cars. This is not the only RISC-V-based CPU available, but it is probably the one with the strongest backing, making China's push to switch to RISC-V look more credible. Nvidia is ensuring that China continues to rely on its GPUs by supporting RISC-V, which is why they are porting CUDA. However, this will also strengthen the viability of the RISC-V platform as a whole and therefore hurt x86 and ARM. The company also has experience with RISC-V; in 2016, it switched from its proprietary Falcon microprocessor, which was used as a logic controller in its GPUs, to RISC-V. This port also signals Nvidia's possible switch to RISC-V for its CPU cores. If the company can pull it off, it would bring a lot of savings from not having to pay for ARM licensing. Related: Meta delivers eye-popping AI announcement The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.


Express Tribune
18-07-2025
- Science
- Express Tribune
Isambard-AI, UK's most powerful supercomputer, goes live in Bristol
A supercomputer, which is now the most powerful in the UK, has been fully activated in Bristol. The Isambard-AI machine, a major leap forward in the country's technological capabilities, was launched by Technology Secretary Peter Kyle, as part of the government's fresh push towards expanding artificial intelligence (AI) resources, as reported by BBC. Kyle officially 'flicked the switch' on the machine during an announcement that also unveiled new AI-related initiatives. The Isambard-AI supercomputer will join a similar machine in Cambridge, called Dawn, as part of the UK's AI Research Resource, which will serve as a vital tool for public sector projects. The supercomputers are expected to aid in tackling critical issues like reducing NHS waiting lists and developing innovative solutions for climate change, despite the known high energy consumption of AI systems. Alongside these announcements, the government revealed plans for substantial investments in AI Growth Zones across Scotland and Wales, aiming to boost regional economies and expand AI's reach nationwide. The Isambard-AI machine had already been involved in a medical project earlier this year, contributing to vaccine development. Now fully operational, it is expected to support a broad range of public sector initiatives. As its name implies, Isambard-AI is a supercomputer designed to complete tasks at an unprecedented speed, utilising thousands of processing units to handle larger datasets faster than traditional computers. This extraordinary capability will enhance the UK's public AI computing infrastructure. The government has announced a training initiative aimed at preparing a million students in AI and providing broad economic training for 7.5 million individuals over the coming years. Isambard-AI, using more than 5,400 Nvidia GH200 Grace Hopper Superchips, was developed by the University of Bristol with public funding, as reported by BBC. It ranks 11th in the world in terms of processing power, according to the most recent Top 500 list of supercomputers. In contrast, Dawn at the University of Cambridge employs over 1,000 Intel chips. David Hogan, Nvidia's European vice president, described Isambard-AI as a "truly transformational machine" but stressed that it is just the beginning of a much broader AI strategy. To bolster these efforts, researchers, academics, and tech leaders, including Google DeepMind's Pushmeet Kohli and the Royal Society's Alison Noble, are collaborating on a national AI strategy set to be unveiled later this year. The UK government believes that further investment in supercomputers will help position the country as a leader in AI development, challenging the global race for dominance in the sector. This move is also aimed at enhancing the UK's competitiveness in the field, with global companies eager to secure the best talent and resources in the AI space.


Business Insider
18-07-2025
- Business
- Business Insider
Nvidia (NVDA) and University of Bristol Launch the UK's Most Powerful AI Supercomputer
Chipmaker Nvidia (NVDA) and the University of Bristol have launched the Isambard- AI supercomputer, which is now the most powerful in the UK and one of the world's most energy-efficient. Interestingly, it uses 5,448 Grace Hopper superchips packed into liquid-cooled HPE (HPE) server cabinets, with 440 GPUs each. As a result, this powerful system will help researchers take on a wide range of projects, from drug development and materials science to building large language models for UK-specific languages like Welsh. Elevate Your Investing Strategy: Take advantage of TipRanks Premium at 50% off! Unlock powerful investing tools, advanced data, and expert analyst insights to help you invest with confidence. Make smarter investment decisions with TipRanks' Smart Investor Picks, delivered to your inbox every week. Although the supercomputer is not as large as the huge data centers used by companies like Amazon (AMZN) and Microsoft (MSFT) —which require gigawatts of power—Isambard-AI uses just 5 megawatts while still being able to run many experiments at once. In addition, the Bristol team chose liquid cooling instead of fans, which allows more GPUs to be placed into smaller spaces while improving energy efficiency. The university also plans to reuse the hot wastewater from the supercomputer to heat campus buildings and nearby homes. It is worth noting that this project is part of Nvidia's 'sovereign AI' focus, which is where countries build their own national AI infrastructure. Indeed, in May, CEO Jensen Huang joined President Trump at a Saudi event to announce that Nvidia would sell thousands of GPUs to companies in the country. Moreover, in June, Nvidia said that it was expanding its AI infrastructure push into France, Germany, Italy, and the UK. This is because Huang believes that AI is the new essential infrastructure, just like electricity and the internet once were. What Is a Good Price for NVDA? Turning to Wall Street, analysts have a Strong Buy consensus rating on NVDA stock based on 34 Buys, three Holds, and one Sell assigned in the past three months, as indicated by the graphic below. Furthermore, the average NVDA price target of $182.06 per share implies 5% upside potential.


South Wales Guardian
17-07-2025
- Business
- South Wales Guardian
UK's most powerful supercomputer comes online in major AI drive
Technology Secretary Peter Kyle flicked the switch on the Isambard-AI machine in Bristol on Thursday, in a move ministers say will help the UK develop new medical cures and tools to cut emissions. The Government has pledged £1 billion to increase Britain's compute capacity 20-fold by 2030, including through the creation of a series of AI 'growth zones' designed to hasten planning approvals for new data centres. One of these will be built in Scotland, where Chancellor Rachel Reeves has also confirmed £750 million of funding will be dedicated to developing another supercomputer in Edinburgh, and another in Wales. Together with a second existing supercomputer in Cambridge, Isambard is expected to be able to process in one second 'what it would take the entire global population 80 years to achieve', the Government said. Businesses and scientists are expected to be able to use the systems to process more of the data required to train and build AI models to make new drug discoveries and breakthroughs in climate change technology. Researchers at the University of Liverpool are already using the machine to sift through tens of millions of chemical combinations in the hopes of finding ways to decarbonise British industry. The plans form part of the new Compute Roadmap, a strategy aimed at reducing reliance on foreign processing power and transform the UK's public compute capacity. By 2030, the Government expects this capacity to increase to 420 AI exaFLOP – the equivalent of one billion people spending 13,316 years doing what the system will do in one second. To support the plans, researchers, academics and tech bosses have been brought together to develop an AI science strategy to be published in the autumn. The group includes Google DeepMind vice-president Pushmeet Kohli, vice-president of the Royal Society Alison Noble and chairwoman of the Engineering and Physical Sciences Research Council Charlotte Deane. Ms Reeves said the plans would 'transform our public services, drive innovation and fuel economic growth that puts money in people's pockets'. Mr Kyle said they would 'put a rocket under our brilliant researchers, scientists and engineers – giving them the tools they need to make Britain the best place to do their work.'

South Wales Argus
17-07-2025
- Business
- South Wales Argus
UK's most powerful supercomputer comes online in major AI drive
Technology Secretary Peter Kyle flicked the switch on the Isambard-AI machine in Bristol on Thursday, in a move ministers say will help the UK develop new medical cures and tools to cut emissions. The Government has pledged £1 billion to increase Britain's compute capacity 20-fold by 2030, including through the creation of a series of AI 'growth zones' designed to hasten planning approvals for new data centres. The Technology Secretary said the plans would 'put a rocket under' the work of UK researchers One of these will be built in Scotland, where Chancellor Rachel Reeves has also confirmed £750 million of funding will be dedicated to developing another supercomputer in Edinburgh, and another in Wales. Together with a second existing supercomputer in Cambridge, Isambard is expected to be able to process in one second 'what it would take the entire global population 80 years to achieve', the Government said. Businesses and scientists are expected to be able to use the systems to process more of the data required to train and build AI models to make new drug discoveries and breakthroughs in climate change technology. Researchers at the University of Liverpool are already using the machine to sift through tens of millions of chemical combinations in the hopes of finding ways to decarbonise British industry. The plans form part of the new Compute Roadmap, a strategy aimed at reducing reliance on foreign processing power and transform the UK's public compute capacity. By 2030, the Government expects this capacity to increase to 420 AI exaFLOP – the equivalent of one billion people spending 13,316 years doing what the system will do in one second. To support the plans, researchers, academics and tech bosses have been brought together to develop an AI science strategy to be published in the autumn. The group includes Google DeepMind vice-president Pushmeet Kohli, vice-president of the Royal Society Alison Noble and chairwoman of the Engineering and Physical Sciences Research Council Charlotte Deane. Ms Reeves said the plans would 'transform our public services, drive innovation and fuel economic growth that puts money in people's pockets'. Mr Kyle said they would 'put a rocket under our brilliant researchers, scientists and engineers – giving them the tools they need to make Britain the best place to do their work.'