@HPC Podcast Archives - OrionX.net - HPC News Bytes – 20260713

Episode Date: July 13, 2026

- Meta Iris accelerator chip - New class of vendors in Chips - ORNL on-prem Quantum Computer - Satoshi Matsuoka’s papers on Precision, Memory, Open Models - Is the U.S. “blowing it” in AI? [audio mp3="https://orionx.net/wp-content/uploads/2026/07/HPCNB_20260713.mp3"][/audio] The post HPC News Bytes – 20260713 appeared first on OrionX.net.

Welcome to HPC Newsbytes, a weekly show about important news in the world of supercomputing, AI, quantum computing, and other advanced technologies. Hi, everyone. Welcome to HBC Newsbytes. I'm Doug Black, and with me, of course, is Shane Khan. There's a rising tide of GPUs on the AI processing market, many of them from some of the richest companies in the world, but the degree to which they pose a threat to NVIDIA's GPU market dominance is very much in question. There was news last week that meta plans to have its new AI chip in production in the September timeframe, according to a Reuters story. The report is based on an internal meta memo. The new chip is called Iris announced four months ago, and it's the first in a

series of meta training and inference accelerator processors to be rolled out over the next year and a half, and they will target generative AI inferencing workloads. Meta's iris chip is one piece of a larger move of meta and other hyperscalers toward owning and optimizing more of their own AI infrastructure stack. That really is the big trend and the big signal here. Broad brush, if you have enough volume, you can build your own chips and systems. Hyperscalers can build their own chips and systems and software, and they increasingly do, while continuing to buy from merchant vendors what they must. And chip vendors, likewise, are becoming system and rack-scale vendors, and invest in cloud providers who are strongly aligned with them. Meta is working with Broadcom on custom

silicon and TSM on fabrication and packaging, while continuing to buy heavily from NVIDIA, AMD, and Intel for merchant compute. The supporting supply chain is just as important. Samsung, SK Hynix, Micron, and Sandysk for memory and storage, Sumitomo Electric for optical interconnect components, W-WIN, Quanta, and Foxcon for servers, motherboards, and racks. The layers point to another market signal here. Hyperscalers building their own hardware and the growing complexity of high-end chips are coming together to change the supply chain for chips. Broadcom, Marvell, and increasingly Media Tech are moving up the stack from implementation partners towards full-service custom chip prime contractors.

Hyperscalers increasingly specify the workloads, economics and system requirements, while specialist partners take on more of the architecture, design, packaging, and production. That could become a major new layer in AI infrastructure. The Oakridge Leadership Computing Facility announced last week that they're deploying a 20-cubit system from IQM quantum computer. Oakridge said the IQM Radiant System, which is named Pathfinder, will play a key role in the lab's efforts to integrate quantum computing technology with their classical HBC resources. IQM was founded in 2018 and is based in Finland. Their radiant system is based on superconducting technology, which means its qubits must be cool to nearly absolute zero.

The Oak Ridge installation reinforces a broader shift already on their way in the U.S. Quantum computing is moving from experiment and engineering into procurement and operations. The 20-cubit system from IQM is modest. They have bigger systems that they sell to. but the strategic value is in integrating quantum hardware with HPC workflows, schedulers, and applications while developing in-house expertise. The larger market signal is also important. We can now identify more than 30 quantum computing vendors

across eight hardware modalities that have collectively built or deployed well over 100 systems, including internal research platforms, prototypes, cloud systems, and customer-installed machines. IQM for its part has said that it has sold 23 systems and installed 15. HBC Luminary Satoshi Matsuoka of Japan's Rican Center for Computational Science has issued three papers of late that, given his stature in the supercomputing industry and the content of his articles has stirred the pot in the HBC community. Matsuo, of course, is a leader in Japanese super computing system strategy and design, including the country's upcoming Fugaku Next leadership system scheduled for 2029 or 2030. His three papers, all available on the archive site,

take on the topics of floating point 64 processing as the HPC Holy Grail. Another argues that FP8 is all you need, and the third looks at the scarcity of memory chips, at open AI models, and at the restructuring of the AI industry through the rest of the decade. We need to cover these topics in more detail, but I will share my high-level take from these papers. On numerical precision, if you assume that most scientific computation can be formulated as matrix and tensor algebra, a reasonable assumption, then algorithmic emulation can do the job faster. and if the fraction of computation that is not matrix algebra is small enough, then it can be done in slower ways and you still come out ahead.

On the memory issues, I see the paper making the following salient points. A trend towards memory bandwidth versus compute. Memory bandwidth is what people are in fact buying, the paper says. Older infrastructure can continue to work well and could dampen the need to upgrade, and especially in situations where the infrastructure that runs, service is not visible to the customer. AI inference costs will trend towards commodity because of open models.

For economic growth, token demand must continue to outstrip new efficiencies. Infference will expand to the edge. AI for science and HPC provide a steady demand floor for capacity, and China can compete. As usual, these are high-quality food for thought. The Wall Street Journal ran a contributed piece from Jim Vandehi, a co-founder of the Axios News platform, in which he contends that the U.S. is at risk of blowing the AI race. His article proposes seven ideas to get America back on track. The article reflects the growing significance on the world stage of AI and its encroachment beyond tech markets into areas of national policy and geopolitics.

Van de Haidtons that, quote, America faces one of those rare historic moments when government, business, schools, and families could be working together to meet a truly generational challenge winning the AI race. So far, we're blowing it, he says. At the core of his argument is that the awesome technological advances of world-leading American engineers at Anthropic, Open AI, and elsewhere could be zipping the country further ahead of China. Instead, it's locked in the panicked and confused paralysis across government, business, campuses, and the workforce. China, meanwhile, has a state-directed countrywide plan to put AI into action and lock down the supply chain for future dominance. This threat is present, real, and intensifying. Shaheen, very interested to hear your take on the article. Right. Well, I'll be the contrarian as I sometimes have to be. The articles claim that America is, quote, blowing the AI race is, well, shall we say, unsupported. The U.S. leads in spending, advanced AI models, much of the stack from chips to apps, and commercialization.

If there are any constraints, they are due to massive adoption and investment that is causing shortages. That does not mean there are no constraints worth addressing. The most credible ones are, one, power and grid expansion. two, attracting more top-tier math and science talent, and three, continued focus on advanced chip manufacturing in the U.S. Even the chip risk is mitigated by Intel's rapidly advancing domestic capability and capacity, and other providers of high-end capacity, Taiwan, South Korea, and soon, Japan, are U.S. allies. So while you can always do better and should, and other countries can make advances,

I just don't see anyone being particularly close to the U.S. in AI at all. All right, that's it for this episode. Thank you all for being with us. HPC Newsbytes is a production of OrionX. Shaheen Khan and Doug Black host the show. Every episode is posted on OrionX.net. If you like the show, please rate and review it. Thank you for listening.