Everybody knows memory has grown dramatically short in supply and therefore more expensive. Now the impact is being felt where the rubber meets the road with edge practitioners.
Professional engineers who design and deploy edge applications that rely on sensors in IoT systems alongside memory are speaking out about the pain they feel. They use terms more graphic and emotional than the ivory tower financial analysts and executives at memory manufacturers who have been warning of the memory crunch for more than a year.
“There’s memory in everything, and we’re not just talking DRAM and PC memory. We’re talking memory at microcontroller levels and every type of memory is affected in some way” by memory shortages and price increases, said Andy Do, president and CEO and CEO of Embedded Works and president of SensorsWorks.
Do, an IoT and sensors evangelist, is actively involved with his companies installing sensors in multiple projects, including a recent one calling for hundreds of end point sensors to retrofit an NFL stadium to monitor sewer covers to prevent illegal entry and to assess multiple restrooms for better maintenance.
“Yes, the memory crunch has affected me,” he told a group of sensors professionals meeting as the Sensors Converge Advisory Board last week. “It’s not price levels of the Covid era but has gotten to the point where my prices have gone up. And suppliers don’t know how long it’s going to last.” He said one large networking OEM had told him price hikes might last until Q1 of 2027. “So go buy Micron stock,” he suggested.
In an interview with Fierce, Do said he is seeing vendors in the IoT sector give him advanced notice that prices for good with embedded memory will increase up to 10% effective immediately and could increase. Meanwhile, demand is growing for IoT sensors used in retails stores, service business like HVAC, and landscaping companies that need to better manage assets. “Our clients’ adoption of IoT and sensors is not just a nice to have technology, but one which gives them a competitive edge and for some to use the data they’ve collected so they don’t have AI FOMO,” he said.
Indeed, memory prices have seen an extreme, record-breaking surge, according to multiple sources. Some RAM and SSD costs have risen up to 400%. Market analysts describe what’s happening as an unprecedented surge and blame the massive demand on an explosion of AI data centers. Micron and other memory makers have shifted their focus to making HBM (High Bandwidth Memory) for AI with less production for memory products that service microcontrollers in edge applications, places where sensors and processors and memory work together, often on the same printed circuit board.
Micron, SK Hynix and Samsung Electronics make up nearly the entire RAM market and their companies have benefited from the demand. Micron’s stock is more than 200% over the past year. “We have seen a very sharp, significant surge in demand for memory and it has far outpaced our ability to supply that memory and in our estimation, the supply capability of the whole memory industry,” Micron business chief Sumit Sadana told CNBC in early January.
“The whole memory thing is affecting everyone in some way, shape or another, but especially with sensors,” said Azita Arvani, founder of Arvani Group and past CEO of Rakuten Symphony Americas. She consults with companies involved in edge AI projects. “If you’re adding more capability to sensors, if you want to grab more data, if you want to do more AI on device, if you want to add new certifications for security or confidentiality, all of that means have to have more memory when you capture the data, so the [memory crunch] definitely affects it.”
In an interview with Fierce, Arvani added, “We see memory shortages are being driven by AI infrastructure with data center consuming much of the supply, causing DRAM/NAND costs to rise materially. The impact will be on smart sensors.”
She added: “Smart sensors manufacturers, especially those embedding AI on-chip are exposed. Image and motion sensors, for example, Sony’s IMX500 or Bosch’s BHI260AP, integrate memory to do processing. While these companies haven’t reported [shortage] issues yet, their designs rely on stable memory supply. The impact will be higher memory prices that can delay new sensor launches or force leaner designs.”
Arvani said smart cameras running AI locally or industrial motion sensors processing data on device that depend on memory could see supply constraints and rising costs or delayed enhancements. “Memory strategy is critical for sensor firms,” she said. “We have already heard that device makers like Apple will be impacted by this supply chain issue.”
No single sensor firm has sounded alarm bells about the memory crunch but Arvani added, “the trend is unmistakable. Any advanced sensor leveraging embedded memory, whether in cameras, industrial IoT, phones or wearables, will need to navigate these memory-driven headwinds wisely.”
One engineer for a major consumer appliance maker said it is difficult to plan annual home appliance upgrades that add more features when memory might have to be reduced in capacity due to added cost. “It could mean a tradeoff with features and sensors,” he said. For a future product, “you already better have provisioned a lot of memory and a lot of other stuff, otherwise your product will become a brick in the market.”