{
  "id": 139714,
  "title": "Designing memory-centric chips for future space telescopes",
  "url": "https://urgent.news/2026/08/04/designing-memory-centric-chips-for-future-space-telescopes",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-08-04T16:40:01.000Z",
  "source": {
    "name": "Phys.org",
    "slug": "phys-org",
    "url": "https://phys.org/news/2026-08-memory-centric-chips-future-space.html"
  },
  "original_language": "en",
  "account": "Two innovative computer chip designs could efficiently process data for future space telescopes while searching for Earth-like exoplanets, according to a University of Michigan Engineering study set to be presented at the IEEE Space Computing Conference in August. The Habitable Worlds Observatory, a proposed NASA space telescope, would orbit the Sun-Earth L2 point, allowing a single sunshield to block light and heat from Earth, the moon, and the sun. Real-time data processing and correction of optical distortions are crucial for imaging Earth-like exoplanets, but current radiation-hardened space processors are too slow, and powerful GPU-based systems consume too much power for space applications. The key insight from the research is that future space missions require the right type of computing power, not just more of it. The team demonstrated that custom memory-centric chips could drastically reduce power consumption, with the SRAM architecture cutting power use by up to 59 times compared to GPU-based alternatives. The research team proposed two custom hardware architectures: the high-bandwidth memory (HBM) approach and the distributed SRAM design. The HBM layout uses 27 HBM chips horizontally connected to a custom processor, creating a superhighway for data transfer, while the distributed SRAM design splits data and processing across 56 chiplets, each with 2 GB of SRAM, bringing the processing \"office\" to the \"warehouse\" of data. These architectures reduced power demands from 3,000 watts for GPUs to 928 watts for HBM and 51 watts for SRAM plus reduced precision. The weight reduction from 1,100 kg for a GPU-based system to as little as 193 kg for the SRAM design could save $430 million over a 25-year space mission. The researchers plan to manufacture the SRAM chiplets and test them in a laboratory setting.",
  "summary": "Two proposed computer chip designs could efficiently process data on future space telescopes while they search for Earth–like exoplanets, according to a University of Michigan Engineering study that will be presented at the IEEE Space Computing Conference in August.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}