Good morning. It's Tuesday, July 14, and welcome to this week's Science & Technology newsletter. First time reading? Sign up here or forward to share with friends.
For over a year, consumer electronics have seen substantial price increases as the expansion of data centers has led to shortages of various computer components, including memory and graphics processing units. This week, we're taking a look at the latter to understand what this technology is, its history, and why it's so vital to AI.
PS—Reader feedback is a gift! Whether it's feedback on today's email, suggestions for what we should cover, or anything else, we're happy to hear from readers. Simply reply to this email or reach out at science-technology@join1440.com.
—Marco Daniel Machado, 1440 Science & Technology Section Editor |
|
|
|
A graphics processing unit is computer hardware designed to simultaneously perform the same operations on many pieces of data. While the central processing unit is responsible for general-purpose functions by executing various instructions, one at a time, GPUs are designed to run multiple instances of the same task in parallel. This makes GPUs efficient at handling repetitive, specialized workloads, such as processing the color of millions of pixels dozens of times per second.
> View how the number and organization of processing cores distinguish CPUs from GPUs. (More)
> Take a virtual tour inside a GPU to learn what enables it to perform over 30 trillion calculations per second. (More)
Before the 1990s, manufacturers of components such as circuit boards for video game consoles and arcade cabinets relied on CPUs and custom parts called accelerators to quickly process visual data. However, the rise of 3D gaming introduced performance bottlenecks as CPUs struggled to render more complex polygons, textures, and lighting effects. This motivated the development of dedicated graphics microchips, and, in 1999, Nvidia marketed its GeForce 256 as the world's first GPU.
> Watch a tech demo of the GeForce 256 rendering grass and clouds. (More)
> GPU once stood for geometry processing unit. (More)
> ... and the acronym is often used to refer to graphics cards, not microchips. (More)
GPUs have since facilitated a diverse range of workflows, including video effects and editing, 3D prototyping and animation, weather and climate forecasting, and high-frequency trading. Because they rely on substantial parallel processing, the large language models underlying AI have increased the demand and cost of GPUs and their supporting components, including high-speed memory.
> GPUs became general-purpose components thanks to a Stanford doctoral student. (More)
> See how GPUs enable AI chatbots by processing millions of text parameters simultaneously. (More)
Discover more:
> Memory is organized within GPUs according to ergonomic principles. (More)
> Manufacturers like Nvidia deliberately sell GPUs with defective components, advertising them as lower-tier products. (More)
> Buying a new computer? Learn if you need to purchase a dedicated GPU. (More)
Keep learning by exploring all our findings on the Graphics Processing Unit here. |
|
|
|
|
In partnership with Tello |
|
Phone Coverage That Doesn't Have an Agenda |
|
Historically, major cell carriers had the best coverage … but that’s no longer the case today. In fact, smaller carriers tap into the same towers as the “Big Three”, so you’re getting the same coverage at a more affordable price.
Case in point? Tello offers affordable phone plans without sacrificing service quality. All of those other sacrifices you make with the larger carriers—locked contracts, hidden fees, increasing prices? Not with Tello. Their Unlimited Everything Plan is just $25/month—unlimited talk, text, and data, plus free hotspot, all on nationwide 5G coverage. Prefer something lighter? You have lower plans options such as 10 GB for $15/month, with no contracts and no lock-ins, ever.
Even better: New customers can bring their own phone and number. Tello keeps it simple: build the plan you want, change it when your needs shift, and pay exactly what you expected. See what your next phone bill could look like. |
|
Please support our sponsors! |
|
|
|
|
1440 brings you the knowledge and context behind the week's stories:
SpaceX launched the first nuclear-powered commercial satellite, which creates electricity from the decay of tritium. (More)
> See how electricity can be generated on spacecraft using radioactive elements.
> Tritium is a radioactive isotope of hydrogen that also powers some watches and exit signs.
Large-scale seafloor spreading was directly observed for the first time when monitoring the Australian-Antarctic plate boundary. (More)
> Patterns of magnetic rocks provided the first evidence of seafloor spreading.
> The discovery of the Mid-Atlantic Ridge, where tectonic plates move apart, was once dismissed as "girl talk."
Invasive Manila clams are now everywhere in the Northern Hemisphere, after being identified along the Massachusetts coastline. (More)
> The Manila clam was accidentally spread by human activity, the most common method of spread for invasive species.
> In New England, people are encouraged to catch invasive green crabs, which are used for bait, broths, and crab whiskey.
The Supreme Court allowed Texas to prohibit minors from downloading apps without parental consent. (More)
> Read Texas's App Store Accountability Act.
> Learn about the strategies and challenges to implementing age verification. |
|
|
|
|
Like all great researchers, we love spending time learning about the latest breakthroughs across science and technology. Here's what we found this week.
Starfish reveal how organs grow thanks to their transparent larval stage.
Marine Biological Laboratory | Camille Ledoux. Like the human heart and lungs, organs in many species of sea stars begin as hollow tubes—hydro-vascular organs—that later expand. Analyzing the closest invertebrate relatives of humans may help grow the understanding of organ development across various animal groups. (More)
> Starfish anatomy includes two stomachs and eye spots at the end of each arm.
> Much of your body was shaped by a protein named after Sonic the Hedgehog.
The first open-source simulation environment for space robots has been launched.
Rice University | Silvia Cernea Clark. Developed alongside researchers at NASA's Lyndon B. Johnson Space Center, the iMETRO Dynamic Simulation serves as a digital twin of the physical iMETRO facility and enables the global robotics community to remotely build and deploy software to operate robots at the physical site. (More)
> Since 2019, a robotic system named Astrobee has helped astronauts with chores on the International Space Station.
> ... but the first humanoid robot in space was NASA's Robonaut 2.
Nuclear weapons in space may be detectable by a satellite-based sensor system.
MIT | Zach Winn. Roughly the size of nine Rubik's Cubes, the proposed system would find a weapon by observing neutrons produced from collisions between high-energy protons and radioactive material. Calculations suggest 99% detection accuracy is achievable by orbiting within 4 kilometers of a suspect satellite for a week. (More)
> The high-energy protons the system relies on are trapped by Earth's magnetic field in "radiation doughnuts."
> See what happened in 1962 when the US detonated a nuclear weapon in space.
AI-generated faces found to be more trustworthy than real ones.
Lancaster University | Staff. In the first-ever study of its kind, 169 participants rated the average synthetic face created with the latest diffusion technology as 16.6% more trustworthy than the average real face, raising concerns over their use in online fraud. Despite faces from the latest model appearing most trustworthy, participants also rated them less realistic than those produced from older models. (More)
> How do the diffusion models underlying AI image tools work?
> Our comfort with faces as they approach human likeness may be limited by the uncanny valley. |
|
|
|
|
In partnership with Tello |
|
No Contract. No Catch. No Kidding. |
|
Please support our sponsors! |
|
|
|
|
🔭 Want more? Explore over 100 topics and thousands of resources at 1440's hub for Science & Technology. |
|
|
"That's called CEO math. CEO math is not accurate, but it is correct: The more [GPUs] you buy, the more you save." |
—Jensen Huang, Nvidia CEO |
|
|
Behind the Name. Why 1440? The printing press was invented around the year 1440, spreading knowledge to the masses and changing the course of history. More facts: In every day, there are 1,440 minutes. We’re here to make each one count.
Want to connect with 4.7 million insatiably curious minds? Become a 1440 partner here. |
|
1440 Media 222 W Merchandise Mart Plaza, Suite 1212 Chicago, IL 60654
Copyright © 2026, 1440 Media, All rights reserved. |
|
|
|
|
|
|
|
