#amazon accelerate

[jcmarchi]

Neural Processing Units (NPUs): The Driving Force Behind Next-Generation AI and Computing

New Post has been published on https://thedigitalinsider.com/neural-processing-units-npus-the-driving-force-behind-next-generation-ai-and-computing/

Neural Processing Units (NPUs): The Driving Force Behind Next-Generation AI and Computing

Just as GPUs once eclipsed CPUs for AI workloads, Neural Processing Units (NPUs) are set to challenge GPUs by delivering even faster, more efficient performance—especially for generative AI, where massive real-time processing must happen at lightning speed and at lower cost.

The question is how do NPUs work, and why are they edging out their GPU predecessors for modern AI tasks, and what makes them indispensable for everything from robust data center infrastructure to everyday consumer devices? Whether you’re strategizing your next big AI deployment or simply curious about the cutting edge of tech, it’s important to understand why NPUs could be the breakthrough that redefines AI—and the next generation of computing.

What Is a Neural Processing Unit (NPU)?

A Neural Processing Unit (NPU) is a specialized microprocessor built from the ground up to handle the unique requirements of modern AI and machine learning workloads. While Central Processing Units (CPUs) and Graphics Processing Units (GPUs) have historically powered traditional computing tasks and graphics rendering, they were not originally designed to tackle the computational intensity of deep neural networks. NPUs fill this gap by focusing specifically on parallel, high-throughput operations such as matrix multiplications and tensor math—the underpinnings of AI models.

Key aspects that differentiate NPUs from general-purpose CPUs and GPUs include:

Optimized AI Arithmetic: NPUs commonly use low-precision data types (e.g., 8-bit integer math, or even lower) to balance processing power and energy efficiency, while CPUs and GPUs typically rely on higher-precision floating-point calculations.

Parallelized Architecture: NPUs can break down AI tasks into thousands (or even millions) of smaller computations that run concurrently, dramatically increasing throughput.

Energy Efficiency: By eliminating unnecessary instructions and optimizing specifically for neural network tasks, NPUs can achieve higher performance at lower power compared to GPUs or CPUs performing the same AI workloads.

Also known as AI accelerators, NPUs often appear as discrete hardware attached to server motherboards, or as part of a system-on-chip (SoC) in smartphones, laptops, or edge devices.

Why NPUs Matter for Generative AI

The explosive rise of generative AI—which includes large language models (LLMs) like ChatGPT, image-generation tools like DALL·E, and video synthesis models—demands computational platforms that can handle massive amounts of data, process it in real-time, and learn from it efficiently. Traditional processors can struggle with these requirements, leading to high energy consumption, increased latency, and throughput bottlenecks.

Key NPU Advantages for Generative AI

Real-Time Processing: Generative AI models such as transformers, diffusion models, and generative adversarial networks (GANs) involve extensive matrix and tensor operations. NPUs excel at multiplying matrices and adding vectors in parallel, helping generative models achieve low-latency performance.

Scalability: NPUs are purpose-built for parallel scaling, making them a strong fit for the large-scale architectures used in generative AI. Adding more NPU cores or NPUs to a data center cluster can linearly increase AI performance without drastically increasing energy costs.

Energy Efficiency: As the complexity of generative models grows, so does their power consumption. NPUs help keep the energy footprint in check by focusing on exactly the kind of math that generative AI requires, eliminating overhead from other computations.

Key Features of NPUs

Parallel Processing: By dividing computational tasks into many smaller ones, NPUs can handle extensive matrix operations far faster than CPUs, which typically execute instructions in a more linear or serial manner. This parallelism is critical for deep learning tasks, where training and inference involve large batches of data.

Low Precision Arithmetic: Most neural network computations don’t require the precision of 32-bit or 64-bit floating-point operations. Low-precision data types, such as 8-bit integers, significantly reduce the number of bits processed per operation, allowing for faster and more energy-efficient execution while still maintaining the model’s accuracy.

High-Bandwidth On-Chip Memory: The ability to keep large chunks of training or inference data near the processor is crucial for AI tasks. Many NPUs feature on-chip high-bandwidth memory (HBM) or advanced memory subsystems designed specifically for neural networks, reducing the need to constantly communicate with external memory.

Hardware Acceleration Techniques: Modern NPU architectures often incorporate specialized hardware units like systolic arrays or tensor cores, enabling them to perform matrix multiplication and other AI-centric operations at blazingly fast speeds with minimal overhead.

How NPUs Work: Simulating the Brain

NPUs draw inspiration from the neural networks of the human brain. Just as billions of neurons and synapses process information in parallel, an NPU is composed of numerous processing elements capable of simultaneously handling large datasets. This design is particularly effective for tasks like:

Image Recognition and Processing

Natural Language Processing (NLP) and Speech Recognition

Object Detection and Autonomous Navigation

Generative AI (e.g., image generation and text generation)

Synaptic Weights and Learning

A cornerstone of neural network computation is the concept of weights, which represent the “strength” or “importance” of each neuron’s connection in the network. NPUs integrate these weights directly into hardware, enabling faster and more energy-efficient updates as a model learns.

Simplified High-Capacity Cores

While CPUs have traditionally handled multiple, diverse operations (ranging from web browsing to spreadsheet calculations), NPUs streamline the design to focus on just a few core operations—like matrix multiplication, activation functions, and convolution—executed repeatedly in parallel.

NPUs vs. GPUs vs. CPUs

Each processor type plays a unique role in modern computing, though there is some overlap when it comes to handling AI tasks. Here’s a quick breakdown:

Feature CPU GPU NPU Primary Use General-purpose tasks, logic, and control Rendering graphics, parallel processing for HPC tasks Specialized parallel processing for AI, ML, and deep learning Number of Cores Few (often 2–16 in consumer chips) Hundreds to thousands of smaller cores Highly parallel array of specialized cores Precision Typically high precision (32-bit or 64-bit) Mix of higher and lower precision (FP32, FP16, etc.) Focus on low precision (8-bit or lower) Energy Efficiency (AI) Moderate when scaled for large AI Good, but can be power-hungry at scale Highly optimized, lower power per operation Physical Footprint Integrated into mainboard or SoC Often standalone cards (discrete GPUs) or SoC-based Can be standalone or integrated into SoC (smartphones, etc.)

Takeaway: While CPUs remain crucial for overall system control and traditional workflows, and GPUs offer robust parallel processing power (especially for heavy graphics tasks), NPUs are purpose-built for AI acceleration and often operate at higher performance-per-watt for machine learning workloads.

Real-World NPU Applications

Data Centers and Cloud AI

Large-scale data centers house standalone NPUs that can be attached directly to server motherboards. These accelerate everything from recommendation engines (like those powering Netflix and Amazon) to generative AI like real-time text and image generation.

Smartphones and Consumer Electronics

Many of today’s premium smartphones, laptops, and tablets incorporate an NPU or AI engine directly into the SoC. Apple’s Neural Engine, Qualcomm’s Hexagon NPU, and Samsung’s Neural Processing Engine are examples of integrated solutions. This approach allows for:

Real-time image and video processing (e.g., background blur on video calls)

On-device voice assistants (with speech recognition)

Intelligent camera features like scene detection, face recognition, and advanced image stabilization

Edge Devices and IoT

NPUs have become pivotal in edge computing, where devices need to process data locally rather than sending it to the cloud. This is especially valuable for applications requiring low latency, data privacy, or real-time feedback—think smart home devices, industry 4.0 sensors, drones, autonomous vehicles, and more.

Robotics

From automated warehouse robots to robotic surgical assistants, NPUs can make split-second decisions based on sensor input. Their ability to handle video feeds (object detection and pattern recognition) and other sensor data quickly is transformative for the next generation of autonomous and semi-autonomous robots.

NPUs for Edge Computing and On-Device AI

Why Edge Computing Matters

As AI proliferates into wearables, remote sensors, and other Internet of Things (IoT) devices, the ability to process data near the source (versus the cloud) can be more critical than ever. Edge AI reduces data transfer costs, mitigates latency issues, and keeps sensitive information on the device—improving both security and privacy.

Role of NPUs in Edge AI

Low Power Consumption: Often battery-operated or energy-constrained, edge devices need an AI processor that can function without draining resources. NPUs, optimized for efficient matrix operations, are the perfect fit.

Real-Time Insights: Whether detecting anomalies in a factory or re-routing a drone mid-flight, split-second inference decisions can make or break an application’s viability. NPUs offer this capability with minimal overhead.

Smartphone Applications: With the emergence of on-device generative AI, NPUs in smartphones are already powering advanced camera features, real-time language translation, and context-aware voice assistance.

The Future of NPUs and AI

As generative AI continues to exponentially increase in capability, so will the demands for high-performance, ultra-efficient computing. Already, hardware manufacturers like Intel, AMD, Nvidia, Apple, Qualcomm, and Samsung are racing to incorporate or refine their own NPU architectures. Likewise, data centers are shifting toward heterogeneous computing models—where CPUs, GPUs, and NPUs co-exist—to handle increasingly specialized workloads at scale.

NPUs for Next-Generation Generative AI

Lower Latency: Future NPUs could achieve near-instantaneous real-time inference, making virtual personal assistants and real-time content generation a seamless part of everyday life.

On-the-Fly Model Adjustments: As models become more dynamic—adjusting their architecture and weights on the fly—NPUs will evolve to handle continuous, online learning scenarios.

Beyond Vision and Language: Generative AI will soon extend into complex multisensory outputs, including real-time haptic feedback, 3D object generation, or even audio-visual immersive experiences.

Multi-Processor Collaboration

Heterogeneous computing involves harnessing the right processor for the right job. The CPU handles generalized tasks and orchestration, the GPU tackles large-scale parallel operations (like graphics or large matrix computations), and the NPU powers specialized AI tasks—especially large-scale neural network inference.

In this future scenario, applications become more flexible and powerful:

Generative art can run locally, with your NPU handling style transfer or upscaling tasks in real-time.

Enterprise software that requires AI-based natural language processing can delegate grammar correction and context understanding to NPUs while the CPU coordinates with the GPU for data visualization.

Complex simulations in scientific research can be split among CPU, GPU, and NPUs to efficiently handle billions of data points.

Rapid Hardware and Software Innovation

Because of the need for rapid scaling of AI, hardware and software innovations are accelerating:

Custom Instruction Sets: Many NPUs are developed with proprietary instruction sets aligned with evolving AI algorithms.

Unified AI Frameworks: AI frameworks (e.g., TensorFlow, PyTorch, ONNX) continue to optimize for NPU backends, simplifying developer workflows.

Edge and Cloud Convergence: The same AI workloads once relegated to the cloud can now be spread across cloud GPUs and NPUs, or directly on edge devices.

Conclusion

Neural Processing Units (NPUs) are ushering in a new era of purpose-built AI hardware, directly addressing the challenges posed by deep learning, generative AI, and large-scale data processing. By focusing on parallel, low-precision workloads, NPUs deliver unprecedented performance, energy efficiency, and scalability—benefits that are paramount not only for cutting-edge cloud AI but also for everyday consumer devices and emerging edge applications.

Their importance in the future of AI cannot be overstated. As demand for on-device generative AI surges and heterogeneous computing becomes the standard, NPUs will likely become as integral to AI-driven systems as the CPU has been for traditional computing. Whether enabling real-time language translation on your smartphone or orchestrating large language models in the data center, the NPU is poised to transform how machines learn and interact with the world—offering a glimpse into a future of ever-more-intelligent, personalized, and energy-efficient computing.

[onedigital]

Tres sectores están liderando la transformación en México gracias a la IA Generativa

 Amazon Web Services (AWS) ha seleccionado a tres startups mexicanas que utilizan Inteligencia Artificial Generativa para resolver desafíos complejos y desarrollar sus industrias  Expertos coinciden en que la IA generativa no eliminará empleos sino que aumentará la capacidad de cada trabajador Hero Guest: IA mejora y humaniza los procesos Continue reading Tres sectores están liderando la…

[acvpartners]

Find Out Now, What Should You Do Connecting With ACV Partners? ACV Partners Reviews |  Your Gateway To Unmatched E-Commerce Success

As per latest trends ACV Partners is the go-to choice for every e-commerce business owner, Amazon, or Etsy seller seeking incredible sales growth. As the top-rated e-commerce growth agency, ACV has revolutionized the online retail landscape with its comprehensive services. From seamless logistics and wholesale distribution to expert brand development, This Company handles every detail to ensure your success. Their unique model allows sellers to focus on growth while they manage all operations, from product sourcing to customer service. With cutting-edge technology and a proven track record, ACV Partners ensures maximum efficiency and profitability. Whether you're just starting or scaling your business, ACV provides the expertise and support needed to thrive in the competitive e-commerce market. Join the ranks of successful sellers who trust Accelerated Ecom Ventures to take their business to the next level.

[amzingamazongroup]

From Patent to Profit: Is Amazon’s IP Accelerator Worth Your Bucks?

Introduction To Amazon IP Accelerator Navigating the bustling marketplace of Amazon, sellers quickly realize the paramount importance of safeguarding their intellectual property. The Amazon IP Accelerator program emerges as a beacon for those seeking not just to protect, but also to monetize their innovations efficiently. But, is it worth the investment? In this deep dive, we’ll unravel the…

[reasonsforhope]

Masterpost: Reasons I firmly believe we will beat climate change

Posts are in reverse chronological order (by post date, not article date), mostly taken from my "climate change" tag, which I went through all the way back to the literal beginning of my blog. Will update periodically.

Especially big deal articles/posts are in bold.

Big picture:

Mature trees offer hope in world of rising emissions (x)

Spying from space: How satellites can help identify and rein in a potent climate pollutant (x)

Good news: Tiny urban green spaces can cool cities and save lives (x)

Conservation and economic development go hand in hand, more often than expected (x)

The exponential growth of solar power will change the world (x)

Sun Machines: Solar, an energy that gets cheaper and cheaper, is going to be huge (x)

Wealthy nations finally deliver promised climate aid, as calls for more equitable funding for poor countries grow (x)

For Earth Day 2024, experts are spreading optimism – not doom. Here's why. (x)

Opinion: I’m a Climate Scientist. I’m Not Screaming Into the Void Anymore. (x)

The World’s Forests Are Doing Much Better Than We Think (x)

‘Staggering’ green growth gives hope for 1.5C, says global energy chief (x)

Beyond Catastrophe: A New Climate Reality Is Coming Into View (x)

Young Forests Capture Carbon Quicker than Previously Thought (x)

Yes, climate change can be beaten by 2050. Here's how. (x)

Soil improvements could keep planet within 1.5C heating target, research shows (x)

The global treaty to save the ozone layer has also slowed Arctic ice melt (x)

The doomers are wrong about humanity’s future — and its past (x)

Scientists Find Methane is Actually Offsetting 30% of its Own Heating Effect on Planet (x)

Are debt-for-climate swaps finally taking off? (x)

High seas treaty: historic deal to protect international waters finally reached at UN (x)

How Could Positive ‘Tipping Points’ Accelerate Climate Action? (x)

Specific examples:

Environmental Campaigners Celebrate As Labour Ends Tory Ban On New Onshore Wind Projects (x)

Private firms are driving a revolution in solar power in Africa (x)

How the small Pacific island nation of Vanuatu drastically cut plastic pollution (x)

Rewilding sites have seen 400% increase in jobs since 2008, research finds [Scotland] (x)

The American Climate Corps take flight, with most jobs based in the West (x)

Waste Heat Generated from Electronics to Warm Finnish City in Winter Thanks to Groundbreaking Thermal Energy Project (x)

Climate protection is now a human right — and lawsuits will follow [European Union] (x)

A new EU ecocide law ‘marks the end of impunity for environmental criminals’ (x)

Solar hits a renewable energy milestone not seen since WWII [United States] (x)

These are the climate grannies. They’ll do whatever it takes to protect their grandchildren. [United States and Native American Nations] (x)

Century of Tree Planting Stalls the Warming Effects in the Eastern United States, Says Study (x)

Chart: Wind and solar are closing in on fossil fuels in the EU (x)

UK use of gas and coal for electricity at lowest since 1957, figures show (x)

Countries That Generate 100% Renewable Energy Electricity (x)

Indigenous advocacy leads to largest dam removal project in US history [United States and Native American Nations] (x)

India’s clean energy transition is rapidly underway, benefiting the entire world (x)

China is set to shatter its wind and solar target five years early, new report finds (x)

‘Game changing’: spate of US lawsuits calls big oil to account for climate crisis (x)

Largest-ever data set collection shows how coral reefs can survive climate change (x)

The Biggest Climate Bill of Your Life - But What Does It DO? [United States] (x)

Good Climate News: Headline Roundup April 1st through April 15th, 2023 (x)

How agroforestry can restore degraded lands and provide income in the Amazon (x) [Brazil]

Loss of Climate-Crucial Mangrove Forests Has Slowed to Near-Negligable Amount Worldwide, Report Hails (x)

Agroecology schools help communities restore degraded land in Guatemala (x)

Climate adaptation:

Solar-powered generators pull clean drinking water 'from thin air,' aiding communities in need: 'It transforms lives' (x)

‘Sponge’ Cities Combat Urban Flooding by Letting Nature Do the Work [China] (x)

Indian Engineers Tackle Water Shortages with Star Wars Tech in Kerala (x)

A green roof or rooftop solar? You can combine them in a biosolar roof — boosting both biodiversity and power output (x)

Global death tolls from natural disasters have actually plummeted over the last century (x)

Los Angeles Just Proved How Spongy a City Can Be (x)

This city turns sewage into drinking water in 24 hours. The concept is catching on [Namibia] (x)

Plants teach their offspring how to adapt to climate change, scientists find (x)

Resurrecting Climate-Resilient Rice in India (x)

Edit 1/12/25: Yes, I know a bunch of the links disappeared. I'll try to fix that when I get the chance. In the meantime, read all the other stuff!!

Other Masterposts:

Going carbon negative and how we're going to fix global heating (x)

[updatedideas]

Amazons GPT55X Unveiled

Hey there, tech enthusiast! 🚀 Grab your coffee because we’re about to dive into one of the most exciting innovations in the world of AI: Amazon’s GPT55X. Picture this: you’re chatting with a friend, and they casually mention this groundbreaking piece of tech. Confused? Don’t fret. We’re here to break it down for you, friend-to-friend. Introducing the Rockstar: Amazons GPT55X Ever watched a movie…

[cloudthat-training]

CloudThat is the world's leading training & Consulting services provider on Cloud, DevOps, Security, AI&ML, IoT, and Big Data for midsize and enterprise clients globally.

Mail us at sales@cloudthat.com

What you will learn:

Use the architectural principles of AWS to inform your selections.

Make your infrastructure scalable, dependable, secure, and highly available by using AWS services.

Learn how to manage AWS services to increase an infrastructure's adaptability and resilience.

To improve performance and cut expenses, figure out how to make an AWS-based infrastructure more efficient.

Apply the Well-Architected Framework to AWS applications to enhance architectures and much more

[crimethinc]

Capitalism Is the Dance of Death

A poster

http://crimethinc.com/CapitalismDeath

From Greece to the Amazon, environmentalists and volunteer firefighters are risking their lives to protect ecosystems threatened by urban expansion, agribusiness, and industrially produced climate change. Yet the causes of these disasters are only accelerating, driven by a system that rewards profit at any price. We are swept up in a rhythm we cannot control, moving faster and faster towards our doom. Capitalism is the danse macabre, the dance of death.

[suzukiblu]

Day twenty-three of “Kon meets pink kryptonite and decides to fuck Tim and his boyfriend about it” behind the cut. (( chrono || non-chrono ))

“I realize I am not letting this one go, but what do you think, are you more into the barbell look or should we put a ring on it?” Bernard muses, giving both his nipples a meaningful pinch as he does, and then a tight little twist. Kon isn’t actually sure if it’s more embarrassing to have someone pay so much attention to a part of his body that most of his other partners haven't, or if it’s more embarrassing how affected he is by the attention. How affected he is by the idea, even. 

“Told you I was the kept boy type, man, c’mon,” he says, trying for a laugh as he says it, and Bernard grins at him and cups his pecs and pushes them up a little bit again. Kon tightens his grip on his ass like–a reflex, maybe, or maybe more like a response–and lifts his cheeks a little bit more too. Bernard probably doesn’t, like, see that or anything, but–but it’s kind of embarrassing to realize, and he knows, like–obviously Tim saw, but since he’s straight behind him maybe he didn’t realize it was, like–like why he . . . 

“Yeah, so what if we kept you?” Bernard asks, his grin turning sly, and Kon stops overthinking because he is suddenly literally just incapable of thinking. That–that is–

It doesn’t help that he heard Tim’s heartbeat accelerate when Bernard said that. 

“Just saying, libido like yours, you’d do better if you were doing the kept boy thing for a couple,” Bernard mentions casually, since it’s apparently blatantly obvious that Kon is likelier to fall off his dick and also the bed than string together a coherent reply right now. “Or, you know, just getting regularly gangbanged by an entire superhero team, but that seems harder to arrange on the daily, you know? Like, more a special occasions kind of thing.” 

“I–you–” Kon attempts, face blushing and gut burning as his hips stutter, and Bernard just keeps up the light, easy chatter like he’s not literally on the guy's dick right now. 

“Like normally I’d suggest the Birds of Prey or just any Amazon expats, obviously, but right now I guess you’d be more into . . . I dunno, Young Justice’s always seemed pretty femme-heavy and I feel like Tim would have some weird feelings about letting Batman Inc borrow you, so maybe see about meeting yourself a few nice Titans?” Bernard suggests, squeezing his pecs again; rubbing and massaging at the muscle for a few moments. Kon is pretty sure he’d be boiling in the literal vacuum of space, hot as he suddenly feels. “Or maybe check if anybody in the Green Lantern Corps has some job-related stress to work out? Or like, what’re the Outsiders up to these days, actually, that’s also–” 

“Please,” Kon half-chokes, not even sure what he’s asking for, and Bernard takes his hands off his chest to wrap an arm around his waist and tug him in closer against himself, which isn’t even something the guy’s actually physically capable of doing but also isn’t something Kon is any kind of capable of not letting him do, and then wraps his other arm up around his neck to grip the back of it, and Kon somehow ends up less riding the other and more just–just rocking a little bit in his lap, pulled down fully onto his cock and just–just full of it. All of it. Just–all of it, while Bernard coaxes him into just–just barely rolling their hips together over and over and . . . 

“How’s that, boy?” Bernard asks, scruffing the back of his neck again. “Comfy?” 

Five seconds before the first time Bernard fucked him Kon would absolutely not have had any concept of having a dude’s entire dick up his ass be something that could count as “comfy”, but right now . . . 

“Don’t pull out,” he half-pleads, half just begs, and Bernard tugs his head down just enough to let him press a kiss against his temple and squeezes the arm around his waist again; rolls his hips up again. Kon stifles a whimper. 

“Naw, not gonna,” Bernard promises. “Not ‘til you’ve come again, at least.” 

“Not even then,” Kon half-mumbles, feeling a little bit–dizzier, kind of, every time Bernard’s nails scritch the back of his neck; slumping down heavier into him without really meaning to or anything. “Please? Please just lemme keep it a little longer? Feels so–feels so good, makes me feel like you want me.” 

“Aw, of course I want you, boy,” Bernard coos, and then just starts petting him outright. “You’re super cute. Super fun to pet-sit, too. And I already told you you could cuddle my cock if you wanted, didn’t I?” 

“Wanna,” Kon pleads, feeling all warm and melty and heavy and barely remembering to roll his hips in counter to Bernard’s; barely remembering to keep his hands on his ass where the other put them. “Wanna–lemme cuddle it, lemme cuddle you, I’ll be a good boy, promise, promise, promise–” 

“‘Course you will, boy,” Bernard hums, rubbing a flat palm up the underside of his fade. “Robin told you to be, didn’t he?” 

Kon melts.

[jcmarchi]

AI’s Growing Appetite for Power: Are Data Centers Ready to Keep Up?

New Post has been published on https://thedigitalinsider.com/ais-growing-appetite-for-power-are-data-centers-ready-to-keep-up/

AI’s Growing Appetite for Power: Are Data Centers Ready to Keep Up?

As artificial intelligence (AI) races forward, its energy demands are straining data centers to the breaking point. Next-gen AI technologies like generative AI (genAI) aren’t just transforming industries—their energy consumption is affecting nearly every data server component—from CPUs and memory to accelerators and networking.

GenAI applications, including Microsoft’s Copilot and OpenAI’s ChatGPT, demand more energy than ever before. By 2027, training and maintaining these AI systems alone could consume enough electricity to power a small country for an entire year. And the trend isn’t slowing down: over the last decade, power demands for components such as CPUs, memory, and networking are estimated to grow 160% by 2030, according to a Goldman Sachs report.

The usage of large language models also consumes energy. For instance, a ChatGPT query consumes about ten times a traditional Google search. Given AI’s massive power requirements, can the industry’s rapid advancements be managed sustainably, or will they contribute further to global energy consumption? McKinsey’s recent research shows that around 70% of the surging demand in the data center market is geared toward facilities equipped to handle advanced AI workloads. This shift is fundamentally changing how data centers are built and run, as they adapt to the unique requirements of these high-powered genAI tasks.

“Traditional data centers often operate with aging, energy-intensive equipment and fixed capacities that struggle to adapt to fluctuating workloads, leading to significant energy waste,” Mark Rydon, Chief Strategy Officer and co-founder of distributed cloud compute platform Aethir, told me. “Centralized operations often create an imbalance between resource availability and consumption needs, leading the industry to a critical juncture where advancements could risk undermining environmental goals as AI-driven demands grow.”

Industry leaders are now addressing the challenge head-on, investing in greener designs and energy-efficient architectures for data centers. Efforts range from adopting renewable energy sources to creating more efficient cooling systems that can offset the vast amounts of heat generated by genAI workloads.

Revolutionizing Data Centers for a Greener Future

Lenovo recently introduced the ThinkSystem N1380 Neptune, a leap forward in liquid cooling technology for data centers. The company asserts that the innovation is already enabling organizations to deploy high-powered computing for genAI workloads with significantly lower energy use — up to 40% less power in data centers. N1380 Neptune, harnesses NVIDIA’s latest hardware, including the Blackwell and GB200 GPUs, allowing for the handling of trillion-parameter AI models in a compact setup. Lenovo said that it aims to pave the way for data centers that can operate 100KW+ server racks without the need for dedicated air conditioning.

“We identified a significant requirement from our current consumers: data centers are consuming more power when handling AI workloads due to outdated cooling architectures and traditional structural frameworks,” Robert Daigle, Global Director of AI at Lenovo, told me. “To understand this better, we collaborated with a high-performance computing (HPC) customer to analyze their power consumption, which led us to the conclusion that we could reduce energy usage by 40%.” He added that the company took into account factors such as fan power and the power consumption of cooling units, comparing these with standard systems available through Lenovo’s data center assessment service, to develop the new data center architecture in partnership with Nvidia.

UK-based information technology consulting company AVEVA, said it is utilizing predictive analytics to identify issues with data center compressors, motors, HVAC equipment, air handlers, and more.

“We found that it’s the pre-training of generative AI that consumes massive power,” Jim Chappell, AVEVA’s Head of AI & Advanced Analytics, told me. “Through our predictive AI-driven systems, we aim to find problems well before any SCADA or control system, allowing data center operators to fix equipment problems before they become major issues. In addition, we have a Vision AI Assistant that natively integrates with our control systems to help find other types of anomalies, including temperature hot spots when used with a heat imaging camera.”

Meanwhile, decentralized computing for AI training and development through GPUs over the cloud is emerging as an alternative. Aethir’s Rydon explained that by distributing computational tasks across a broader, more adaptable network, energy use can be optimized, by aligning resource demand with availability—leading to substantial reductions in waste from the outset.

“Instead of relying on large, centralized data centers, our ‘Edge’ infrastructure disperses computational tasks to nodes closer to the data source, which drastically reduces the energy load for data transfer and lowers latency,” said Rydon. “The Aethir Edge network minimizes the need for constant high-power cooling, as workloads are distributed across various environments rather than concentrated in a single location, helping to avoid energy-intensive cooling systems typical of central data centers.”

Likewise, companies including Amazon and Google are experimenting with renewable energy sources to manage rising power needs in their data centers. Microsoft, for instance, is investing heavily in renewable energy sources and efficiency-boosting technologies to reduce its data center’s energy consumption. Google has also taken steps to shift to carbon-free energy and explore cooling systems that minimize power use in data centers. “Nuclear power is likely the fastest path to carbon-free data centers. Major data center providers such as Microsoft, Amazon, and Google are now heavily investing in this type of power generation for the future. With small modular reactors (SMRs), the flexibility and time to production make this an even more viable option to achieve Net Zero,” added AVEVA’s Chappell.

Can AI and Data Center Sustainability Coexist?

Ugur Tigli, CTO at AI infrastructure platform MinIO, says that while we hope for a future where AI can advance without a huge spike in energy consumption, that’s just not realistic in the short term. “Long-term impacts are trickier to predict,” he told me, “but we’ll see a shift in the workforce, and AI will help improve energy consumption across the board.” Tigli believes that as energy efficiency becomes a market priority, we’ll see growth in computing alongside declines in energy use in other sectors, especially as they become more efficient.

He also pointed out that there’s a growing interest among consumers for greener AI solutions. “Imagine an AI application that performs at 90% efficiency but uses only half the power—that’s the kind of innovation that could really take off,” he added. It’s clear that the future of AI isn’t just about innovation—it’s also about data center sustainability. Whether it’s through developing more efficient hardware or smarter ways to use resources, how we manage AI’s energy consumption will greatly influence the design and operation of data centers.

Rydon emphasized the importance of industry-wide initiatives that focus on sustainable data center designs, energy-efficient AI workloads, and open resource sharing. “These are crucial steps towards greener operations,” he said. “Businesses using AI should partner with tech companies to create solutions that reduce environmental impact. By working together, we can steer AI toward a more sustainable future.”

[txttletale]

I feel like a lot of people miss that the likeliest use cases for ai artwork (corporate illustrations, eye-catches for shitty mobile game, amazon book covers, etc) are already effectively algorithmically generated anyway - a human being still makes them, but the demand is set purely by capital and the specifics by the brownian motion of the marketplace, with no pairs of eyes involved especially caring as long as its occupying the place in which art is expected. I think people get caught up enough in a noticeable acceleration of commodified art that they mistake it for a whole new phenomenon.

absolutely 100%. i always come back to a very astute thing podcaster riley quinn said which is "if your job gets replaced by AI that means it was already being done by an AI" (ie, mindlessly and formulaicly)

[toxicanonymity]

bonus snapchats 🍆🥺

350 words, stepdad!joel x f!reader

stepdad master list

More horny and pensive snaps from Stepdad!Joel to go with his latest story, Amazon but you can read it as any Joel who's begging for your forgiveness. WARNINGS: JACKING OFF

Video: A POV shot taken by bracing his phone on one of his knees. His hard, shiny cock is in the foreground with his veiny hand wrapped around it, but you can also see his agonized face and his chest rising and falling with heavy breaths. He moans "ohh," as he strokes himself at a moderate but accelerating pace, then pulls up his shirt, closes his eyes. He groans "uggggh," as he slows his hand way down. His stomach flexes outward and he shudders and comes all over himself with a long, low sigh.  It's trickling into his belly button when the video cuts off. 

Video: A minute later, he’s cleaned up and it's a shot of his head and chest.  He slowly shakes his head and says “I’m. . ." (Deep breath) "so stupid."

Video: Lying in bed in a white t-shirt with his head propped up on his hand. Bicep bulging out of his sleeve.  Caption: had a dream that you forgave me. He sighs and pouts with his eyes, but not with his lips. ”I guess it was. . .” (sighs) “I dunno. I'm sorry."

Video: Same position but his glasses are on. "Then it wasn't real, and I was like." (sighs). "Shit."

Photo: a minute later, his hand is cupping a thick protrusion in his pajama pants, holding it down onto his body and there's a wet spot at the tip. 

Video: Looking at himself in the bathroom mirror. You can tell from the audio that the water is on in the background.  He's shirtless, pj pants pulled down under his cock. Neck and forearm veins blazing. He's breathing heavily and softly moaning "ohh." He holds his phone in one hand and jacks off with the other as the mirror fogs up. He strokes himself at a slow beat and his lips part into a louder moan.  By the time he groans and comes, you can’t really see him through the condensation. But his hand comes into the frame and braces on the mirror as if he came in the sink, and you hear him sigh "fuck" before it cuts off. 

Check out my #joel jacks off tag for more of him j-ing o.

[allthebrazilianpolitics]

Megafires in the Amazon Accelerate the Risk of Biome Collapse

Increase in fires in native forest areas and extreme drought could lead to an irreversible situation

A fire that began on August 8 has already burned over 67,000 hectares in the Kayapó Indigenous Land, in the Xingu region of Pará. The data comes from the Servir-Amazonia program, run by NASA, which monitors the region via satellites. The size of the megafire is equivalent to that of Florianópolis and is just one of several fire fronts in the Amazon.

The classification "megafire" is used for fires exceeding 10,000 hectares —something that is becoming increasingly common. "We are not just entering the era of fire but the era of megafires. It's quite catastrophic," says Erika Berenguer, a senior scientist at the University of Oxford.

Continue reading.

[mostlysignssomeportents]

The super-rich got that way through monopolies

Catch me in Miami! I'll be at Books and Books in Coral Gables on Jan 22 at 8PM.

Just in time for Davos, here's 'Taken, not earned: How monopolists drive the world’s power and wealth divide," a report from a coalition of international tax justice and anti-corporate activist groups:

https://www.balancedeconomy.net/wp-content/uploads/2024/01/Davos-Taken-not-Earned-full-Report-2024-FINAL.pdf

The rise of monopolies over the past 40 years came about as the result of specific, deliberate policy choices. As the report documents, the wealthiest people in America funneled a fortune into neutering antitrust enforcement, through the "consumer welfare" doctrine.

This is an economic theory that equates monopolies with efficiency: "If everyone is buying the same things from the same store, that tells you the store is doing something right, not something criminal." 40 years ago, and ever since, the wealthy have funded think-tanks, university programs and even "continuing education" programs for federal judges to push this line:

https://pluralistic.net/2021/08/13/post-bork-era/#manne-down

They didn't do this for ideological reasons – they were chasing material goals. Monopolies produce vast profits, and those profits produce vast wealth. The rise and rise of the super rich cannot be decoupled from the rise and rise of monopolies.

If you're new to this, you might think that "monopoly" only refers to a sector in which there is only one seller. But that's not what economists mean when they talk about monopolies and monopolization: for them, a monopoly is a company with power. Economists who talk about monopolies mean companies that "can act independently without needing to consider the responses of competitors, customers, workers, or even governments."

One way to measure that power is through markups ("the difference between the selling price of goods or services and their cost"). Very large companies in concentrated industries have very high markups, and they're getting higher. From 2017-22, the 20 largest companies in the world had average markups of 50%. The 100 largest companies average 43%. The smallest half of companies get average markups of 25%.

Those markups rose steeply during the covid lockdowns – and so did the wealth of the billionaires who own them. Tech billionaires – Bezos, Brin and Page, Gates and Ballmer – all made their fortunes from monopolies. Warren Buffet is a proud monopolist who says "the single most important decision in evaluating a business is pricing power… if you have to have a prayer session before raising the price by 10 percent, then you’ve got a terrible business."

We are living in the age of the monopoly. In the 1930s, the top 0.1% of US companies accounted for less than half of America's GDP. Today, it's 90%. And it's accelerating, with global mergers climbing from 2,676 in 1985 to 62,000 in 2021.

Monopoly's cheerleaders claim that these numbers vindicate them. Monopolies are so efficient that everyone wants to create them. Those efficiencies can be seen in the markups monopolies can charge, and the profits they can make. If a monopoly has a 50% markup, that's just the "efficiency of scale."

But what is the actual shape of this "efficiency?" How is it manifest? The report's authors answer this with one word: power.

Monopolists have the power "to extract wealth from, to restrict the freedoms of, and to manipulate or steer the vastly larger numbers of losers." They establish themselves as gatekeepers and create chokepoints that they can use to raise prices paid by their customers and lower the payout to their suppliers:

https://chokepointcapitalism.com/

These chokepoints let monopolies usurp "one of the ultimate prerogatives of state power: taxation." Amazon sellers pay a 51% tax to sell on the platform. App Store suppliers pay a 30% tax on every dollar they make with their apps. That translates into higher costs. Consider a good that costs $10 to make: the bottom 50% of companies (by size) would charge $12.50 for that product on average. The largest companies would charge $15. Thus monopolies don't just make their owners richer – they make everyone else poorer, too.

This power to set prices is behind the greedflation (or, more politely, "seller's inflation"). The CEOs of the largest companies in the world keep getting on investor calls and bragging about this:

https://pluralistic.net/2023/03/11/price-over-volume/#pepsi-pricing-power

The food system is incredibly monopolistic. The Cargill family own the largest commodity trader in the world, which is how they built up a family fortune worth $43b. Cargill is one of the "ABCD" companies ("Archer Daniels Midland, Bunge, Cargill and Louis Dreyfus") that control the world's food supply, and they tripled their profits during the lockdown.

Monopolies gouge everyone – even governments. Pfizer charged the NHS £18-22/shot for vaccines that cost £5/shot to make. They took the British government for £2bn – that's enough to pay last year's pay hike for NHS nurses, six times over,

But monopolies also abuse their suppliers, especially their employees. All over the world, competition authorities are uncovering "wage fixing" and "no poaching" agreements among large firms, who collude to put a cap on what workers in their sector can earn. Unions report workers having their pay determined by algorithms. Bosses lock employees in with noncompetes and huge repayment bills for "training":

https://pluralistic.net/2022/08/04/its-a-trap/#a-little-on-the-nose

Monopolies corrupt our governments. Companies with huge markups can spend some of that money on lobbying. The 20 largest companies in the world spend more than €155m/year lobbying in the US and alone, not counting the money they spend on industry associations and other cutouts that lobby on their behalf. Big Tech leads the pack on lobbying, accounting for 82% of EU lobbying spending and 58% of US lobbying.

One key monopoly lobbying priority is blocking climate action, from Apple lobbying against right-to-repair, which creates vast mountains of e-waste, to energy monopolist lobbying against renewables. And energy companies are getting more monopolistic, with Exxonmobil spending $65b to buy Pioneer and Chevron spending $60b to buy Hess. Many of the world's richest people are fossil fuel monopolists, like Charles and Julia Koch, the 18th and 19th richest people on the Forbes list. They spend fortunes on climate denial.

When people talk about the climate impact of billionaires, they tend to focus on the carbon footprints of their mansions and private jets, but the true environmental cost of the ultra rich comes from the anti-renewables, pro-emissions lobbying they buy with their monopoly winnings.

The good news is that the tide is turning on monopolies. A coalition of "businesses, workers, farmers, consumers and other civil society groups" have created a "remarkably successful anti-monopoly movement." The past three years saw more regulatory action on corporate mergers, price-gouging, predatory pricing, labor abuses and other evils of monopoly than we got in the past 40 years.

The business press – cheerleaders for monopoly – keep running editorials claiming that enforcers like Lina Khan are getting nothing done. Sure, WSJ, Khan's getting nothing done – that's why you ran 80 editorial about her:

https://pluralistic.net/2023/07/14/making-good-trouble/#the-peoples-champion

(Khan's winning like crazy. Just last month she killed four megamergers:)

https://www.thesling.org/the-ftc-just-blocked-four-mergers-in-a-month-heres-how-its-latest-win-fits-into-the-broader-campaign-to-revive-antitrust/

The EU and UK are taking actions that would have been unimaginable just a few years ago. Canada is finally set to get a real competition law, with the Trudeau government promising to add an "abuse of dominance" rule to Canada's antitrust system.

Even more exciting are the moves in the global south. In South Africa, "competition law contains some of the most progressive ideas of all":

It actively seeks to create greater economic participation, particularly for ‘historically disadvantaged persons’ as part of its public interest considerations in merger decisions.

Balzac wrote, "Behind every great fortune there is a crime." Chances are, the rapsheet includes an antitrust violation. Getting rid of monopolies won't get rid of all the billionaires, but it'll certainly get rid of a hell of a lot of them.

I'm Kickstarting the audiobook for The Bezzle, the sequel to Red Team Blues, narrated by @wilwheaton! You can pre-order the audiobook and ebook, DRM free, as well as the hardcover, signed or unsigned. There's also bundles with Red Team Blues in ebook, audio or paperback.

If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2024/01/17/monopolies-produce-billionaires/#inequality-corruption-climate-poverty-sweatshops

[septimaseverinawannawrite]

Saturday Seven 04

Fandom: Tom Hiddleston's Loki

Type: Mini-Drabble

Pairing: Loki x F!Reader

Warning: NSFW. My English (is my second language).

Rate: M

Words: 303

A/N: *gasping* I can resume my own event again! May this inspiration be with me always, and make me be able to write contless one-shots to mini-series! Annnnnnnnnd don't hit me for cliffhanging, 'cause I might continue what I've started 😉

🌹Click to My AO3

Loki found that a space beside him – that you supposed to sleep and hold him after sensual night full of burning fire, passion, and fervent in this room, your room, happened every Friday night – was emptied and cold. He did not here sound of shower running in bathroom, or clanking at small kitchen of your tiny flat. This made him curious; how strange.

He grabbed a towel at the end of the king-size bed, draped it loose before finding you, sat on a sofa in his black shirt from last night, definitely without any piece of clothes underneath, but you were earnestly reading a small and quite rather vintage book without awareness of present circumstance; Loki had been approaching you with smirk.

“Who is this Ms. Ellen Conroy? She shouldn’t vie to the mighty god of Asgard by drawing your attention from me.”

“Hmm, I thought you knew her.” You raised your eyebrows as Loki sat beside you, snuggling his arms around you. “Oh, here is it! ‘Green in its degraded sense gives us "the green-eyed monster jealousy’…” Your voice was exciting, on contrary to Loki, his eyes narrowed down.

“I must object Ms. Conroy.” He whispered into your ears; a hand fondled your bosom, starting with slow and light paces, accelerating to firm and hard. “Green, particularly with Emerald green: the colour of glorious, graceful and gleaming….and for King and his queen.”

A gasp and moan slipped from your lips; trying to turn a page and muttering under hitched breath. “Y-y- you h-h-haven’t…. Loki…Not now…I-I-I promise P-p-patsy to return this on…hnnnng!”

“Surely she can wait.” He chuckled. Another hand running upon your abdomen, sliding down along underbelly to your, beginning to moist, fold. Finally, his slender fingers passed through, while the book was falling from your hand as your body tensed up.

P.S. The book is The Symbolism of Colours (1921) by Ellen Conroy Mccaffery. Available on Amazon, but if you don't mind reading in typewriting fonts, click here. 💛

[mariacallous]

In March 2007, Google’s then senior executive in charge of acquisitions, David Drummond, emailed the company’s board of directors a case for buying DoubleClick. It was an obscure software developer that helped websites sell ads. But it had about 60 percent market share and could accelerate Google’s growth while keeping rivals at bay. A “Microsoft-owned DoubleClick represents a major competitive threat,” court papers show Drummond writing.

Three weeks later, on Friday the 13th, Google announced the acquisition of DoubleClick for $3.1 billion. The US Department of Justice and 17 states including California and Colorado now allege that the day marked the beginning of Google’s unchecked dominance in online ads—and all the trouble that comes with it.

The government contends that controlling DoubleClick enabled Google to corner websites into doing business with its other services. That has resulted in Google allegedly monopolizing three big links of a vital digital advertising supply chain, which funnels over $12 billion in annual revenue to websites and apps in the US alone.

It’s a big amount. But a government expert estimates in court filings that if Google were not allegedly destroying its competition illegally, those publishers would be receiving up to an additional hundreds of millions of dollars each year. Starved of that potential funding, “publishers are pushed to put more ads on their websites, to put more content behind costly paywalls, or to cease business altogether,” the government alleges. It all adds up to a subpar experience on the web for consumers, Colorado attorney general Phil Weiser says.

“Google is able to extract hiked-up costs, and those are passed on to consumers,” he alleges. “The overall outcome we want is for consumers to have more access to content supported by advertising revenue and for people who are seeking advertising not to have to pay inflated costs.”

Google disputes the accusations.

Starting today, both sides’ arguments will be put to the test in what’s expected to be a weekslong trial before US district judge Leonie Brinkema in Alexandria, Virginia. The government wants her to find that Google has violated federal antitrust law and then issue orders that restore competition. In a best-case scenario, according to several Google critics and experts in online ads who spoke with WIRED, internet users could find themselves more pleasantly informed and entertained.

It could take years for the ad market to shake out, says Adam Heimlich, a longtime digital ad executive who’s extensively researched Google. But over time, fresh competition could lower supply chain fees and increase innovation. That would drive “better monetization of websites and better quality of websites,” says Heimlich, who now runs AI software developer Chalice Custom Algorithms.

Tim Vanderhook, CEO of ad-buying software developer Viant Technology, which both competes and partners with Google, believes that consumers would encounter a greater variety of ads, fewer creepy ads, and pages less cluttered with ads. “A substantially improved browsing experience,” he says.

Of course, all depends on the outcome of the case. Over the past year, Google lost its two other antitrust trials—concerning illegal search and mobile app store monopolies. Though the verdicts are under appeal, they’ve made the company’s critics optimistic about the ad tech trial.

Google argues that it faces fierce competition from Meta, Amazon, Microsoft, and others. It further contends that customers benefited from each of the acquisitions, contracts, and features that the government is challenging. “Google has designed a set of products that work efficiently with each other and attract a valuable customer base,” the company’s attorneys wrote in a 359-page rebuttal.

For years, Google publicly has maintained that its ad tech projects wouldn’t harm clients or competition. “We will be able to help publishers and advertisers generate more revenue, which will fuel the creation of even more rich and diverse content on the internet,” Drummond testified in 2007 to US senators concerned about the DoubleClick deal’s impact on competition and privacy. US antitrust regulators at the time cleared the purchase. But at least one of them, in hindsight, has said he should have blocked it.

Deep Control

The Justice Department alleges that acquiring DoubleClick gave Google “a pool of captive publishers that now had fewer alternatives and faced substantial switching costs associated with changing to another publisher ad server.” The global market share of Google’s tool for publishers is now 91 percent, according to court papers. The company holds similar control over ad exchanges that broker deals (around 70 percent) and tools used by advertisers (85 percent), the court filings say.

Google’s dominance, the government argues, has “impaired the ability of publishers and advertisers to choose the ad tech tools they would prefer to use and diminished the number and quality of viable options available to them.”

The government alleges that Google staff spoke internally about how they have been earning an unfair portion of what advertisers spend on advertising, to the tune of over a third of every $1 spent in some cases.

Some of Google’s competitors want the tech giant to be broken up into multiple independent companies, so each of its advertising services competes on its own merits without the benefit of one pumping up another. The rivals also support rules that would bar Google from preferencing its own services. “What all in the industry are looking for is fair competition,” Viant’s Vanderhook says.

If Google ad tech alternatives win more business, not everyone is so sure that the users will notice a difference. “We’re talking about moving from the NYSE to Nasdaq,” Ari Paparo, a former DoubleClick and Google executive who now runs the media company Marketecture, tells WIRED. The technology behind the scenes may shift, but the experience for investors—or in this case, internet surfers—doesn’t.

Some advertising experts predict that if Google is broken up, users’ experiences would get even worse. Andrey Meshkov, chief technology officer of ad-block developer AdGuard, expects increasingly invasive tracking as competition intensifies. Products also may cost more because companies need to not only hire additional help to run ads but also buy more ads to achieve the same goals. “So the ad clutter is going to get worse,” Beth Egan, an ad executive turned Syracuse University associate professor, told reporters in a recent call arranged by a Google-funded advocacy group.

But Dina Srinivasan, a former ad executive who as an antitrust scholar wrote a Stanford Technology Law Review paper on Google’s dominance, says advertisers would end up paying lower fees, and the savings would be passed on to their customers. That future would mark an end to the spell Google allegedly cast with its DoubleClick deal. And it could happen even if Google wins in Virginia. A trial in a similar lawsuit filed by Texas, 15 other states, and Puerto Rico is scheduled for March.