Microsoft has begun deploying its home‑grown Maia 200 AI inference chip in its data centers while confirming it will continue purchasing GPUs from Nvidia and AMD. CEO Satya Nadella emphasized the company’s ongoing partnerships and the need to stay ahead in AI hardware. The Maia 200, designed for high‑performance AI model inference, will first be used by Microsoft’s Superintelligence team as they develop frontier models, and will also support OpenAI models on Azure. The move reflects a hybrid strategy of building in‑house chips while leveraging external suppliers. Leer más →

OpenAI is preparing to announce its first hardware product in the second half of the year, with reports suggesting the device will be earbuds codenamed “Sweet Pea.” The wearable is expected to feature a custom 2‑nanometer processor that handles AI tasks locally, and the company is weighing manufacturing partners such as Luxshare and Foxconn. OpenAI aims to ship tens of millions of units in the first year, seeking greater control over the distribution of its ChatGPT assistant while facing integration challenges with existing ecosystems. Leer más →

DeepSeek, in partnership with Peking University, unveiled Engram, a new training method that separates static memory from computation in large language models. By using hashed N‑gram lookups and a context‑aware gating mechanism, Engram reduces reliance on high‑bandwidth memory (HBM), allowing models to operate efficiently on standard GPU memory while scaling parameter counts. Tests on a 27‑billion‑parameter model showed measurable gains across industry benchmarks, and the approach integrates with existing hardware solutions such as Phison’s SSD‑based accelerators and emerging CXL standards. Engram could ease pressure on costly memory hardware and stabilize DRAM price volatility. Leer más →

A recent leak on X reveals that OpenAI is developing a behind‑ear wearable called Sweetpea, designed to integrate ChatGPT functionality directly into a discreet ear‑mounted device. The prototype reportedly includes a custom chip that can issue commands to Siri, an ultrasonic transmitter for environmental sensing, and a larger battery positioned behind the ear for extended use. The leaker, known as Smart Pikachu, suggests the device could launch near September 2026 with a high price point reflecting its smartphone‑grade components. Sweetpea appears aimed at competing with mainstream earbuds while offering AI‑first capabilities. Leer más →

OpenAI announced an acqui‑hire of the three co‑founders behind Convogo, a platform that helped executive coaches and HR teams automate leadership assessments. The deal does not include Convogo’s technology or product, which will be discontinued, but brings the founders—Matt Cooper, Evan Cater, and Mike Gillett—into OpenAI’s AI cloud efforts under an all‑stock arrangement. The move reflects OpenAI’s ongoing strategy of using acquisitions to acquire talent and accelerate capabilities, adding to a series of recent purchases that have either integrated products into OpenAI’s ecosystem or shuttered them as teams joined the company. Leer más →

OpenAI is reportedly developing a new AI‑powered hardware device that could take the form of a smart pen. The project, internally dubbed “Gumdrop,” is linked to former Apple design chief Jony Ive and may shift manufacturing from Luxshare to Foxconn. Tips suggest the pen would feature always‑on listening and ChatGPT assistance, while other concepts such as a wearable pin and smart speaker have also been mentioned. CEO Sam Altman has described the envisioned experience as “a cabin by a lake,” indicating a calm, background‑focused interaction. Leer más →

Tech leaders are shifting artificial intelligence processing from cloud data centers to users' devices. On‑device AI promises faster response times, stronger privacy protection, and lower ongoing costs by eliminating the need for constant cloud compute. Companies such as Apple, Google, and Qualcomm are deploying specialized models and custom hardware to handle tasks like facial recognition, language summarization, and contextual assistance locally. While current models excel at quick tasks, more complex operations still rely on cloud offloading. Researchers at Carnegie Mellon highlight the trade‑offs and anticipate rapid advances in both hardware and algorithms over the next few years. Leer más →

Naveen Rao, former head of AI at Databricks, has closed a $475 million seed financing for his new venture Unconventional AI, valuing the company at $4.5 billion. The round was led by Andreessen Horowitz and Lightspeed Ventures, with Lux Capital and DCVC also participating. Rao’s aim is to build an energy‑efficient computer for artificial intelligence, a goal he describes as “as efficient as biology.” The funding represents the first installment toward a potential $1 billion total raise. Rao’s prior successes include MosaicML, acquired by Databricks for $1.3 billion, and Nervana Systems, acquired by Intel for more than $400 million. Leer más →

Meta announced the acquisition of Limitless, the AI‑focused hardware startup formerly known as Rewind. The deal will see Limitless cease sales of its AI‑powered pendant and transition existing users to Meta’s Unlimited Plan while maintaining support for a year. Limitless, founded by Brett Bejcek and former Optimizely co‑founder Dan Siroker, pivoted to wearable AI devices last year with its $99 pendant. Meta says the addition will help accelerate its work on AI‑enabled wearables, though details on future product integration remain limited. Leer más →

Amazon executives highlighted the rapid growth of the company’s Trainium AI chip line, noting that Trainium2 is already generating multi‑billion‑dollar revenue with over a million chips in production and more than 100,000 customers. The chip’s price‑performance edge has attracted major partners such as Anthropic, which is using hundreds of thousands of chips for its Project Rainier. At AWS re:Invent, Amazon unveiled the next‑generation Trainium3, promising four‑times the speed and lower power usage, underscoring Amazon’s ambition to challenge Nvidia’s dominance in the AI‑hardware market. Leer más →

OpenAI CEO Sam Altman and former Apple designer Jony Ive disclosed that their first AI hardware device is in the prototyping stage and could be ready in less than two years. Described as a simple, playful, screen‑free unit roughly the size of a smartphone, the design aims for an intuitive, almost naive elegance that invites users to pick it up and use it without hesitation. Both executives emphasized the product’s tactile appeal and the hope that observers will instantly recognize it as the solution they’ve been seeking. Leer más →

Microsoft is turning to its AI partner OpenAI to address challenges in its chip strategy. Under a revised agreement, Microsoft will gain full access to OpenAI’s custom AI chip designs, which are being co‑developed with Broadcom, and secure intellectual‑property rights to those designs. The deal also preserves Microsoft’s access to OpenAI’s models through 2032 while excluding OpenAI’s consumer‑hardware initiatives. Executives describe the move as a pragmatic response to the high cost and complexity of building cutting‑edge AI chips, positioning OpenAI’s expertise as a key accelerator for Microsoft’s semiconductor ambitions. Leer más →

Qualcomm announced the AI200 and AI250 processors, built on its Hexagon neural processing technology, to target AI inference workloads and compete with Nvidia's dominance. The AI200, slated for release next year, offers 768 GB of RAM and can be combined in racks of up to 72 chips. The AI250, planned for 2027, promises a generational leap in efficiency and lower power consumption. Saudi Arabia's Public Investment Fund‑backed AI firm Humain will use both chips in its upcoming AI datacenters, signaling Qualcomm's push into the high‑performance AI market. Leer más →

Nvidia has introduced the DGX Spark, a small‑form‑factor AI system that runs an ARM‑based DGX OS built on Ubuntu Linux and includes the company’s AI software stack. Priced from US $3,999, the Spark offers performance comparable to an RTX 5070 GPU but with 128 GB of unified memory, enabling larger AI models. Nvidia CEO Jensen Huang highlighted the device’s mission by personally delivering a unit to Elon Musk at SpaceX’s Starbase, recalling a similar hand‑off to OpenAI in 2016. Leer más →

At OpenAI’s developer conference, CEO Sam Altman and former Apple chief design officer Jony Ive shared their vision for a new class of AI‑powered devices. While details remain limited, the duo emphasized a focus on redesigning the human‑technology relationship, aiming for products that bring happiness, fulfillment, and reduced anxiety. Ive highlighted a portfolio of compelling concepts, and Altman acknowledged the technical challenges of creating novel form factors. The collaboration builds on OpenAI’s acquisition of Ive’s design firm LoveFrom and its Io project, positioning the company among other tech giants exploring consumer AI hardware. Leer más →

Cerebras Systems, the AI‑hardware startup founded in 2015, announced a $1.1 billion Series G round that values the company at $8.1 billion. The financing, co‑led by Fidelity and Atreides Management with participation from Tiger Global, Valor Equity Partners and 1789 Capital, follows a $250 million Series F round in 2021. CEO and co‑founder Andrew Feldman said the round will fund new data‑center expansions, U.S. manufacturing hubs and continued technology development. Despite an earlier plan to go public by 2025 that was delayed by regulatory review, Cerebras remains focused on scaling its AI inference services and cloud offering. Leer más →

Google introduced its seventh‑generation Tensor Processing Unit, dubbed Ironwood, at a recent Hot Chips event. The dual‑die chip delivers 4,614 TFLOPs of FP8 performance and pairs each die with eight stacks of HBM3e, providing 192 GB of memory per chip. When scaled to a 9,216‑chip pod, the system reaches 1.77 PB of directly addressable memory—the largest shared‑memory configuration ever recorded for a supercomputer. The architecture includes advanced reliability features, liquid‑cooling infrastructure, and AI‑assisted design optimizations, and is already being deployed in Google Cloud data centers for large‑scale inference workloads. Leer más →

At Meta Connect, the company showcased upgraded Ray‑Ban Gen 2 smart glasses with a neural wristband, doubled battery life, and improved cameras, alongside the sporty Oakley Vanguard model. CEO Mark Zuckerberg introduced new AI features such as Live AI, but live demonstrations faltered, with three notable failures during cooking, WhatsApp call, and translation demos. While the hardware upgrades signal a strong push into wearable tech, the shaky AI performance underscores challenges Meta faces in delivering seamless, AI‑driven experiences on its glasses. The event leaves observers weighing the promise of the hardware against the current limitations of its software. Leer más →

Morningstar Equity Research forecasts Nvidia's AI‑related sales could approach $400 billion by 2028, driven largely by its accelerator products. The analysis highlights that hyperscaler cloud providers are expected to push annual capital expenditures beyond $450 billion by 2027, fueling demand for AI hardware. While growth remains robust, the pace is projected to decelerate after 2024, raising questions about sustainability. Competitive pressures from Broadcom, AMD and other semiconductor firms, together with rising energy needs, regulatory scrutiny and geopolitical factors, could temper Nvidia's dominance in the long run. Leer más →

OpenAI is developing a range of AI-enabled hardware, including smart glasses, a digital voice recorder, a wearable pin, and a smart speaker without a display. The company is partnering with manufacturers such as Luxshare and Goertek—suppliers also used by Apple—and has attracted former Apple executives, including Jony Ive, to the effort. A launch is targeted for late 2026 or early 2027, with the possibility that a single device will debut first. The goal is to create accessories that complement existing phones, tablets and laptops rather than replace them, learning from earlier AI gadgets that saw limited success. Leer más →