Tensor Processing Unit
PulseAugur coverage of Tensor Processing Unit — every cluster mentioning Tensor Processing Unit across labs, papers, and developer communities, ranked by signal.
- 2026-06-09 product_launch Google has reportedly ordered over three million Intel-manufactured TPUs for delivery in 2028. source
5 day(s) with sentiment data
-
Why GPUs Dominate AI Hardware Despite Specialized Alternatives
Despite the existence of specialized hardware like Groq's LPU and Cerebras' WSE, GPUs continue to dominate the AI landscape due to significant economic and structural barriers. These include the immense capital required…
-
NVIDIA faces backlash from neoclouds over alleged pressure tactics · 3 sources tracked
Executives at "neocloud" companies are reportedly feeling pressured by NVIDIA, leading some to consider offering alternative hardware like AMD GPUs and TPUs. This strategy aims to avoid NVIDIA's alleged high-pressure ta…
-
Bambu Lab A2L 3D printer review: The A1 grows up
Bambu Lab has released the A2L, an updated version of its A1 3D printer, featuring a larger build volume and enhanced capabilities. The A2L is priced at $469 for the standalone unit and $569 for a four-color combo, maki…
-
Google Colab CLI launches for terminal AI agents
Google has released a new command-line interface (CLI) tool for Google Colaboratory, enabling terminal-based AI agents to access GPUs and TPUs. This tool, available under the Apache-2.0 license, allows agents like Claud…
-
Groq LPU gains traction in AI inference, challenging GPU dominance
Groq's Language Processing Unit (LPU) is gaining traction in the AI inference market, moving beyond niche applications to become a recognized component in AI infrastructure. This shift is driven by the increasing demand…
-
Qualcomm acquires AI chip software firm Modular for $4B
Qualcomm is acquiring chip software startup Modular for nearly $4 billion in a deal that includes $300 million for Modular employees. This acquisition aims to bolster Qualcomm's expansion beyond mobile chips into areas …
-
Google taps Intel for 3M+ TPUs in 2028 amid TSMC packaging crunch
Google has reportedly placed an order with Intel for over three million Tensor Processing Units (TPUs) to be manufactured in 2028. This move comes as TSMC's advanced packaging solutions, particularly CoWoS, are heavily …
-
OpenAI's AGI philosophy, advanced AI cyber threats, and Google's Intel chip deal
OpenAI's Jakub Pachocki emphasized that AGI should benefit all of humanity and augment human agency, rather than diminish people's importance. Separately, an analysis of Anthropic-blocked accounts revealed attackers are…
-
Google Colab CLI enables terminal-based remote GPU/TPU code execution
Google has released an open-source command-line interface (CLI) for Colab, allowing developers and AI agents to execute Python code on remote GPUs and TPUs directly from their terminals. This tool simplifies the process…
-
ZML releases hardware-agnostic AI inference stack
ZML has released a new production inference stack designed to run AI workloads independently of specific hardware. The system aims to compile any AI model directly to NVIDIA, AMD, TPU, or Trainium accelerators without r…
-
Google taps Marvell to boost AI TPU performance
Google has partnered with Marvell to enhance its Tensor Processing Units (TPUs) used for AI workloads. This strategic alliance aims to bolster the development and performance of AI-specific chips. The collaboration is e…
-
New MARS LLM architecture uses internal state to override prompts
A researcher has developed a new language model architecture called MARS, which incorporates "proprioceptive channels" to allow the model to perceive its own internal state, such as memory salience or caution level. Ini…
-
Alphabet raises $80B equity to fund AI infrastructure expansion
Alphabet, Google's parent company, announced plans to raise $80 billion in equity capital to expand its AI infrastructure and compute capacity. This significant funding round, which includes a $10 billion investment fro…
-
AI Infrastructure Buildout Faces Power, Supply Chain, and Investment Challenges
The AI infrastructure buildout is entering a more complex phase, with vendors and operators facing challenges in power, supply chains, and geopolitical risks. Companies are shifting focus from solely GPUs to broader AI …
-
BYD launches 4nm Xuanji A3 AI chip, challenging NVIDIA
BYD has unveiled its self-developed "Xuanji A3" AI chip, manufactured on a 4nm process, positioning it as a direct competitor to NVIDIA's high-end offerings. This chip boasts a combined computing power exceeding 2100 TO…
-
AI drives demand for optical comms, Anthropic secures $36B for Google TPUs
Chinese optical communication product manufacturers are experiencing a surge in international demand, with orders extending to 2028 and significant stock market gains in related sectors. Concurrently, AI infrastructure …
-
Apollo, Blackstone eye $36B debt deal for Anthropic AI chip acquisition
Apollo Global Management and Blackstone are reportedly seeking additional investors for a $36 billion debt financing deal. This funding is intended to help Anthropic PBC acquire AI chips, specifically Google's custom TP…
-
Google AI animates physical materials into short film for I/O 2026
Google AI has created a short film using physical materials like cardboard and markers, animated with AI. The film features Google's TPUs preparing for the Google I/O 2026 event. This project was a collaboration with di…
-
Pallas DSL Enables Custom Kernel Optimization on Google TPUs
This guide introduces Pallas, a Python-based domain-specific language (DSL) designed for writing custom kernels that can be compiled and optimized for Google's Tensor Processing Units (TPUs). Pallas aims to provide deve…
-
Orbax checkpointing system speeds up AI model training
Training large AI models is vulnerable to hardware failures and other disruptions, making robust checkpointing systems essential. Orbax is a high-performance saving system designed to handle massive AI models by breakin…