OpenAI Develops Custom Chip to Reduce Costs
In brief
- OpenAI has developed a custom chip called the Jalapeño to lower its infrastructure expenses.
- This ASIC, created with Broadcom, aims to cut costs tied to third-party hardware.
- Currently, Nvidia holds an estimated 75% profit margin in this sector.
- By using their own chips, OpenAI can reduce capital spending and improve efficiency.
- The move underscores a shift towards more cost-effective solutions in AI development.
- Watch for further details on how this impacts the industry's economics.
Terms in this brief
- ASIC
- An Application-Specific Integrated Circuit is a type of computer chip designed for a specific task, like processing AI computations. OpenAI created Jalapeño, an ASIC, to reduce costs by optimizing hardware for their needs, rather than using general-purpose chips from companies like Nvidia.
- Jalapeño
- OpenAI's custom chip, developed with Broadcom, designed to lower infrastructure expenses by improving efficiency and reducing reliance on third-party hardware. This move aims to cut capital spending and challenge Nvidia's dominance in AI hardware.
Read full story at AI News →
More briefs
News Outlets Sue OpenAI for Copyright Infringement
News outlets updated their lawsuit against OpenAI, saying Microsoft encouraged users to plagiarize their work. The news outlets claim OpenAI's chatbots distort their work by providing incomplete or inaccurate summaries. This hurts their ability to sell original content. Over 10 news outlets are involved in the lawsuit. The lawsuit could cost OpenAI billions of dollars in damages. The case will continue in court.
Hartford HealthCare Launches AI-Powered Chatbot
Hartford HealthCare has launched an AI-powered chatbot that interprets lab results and answers patients' questions based on their medical records. The tool is built directly into the patient portal, giving more personalized responses. The chatbot is significant because around 32% of adults nationwide use AI for health information or advice. This tool provides a seamless and real-time conversation with a patient's medical record. The new tool will help patients understand their health information better. It will be available 24/7 to help users interpret lab results and answer health questions.
Qualcomm Takes Aim at Nvidia in AI Chip Market
Qualcomm plans to challenge Nvidia's dominance in the AI chip market. Qualcomm's CEO Cristiano Amon presented a five-year plan to investors to increase sales of AI components in data centers. Qualcomm's goal is to make over $15 billion from AI components by 2029. The company also expects to make $40 billion from businesses outside of handsets by 2029. This is double what was forecast two years ago. Qualcomm will offer power-efficient CPUs to stand out in the market. The company's shares went up 15% after the announcement. Qualcomm is also expanding into other areas like automotive and PC chips. The company just bought AI software company Modular for $3.9 billion. Qualcomm will now compete with Nvidia in the AI chip market. Qualcomm will continue to work on its AI chip plans in the coming years.
AI Agents Now Remember Conversations Across Days
AI agents can now remember details from previous interactions across days, marking a significant leap in their capabilities. Previously limited to handling single questions or short exchanges, these advancements allow agents like NVIDIA's to maintain context over extended periods. For instance, an agent can recall past user preferences and tailor responses accordingly. This improvement is crucial for developers and researchers aiming to create more intuitive AI systems that match human-like interaction. By retaining information across multiple exchanges, these agents can provide more coherent and personalized assistance, enhancing user experience in applications like customer support or personal assistants. Looking ahead, expect further developments in memory retention and contextual understanding, potentially enabling even longer-term recall and more sophisticated conversational flows.
Google's New AI Tech Speeds Up Features on Pixel Phones
Google has found a way to make AI features like notification summaries and message proofreading faster and more efficient on its Pixel phones. The company retrofitted Multi-Token Prediction (MTP) onto existing Gemini Nano v3 models, which are "frozen" and optimized for mobile devices. This new architecture allows the phone's AI to generate multiple tokens of text at once, significantly reducing the time and energy it takes to perform these tasks. This advancement is particularly important because mobile devices have limited processing power and battery life compared to servers. Traditional language models process one word at a time, creating a bottleneck that slows down performance and drains battery. By using MTP, Google claims that features like AI Notification Summaries and Proofread now work faster and consume less energy. For developers, this means they can build high-speed on-device AI features without needing to create separate, memory-heavy models for each task. The new approach is already available on the Pixel 9 and 10 series. Google says this marks a major step forward in making AI more accessible and efficient for everyday users.