OpenAI and Broadcom Launch Custom Chip for LLM Inference — IT News, June 25, 2026
Today's technology news highlights significant advancements in AI infrastructure, particularly with the introduction of a new custom chip by OpenAI and Broadcom designed to optimize large language model (LLM) inference. As demand for AI capabilities grows, builders must focus on efficient architecture and resource management to leverage these innovations effectively.
Software & Development
Companies are scrambling to stop employees from maxing out AI budgets with small tasks
Organizations are increasingly concerned about employees overusing AI resources for minor tasks, leading to budget overruns. This trend highlights the need for better management of AI usage within companies to ensure cost efficiency.
Hardware & Devices
OpenAI and Broadcom announce chip designed for LLM inference at scale
OpenAI has partnered with Broadcom to unveil a new chip specifically engineered for large language model inference, addressing the increasing demand for AI processing power. This collaboration signifies a pivotal moment in the AI hardware landscape, as companies seek to enhance performance and efficiency in AI applications.
OpenAI unveils its first custom chip, built by Broadcom
The new Jalapeño chip from OpenAI, developed in collaboration with Broadcom, marks a significant step in AI hardware development, aimed at improving the efficiency of AI models. This launch highlights the growing importance of tailored hardware solutions in meeting the demands of advanced AI workloads.
45°C cooling design cuts data center water use to near zero
A new cooling design for data centers has been introduced, significantly reducing water consumption while maintaining efficient operation at higher temperatures. This innovation is crucial for sustainable data center management, especially in regions facing water scarcity.
Acquisitions & Funding
Qualcomm to Acquire Modular
Qualcomm has announced its acquisition of Modular, a move aimed at enhancing its capabilities in modular software solutions. This acquisition reflects the ongoing trend of consolidation in the tech industry as companies seek to bolster their software offerings.
Cybersecurity
Anthropic says Alibaba illicitly extracted Claude AI model capabilities
Anthropic has accused Alibaba of unauthorized extraction of its Claude AI model capabilities, raising concerns about intellectual property in AI development. This incident underscores the vulnerabilities that AI companies face regarding their proprietary technologies.
New website names and shames companies that still don't offer passkeys to users
A new initiative has emerged to highlight companies that fail to implement passkey support, emphasizing the importance of secure authentication methods. As cybersecurity threats increase, the adoption of passkeys is becoming critical for protecting user data.
One-two punch delivered in global operation disrupts cybercrime 'assembly line'
A coordinated global effort has successfully disrupted key components of cybercrime operations, significantly impacting the tools used by cybercriminals. This operation highlights the ongoing battle against cyber threats and the importance of international cooperation.
Big Tech & Policy
AI was supposed to kill engineering jobs, but new data suggests they're the most resilient
Recent data indicates that engineering jobs are not only surviving but thriving in the age of AI, countering fears of widespread job losses. This resilience suggests a growing demand for skilled engineers who can work alongside AI technologies.
Google starts lowering Play Store fees, making good on Epic Games settlement
Google has begun to reduce fees on its Play Store as part of a settlement agreement with Epic Games, signaling a shift in its approach to app store monetization. This move may influence how developers strategize their app pricing and revenue models.
Summaries are original; all facts and full reporting belong to the linked sources. Compiled June 25, 2026 by Yogreet Global.
Every headline above is someone scaling — or paying for scale they didn't plan.
Yogreet Global is an infrastructure-first product engineering studio. We design AI-native products on microservices, structured functions and right-sized infrastructure — with the cost curve mapped before code ships, so you scale from 100 to 100,000 users at a price you planned for. The same lens we read the news with, we bring to your build.
Book a Build Audit