Briefings

The lunch briefing.

Mid-day check: Broadcom's OpenAI chip dilemma, DeepSeek's massive funding round, and China's new algorithm regulations.

RIGHT NOW, IN ONE BREATH

AI Investment & Infrastructure. The AI sector continues to attract significant capital, with DeepSeek reportedly raising $7.3 billion at a $51.5 billion valuation, marking a potential record for Chinese AI startups. Meanwhile, Nvidia is solidifying its ecosystem role by investing over $40 billion in equity bets this year across the AI infrastructure stack. However, challenges persist as Broadcom is reportedly hesitant to build OpenAI's custom chip without a 40% purchase commitment from Microsoft.

Frontier AI Capabilities. Breakthroughs in AI model capabilities are pushing boundaries, as evidenced by Subquadratic's debut of a 12 million token context window, significantly expanding processing capacity. A Fields Medalist also reported that ChatGPT 5.5 Pro delivered "PhD-level" mathematical research in under two hours with zero human assistance, highlighting advanced problem-solving potential. This rapid progress raises questions about the future of human-AI collaboration in complex research fields.

AI Regulation & Ethics. Governments are increasingly addressing the societal implications of AI, with China ordering major tech platforms to tighten algorithm rules to ensure compliance and user protection. Concurrently, the "Wild West" of AI kids' toys is drawing scrutiny, with lawmakers considering bans due to concerns over data privacy and developmental impact. These regulatory actions underscore a global effort to balance innovation with responsible AI deployment.

Embodied AI & Autonomy. The development of embodied AI is accelerating, with SynapX launching SYNData, a multimodal data collection system designed to gather human manipulation data for robot learning. In the automotive sector, Volvo plans to integrate Geely's Qianli HaoHan G-ASD autonomous driving solution into its vehicles by 2027, signaling a push towards advanced L2-L4 capabilities. Chinese industrial embodied AI company Xiaoyubot also secured a new multi-hundred-million yuan funding round, backed by major industrial capitals like Xiaomi and Didi.

55
Sources scanned
4,300
Headlines processed
#10
Edition
12k
Discussing now

Broadcom reportedly won't build OpenAI's custom chip without Microsoft's 40% purchase

Broadcom will not finance production of OpenAI's custom AI chip unless Microsoft commits to buying 40 percent of the chips. OpenAI manager Sachin Katti called this dependency "financially unattractive" in an internal message.

DeepSeek reportedly raising $7.3B at $51.5B valuation, V4.1 debuts in June

DeepSeek is in talks to raise its first external round of up to 50 billion RMB, with founder Liang Wenfeng contributing 20 billion RMB personally. This deal would mark the largest single financing in China AI startup history.

Nvidia embraces role of AI investor, pushing past $40 billion in equity bets this year

Nvidia is pouring billions of dollars into companies across the AI infrastructure stack. The company is also signing commercial deals with these firms.

Fields Medalist says ChatGPT 5.5 Pro delivered PhD-level math research in under two hours

Fields Medalist Timothy Gowers had ChatGPT 5.5 Pro tackle open problems in number theory, with the model improving an exponential bound to a polynomial one. An MIT researcher called the key idea "completely original."

The context window has been shattered: Subquadratic debuts a 12M token window

Subquadratic has introduced a 12 million token context window, a significant advancement in large language model capabilities. This development expands the amount of information AI models can process at once.

China orders major tech platforms to tighten algorithm rules

China's cyberspace regulator announced that a nationwide campaign targeting problematic recommendation algorithms has yielded initial results. 14 major platforms, including Meituan and Taobao, have implemented 63 optimization measures.

The new Wild West of AI kids’ toys draws lawmaker scrutiny

Connected AI kids' toys could disrupt make-believe and bedtime stories, leading some lawmakers to consider banning them. Concerns revolve around data privacy and the potential impact on child development.

SynapX launches SYNData: Multimodal data collection system for embodied AI era

SynapX has released SYNData, a multimodal data collection system for dexterous manipulation. The system covers ego vision, EMG signals, and exoskeleton data gloves to enable scalable collection of human manipulation data for robot learning.

Xiaoyubot raises new round, backed by Xiaomi, Didi, BAIC

Xiaoyubot, a leading industrial embodied AI company in China, secured a new multi-hundred-million yuan funding round. Major industrial capitals including Xiaomi, Didi, BAIC, Fosun, and CCDC participated.

Volvo to adopt Geely's Qianli HaoHan G-ASD autonomous driving solution, launching 2027

Volvo Cars plans to integrate Geely's self-developed Qianli HaoHan G-ASD autonomous driving system, co-developed with Stepfun. The L2-L4 capable solution is expected to debut on Volvo vehicles in early 2027.

Banned drones and routers in the US will still get critical updates until 2029

The Federal Communications Commission issued a notice to allow software and firmware updates for banned drones and routers until January 2029. This measure ensures continued security and functionality for existing devices.

ICANN announces new gTLD round 14 years after the last one

ICANN has reopened applications for new top-level domains, allowing organizations to create custom extensions. This marks the first new round of gTLD applications in 14 years.

GrapheneOS fixes Android VPN leak Google refused to patch

GrapheneOS has implemented a fix for an Android VPN leak that Google had previously declined to patch. This addresses a significant security vulnerability for users.

LLMs corrupt your documents when you delegate

Research indicates that large language models can corrupt documents when users delegate tasks to them. This highlights potential issues with data integrity and reliability in AI-assisted workflows.