Briefings

The morning briefing.

While you slept: Google's AI model costs surge, Samsung workers prepare for a major strike, and the UAE pivots oil wealth to AI infrastructure.

RIGHT NOW, IN ONE BREATH

AI Model Economics. The cost of advanced AI models is a growing concern, with Google's Gemini 3.5 Flash reportedly costing 5.5 times more to run than its predecessor. This trend, also observed with Anthropic and OpenAI, suggests that the massive investments in AI development are now translating into higher operational expenses for users. The increased cost per interaction, particularly for agent tasks, indicates a shift towards more expensive, yet potentially more capable, AI services.

Global AI Infrastructure. Nations and companies are making significant moves to bolster their AI infrastructure. The UAE's exit from OPEC is expected to free up oil wealth, which the nation plans to redirect towards substantial AI investments and data center expansion. Meanwhile, China's Alibaba subsidiary, T-Head, has unveiled its Zhengwu M890 AI chip, boasting a 3x performance leap, signaling continued advancements in specialized AI hardware. These developments highlight a global race to secure compute and energy resources for AI.

AI Policy and Ethics. The regulatory landscape for AI is taking shape, with reports indicating that Trump's AI Executive Order will make sharing models with the government voluntary, potentially easing compliance burdens for developers. However, ethical concerns persist, as seen with an AI engineer claiming unfair dismissal from Google after protesting its work for the Israeli government. Public sentiment also reflects caution, with a study finding more fear than hope regarding AI's impact on the future of work.

Tech Labor and Platform Scrutiny. Labor disputes are impacting major tech players, with Samsung workers preparing for an 18-day strike in Korea over bonus payment negotiations, potentially affecting memory chip supply. Simultaneously, online child safety campaigners are urging a US inquiry into Roblox, alleging its design and business model conflict with children's developmental needs. These events underscore growing scrutiny on tech companies regarding both their internal labor practices and the societal impact of their platforms.

55
Sources scanned
5,448
Headlines processed
#53
Edition
10.5k
Discussing now

Samsung workers set to strike at worst possible time

More than 47,000 Samsung Electronics workers are preparing for an 18-day strike after bonus payment negotiations failed. The strike will be limited to Samsung's domestic chipmaking plants, raising concerns about memory chip production.

Google's Gemini 3.5 Flash is significantly pricier than previous models

Google's Gemini 3.5 Flash costs 5.5 times more to run than its predecessor, with total costs for agent tasks exceeding Gemini 3.1 Pro by 75 percent. This trend of increasing AI model prices is also observed with Anthropic and OpenAI.

Alibaba's T-Head unveils Zhengwu M890 AI chip with 3x performance leap

Alibaba's chip design subsidiary T-Head has launched the Zhengwu M890, a new training-and-inference AI chip. Official data indicates overall performance improvements of up to 3x compared to the previous generation.

Trump’s AI Executive Order will reportedly make sharing models with government voluntary

Reports suggest that Trump's upcoming AI Executive Order will make sharing AI models with the government voluntary. This approach differs from previous proposals that mandated such sharing.

‘Ask YouTube’ brings AI-powered conversational search to video, adds Gemini Omni to Shorts

Google is revamping its search experience by introducing 'Ask YouTube,' an AI-powered conversational search feature for videos. The company is also integrating Gemini Omni into YouTube Shorts.

AI coworker startup Viktor raises €64.7 million Series A

Viktor, an AI startup developing an AI coworker for Slack and Microsoft Teams, has raised €64.7 million in Series A funding. The company achieved a €12.9 million revenue run rate within 10 weeks of launch.

Zurich-based AVIAN raises €2.2 million for AI thermal monitoring

AVIAN, a Zurich-based industrial AI company, has raised €2.2 million in pre-Seed funding to prevent industrial fires with 24/7 AI thermal monitoring. The company plans to expand its engineering and deployment capacity.

Tianji Intelligence ships 10,000+ force-controlled humanoid arms in Q1

Guangdong Tianji Intelligence System secured over 10,000 orders for force-controlled humanoid robot arms in Q1 2026. The company also unveiled four new force-controlled humanoid arm products.

GitHub says internal repos exfiltrated after poisoned VS Code extension attack

GitHub has reported that internal repositories were exfiltrated following an attack involving a poisoned VS Code extension. Initial assessments suggest that customer data was not affected.

Online child safety campaigners call for US inquiry into Roblox

Online child safety campaigners, including Jonathan Haidt, have urged the Trump administration to investigate Roblox. They claim the platform's design and business model conflict with children’s developmental needs.

Lenovo unveils Tianxi AI 4.0 Ecosystem with palm-sized AI host

Lenovo Group has released the Tianxi AI 4.0 intelligent agent system, featuring a palm-sized AI host and a full ecosystem of devices. The system boasts bionic memory, a personal knowledge base, and AI simultaneous interpretation.

AI engineer says Google unfairly sacked him after protesting work for Israel

An AI engineer is challenging Google legally, claiming unfair dismissal after protesting the company's work for the Israeli government. He distributed flyers in Google DeepMind's London offices regarding military AI.

The UAE’s OPEC exit frees up oil wealth as it bets big on AI

The UAE's exit from OPEC will free up oil revenue, which the nation plans to invest heavily in AI infrastructure. This move will provide deeper pockets for AI investment funds and more gas for data centers.

Learnings from 100K lines of Rust with AI (2025)

A new report details insights gained from developing 100,000 lines of Rust code with the assistance of AI. The findings highlight best practices and challenges in AI-assisted software development.