Microsoft data centers, China's chips & more (January 14, 2026)

Good morning,

AI momentum keeps accelerating across chips, infrastructure, and enterprise workflows. China is pushing hardware independence, Big Tech is racing to scale compute, and startups are still pulling in serious capital.

Let’s dive in 👇

━━━

Want to engage in conversation about AI news? Join the Byte-Sized Reddit Community

🚀 Platform & Infrastructure Moves

⚡ Microsoft announces massive data center expansion

Microsoft announces a glut of new data centers while promising consumers their electricity bills will not increase. The company is framing efficiency gains and long-term power contracts as offsets to soaring AI compute demand. This underscores how energy costs are now a first-order constraint in AI strategy, not just a back-office concern.

🇨🇳 China’s Zhipu rolls out AI model trained on Huawei chips

China’s Zhipu rolls out AI models trained entirely on Huawei’s domestic chips, signaling progress toward reducing reliance on U.S. semiconductors. The release highlights China’s push for vertically integrated AI stacks, from silicon to models. It also raises questions about performance tradeoffs, scalability, and how quickly domestic ecosystems can close gaps with Nvidia-backed systems.

🧑‍💼 Airbnb hires Meta’s former AI head as CTO

Airbnb hires Meta’s former AI chief to lead its technology organization, signaling deeper AI integration across product and operations. The move suggests Airbnb is shifting from experimentation to execution. Talent migration from Big Tech to consumer platforms continues as AI becomes core infrastructure, not a feature.

🏢 Enterprise, Chips & Funding

💬 Salesforce launches new Slackbot AI agent

Salesforce rolls out new Slackbot AI agent as competition with Microsoft Copilot intensifies. The agent focuses on task execution inside conversations, not just summarization. This reflects a broader shift toward embedded agents living where work already happens.

🧠 Chinese chip designer Montage backed by Alibaba

Chinese chip designer Montage secured Alibaba backing ahead of a potential Hong Kong listing. The investment highlights strategic alignment between cloud platforms and domestic chip suppliers. Capital is increasingly flowing to firms positioned as national infrastructure plays.

🎙️ Deepgram raises $130M, hits $1.3B valuation

Deepgram raises $130M at a $1.3B valuation and acquires a Y Combinator AI startup. Speech-to-text demand continues to surge across call centers, media, and agents. The deal signals consolidation as mature AI verticals prioritize scale and defensibility.

🛠️ Tools of the Day

→ Remio AI Note Taker – Automatic meeting notes and summaries with minimal setup.
→ Vellum AI – LLM experimentation, prompt evaluation, and production testing.
→ Papercuts – Turn long documents into structured, usable outputs.

⚡ Quick Hits

→ Google’s Veo update adds vertical video creation from reference images.
→ Hegseth pushes Grok into U.S. military networks.
→ Skild Robotics valued above $14B after SoftBank round.
→ Converge Bio raises $25M for AI drug discovery.
→ Watchdog warns on Google AI shopping agents.
→ ElevenLabs crosses $330M ARR milestone.
→ DeepSeek memory fix cuts wasted GPU cycles.
→ Nvidia DLSS 4.5 update goes live.

🧠 TLDR

China is accelerating AI self-reliance across chips and models, while U.S. tech giants race to scale data center infrastructure without triggering energy backlash. Enterprise AI competition is intensifying inside everyday tools like Slack, and talent continues flowing from Big Tech to consumer platforms. Funding remains strong for speech, robotics, and biotech, signaling where AI value is concentrating.

Cheers,
David