March 9 – 15, 2026

Caitlin Kalinowski, OpenAI's head of robotics and consumer hardware, resigned following the company's partnership with the Pentagon. Her departure highlighted growing internal tensions at AI companies over defence collaboration.

Why it matters: Shows internal conflicts at AI labs over military partnerships, signalling governance challenges as defence becomes a major revenue driver.

Australian software company Atlassian laid off approximately 1,600 employees (10% of its workforce) to redirect nearly $236 million toward AI development. The company also replaced its CTO with two AI-focused CTOs.

Why it matters: Demonstrates enterprise software firms are restructuring teams and budgets for AI integration, with restructuring costs rising across the sector.

Meta unveiled four new generations of its MTIA (Meta Training and Inference Accelerator) custom AI chips — the 300, 350, 400, and 500 series — designed to reduce dependency on NVIDIA accelerators by 2027. The roadmap signals major tech companies are vertically integrating AI infrastructure to cut costs and escape supplier constraints.

Why it matters: Major tech companies vertically integrating AI infrastructure, reducing NVIDIA's monopoly power and reshaping the compute supply chain.

NVIDIA released Nemotron 3 Super, a 120-billion-parameter hybrid MoE (mixture-of-experts) model with 12B active parameters optimised for agentic reasoning. The open-source model is designed to run efficiently on smaller hardware, enabling smaller organisations to deploy frontier-class AI agents affordably.

Why it matters: MoE architecture lets smaller orgs run large models affordably, democratizing frontier AI and accelerating agent adoption across mid-market.

Google Maps introduced "Ask Maps" for conversational natural-language place searching and "Immersive Navigation" with AI-analysed imagery for enhanced driving routes. The update rolled out on Android and iOS in the US and India — the largest Maps overhaul in over a decade.

Why it matters: Conversational interfaces becoming table-stakes for consumer apps; LLM integration driving engagement and distinguishing products in mature markets.

Google officially completed its acquisition of cloud security platform Wiz for $32 billion, integrating AI-powered security tools into Google Cloud. Wiz is used by 40% of Fortune 500 companies, making it a major security-first foothold for Google's enterprise AI strategy.

Why it matters: Demonstrates big tech paying premium valuations for enterprise access; security becomes a key entry point for AI-driven cloud services.

Anthropic announced the Claude Partner Network with a $100 million commitment for 2026 to help consulting firms and system integrators deploy Claude in enterprise settings. Claude also gained cross-application capabilities for Excel and PowerPoint workflows.

Why it matters: Enterprise AI adoption depends on partner ecosystems; direct integration into office tools signals Anthropic's push into workplace productivity warfare with OpenAI.

Advanced Machine Intelligence Labs, chaired by Meta AI chief Yann LeCun, secured a $1.03 billion seed round co-led by Cathay Innovation, Greycroft, Hiro Capital, and Bezos Expeditions. The startup focuses on "world models" that learn physical laws rather than relying on language model approaches.

Why it matters: Validates alternative paradigms to LLMs; world models could unlock embodied AI and robotics, attracting mega-scale funding from tech leaders.

ByteDance and Peking University released Helios, a 14-billion-parameter video diffusion model capable of generating 60-second videos on a single H100 GPU. The model is available under an Apache 2.0 open-source license, competing directly with proprietary tools like OpenAI's Sora.

Why it matters: Open-source video generation competes with proprietary Sora, reducing barriers to video AI and threatening premium positioning of closed models.

Morgan Stanley published a report warning that a "Transformative AI" breakthrough is expected in the first half of 2026, projecting a US power shortfall of 9–18 gigawatts through 2028 as data centres strain energy infrastructure.

Why it matters: Power constraints may become the binding limit on AI scaling; energy costs and grid capacity will reshape competition and investment priorities.

Swansea University research studying 800+ participants designing virtual environments found that AI functions as a creative collaborator, enhancing ideation rather than simply replacing human work. Participants using AI tools reported higher satisfaction and more ambitious design goals.

Why it matters: Challenges pure displacement narrative, informing workplace policy and supporting evidence-based arguments for AI augmentation strategies.