Open-weight AI got serious, AI infra startups got richer, and the physical cost of compute became impossible to ignore. Moonshot released Kimi K3, a 2.8 trillion-parameter open-weight model. Fireworks raised $1.51B at a $17.5B valuation to expand engineering and global compute. Helsing raised $1.8B at $18B, making European defence tech a premium hiring lane. Data centre opponents staged coordinated protests across 42 US states. And Thomson Reuters cut engineering roles while planning hundreds of senior AI-focused hires. The hiring signal: AI is fragmenting the market toward model deployment, inference efficiency, infrastructure, security, defence, and people who can turn AI into useful work without detonating cost, trust, or compliance.
Key Takeaways
Open-weight AI is enterprise-grade: hire open-model platform engineers, model evaluation engineers, inference optimisation engineers, AI governance and model-risk specialists, and multi-provider routing engineers
Model serving is the next hiring battleground: distributed systems, GPU infra, reliability and observability, and enterprise solutions engineers around Fireworks-style inference platforms
AI infra is now political infra: demand for data centre programme managers, energy and grid specialists, permitting leads, sustainability and community relations, and capacity planning and FinOps
Thomson Reuters shows the mix is changing, not just shrinking: senior product, AI workflow, legal-tech and tax-tech AI, and applied AI engineers in regulated workflows are still in demand
European defence tech is now competing with AI labs for elite engineers: robotics, autonomy, computer vision, embedded systems, C++/Rust systems, and security-cleared engineers
Add a 30-minute 'AI deployment judgement' interview station scoring model selection, cost and latency awareness, evaluation design, data privacy, escalation paths, and rollback planning
This week: build an open-model skills map, add inference efficiency to every AI JD, source defence-tech engineers, and audit your own CRM/ATS before buying more sourcing tools
Show Notes
Moonshot unveils Kimi K3, a 2.8 trillion-parameter open-weight model claimed as the world's largest, performing close to top US frontier systems
Fireworks raises $1.51B at a $17.5B valuation (Nvidia-backed) to expand engineering and global inference capacity
Data centre opponents stage coordinated protests across 42 US states with 140+ events focused on electricity, water, land, and community impact
Thomson Reuters cuts up to 500 engineering roles while planning 250+ net-new engineering hires, mostly senior and AI-focused across legal, tax, and regulatory workflows
Helsing raises $1.8B at an $18B valuation, cementing European defence tech as a premium hiring market
Smaller signals: TYLSemi $43M (custom AI chip blocks), Instalily AI $60M Series B (AI forward-deployed engineers), PixVerse $439M Series C extension (AI video), Emerald AI ~$100M (data centre management), Nous Research reported open-source agent funding talks
Quick bytes: 200+ experts including Nobel laureates urge action on AI economic impact; US IPO market near record proceeds on AI and data centre listings; IBM's AI transition under pressure; China's open-weight push turning into strategic pressure on US model companies
AI Tool Spotlight: Gem AI Sourcing — turns JDs and scoring criteria into candidate searches across CRM, ATS, and public profiles
Frequently Asked Questions
What does Kimi K3 change for hiring?
Open-weight AI is now a serious enterprise lane. 'OpenAI experience' is no longer enough — strong candidates must evaluate trade-offs across GPT, Claude, Kimi, DeepSeek, Llama, Qwen, and custom internal models, and know inference stacks like vLLM and TensorRT-LLM.
Why does Fireworks' raise matter for recruiters?
It confirms model serving, not just training, is where enterprise AI value is created or burned. Expect competitive demand for distributed systems, GPU infra, model routing, latency and cost control, and enterprise solutions engineers.
How should teams read the data centre backlash?
AI infra hiring cannot stop at 'more SREs'. Roadmaps now depend on power, permits, grid access, and public trust — hire programme managers, energy strategists, permitting leads, sustainability roles, and capacity planning specialists.
What does Thomson Reuters' cut-and-hire signal?
AI is changing the mix, not simply deleting engineering. Low-leverage roles get squeezed; senior engineers who build AI-enabled workflow products in regulated domains stay in demand.
How should a team pilot Gem AI Sourcing?
Pick one hard AI infrastructure, platform, or security role. Define six scoring attributes (not keywords). Run Gem across CRM, ATS, and public profiles. Compare against a LinkedIn-only search. Track relevant candidates per hour, % of shortlist from existing database, HM approval rate, outreach reply rate, false-negative rate, and time-to-first-qualified-shortlist.