Before we get into tech hiring, the important national update: England beat Norway 2-1 in the World Cup quarter-final, Jude Bellingham dragged the country through extra time, and it is, obviously, coming home.

Back to the tech market, which has somehow found a way to be almost as dramatic. OpenAI launched ChatGPT Work powered by GPT-5.6. Meta is pushing its own AI chip into production while planning to double compute capacity to 14GW. Microsoft cut 4,800 jobs as AI spending keeps climbing. China is considering restrictions on overseas access to its top AI models. And Meta’s own AI image detector failed to identify some cropped images created by Meta’s own model, because apparently the machines are now confusing themselves. Inspiring.

The hiring signal this week: AI is not just creating new jobs. It is forcing companies to decide which roles build leverage, which roles get automated, and which roles keep the whole increasingly expensive machine from breaking.

The Drop

1) OpenAI launches ChatGPT Work powered by GPT-5.6

What happened: OpenAI launched ChatGPT Work, a new enterprise agent combining ChatGPT with Codex-style coding capabilities and broader workplace task automation. It is powered by GPT-5.6, which launched after a short delay linked to US government national security concerns.

Why it matters for hiring: The AI agent battle has moved from developer tools into white-collar workflows. This is no longer just “engineers using coding assistants.” It is employees asking AI to create websites, documents, presentations, spreadsheets, and operational outputs across business systems.

Roles likely to spike:

• AI workflow product engineers

• Enterprise solutions engineers

• AI implementation consultants

• Internal tools engineers

• Governance and auditability specialists

Hiring takeaway: If your company is hiring for “AI experience,” get more specific. The bar is shifting toward people who can turn messy business workflows into measurable AI-assisted systems.

2) Meta pushes its own AI chip into production and targets 14GW of compute

What happened: Meta plans to put its in-house AI chip, Iris, into production in September. The chip is part of Meta’s MTIA programme and is aimed at reducing reliance on external suppliers. Meta is also targeting 14GW of computing capacity by next year and could spend up to $145B on AI infrastructure this year.

Why it matters for hiring: Big Tech is vertically integrating the AI stack: chips, data centres, energy, memory, networking, and software. The “AI engineer” market is becoming much wider than LLM app development.

Roles likely to spike:

• ML systems engineers

• Hardware-aware performance engineers

• Compiler and runtime engineers

• Data centre infrastructure engineers

• AI platform and capacity planning leads

Hiring takeaway: Companies that understand the full AI infrastructure stack will hire better than companies still lumping everything under “machine learning.”

3) Microsoft cuts 4,800 roles while AI investment keeps rising

What happened: Microsoft cut roughly 4,800 jobs, around 2.1% of its workforce, including a major restructuring of its Xbox business. The cuts are part of the wider trend of large tech companies reducing headcount while redirecting investment into AI infrastructure and higher-leverage teams.

Why it matters for hiring: This is the new labour market pattern: cut in one pocket, invest in another. Generalist and lower-leverage roles are under pressure, while AI infra, security, enterprise AI adoption, and platform roles keep getting budget.

Roles likely to be available:

• Gaming engineers and producers

• Commercial and sales operations

• Programme managers

• Enterprise software specialists

Roles likely to stay competitive:

• AI infrastructure

• Platform engineering

• Cloud security

• Data centre and compute efficiency

• Enterprise AI implementation

Hiring takeaway: Layoff markets are not automatically easy markets. The best candidates still get snapped up quickly, especially if they can tie their work to AI leverage, cost reduction, or revenue impact.

4) China weighs a “silicon curtain” around its top AI models

What happened: Beijing is considering curbs on overseas access to China’s most advanced AI models, including some open-weight models. This would mirror the logic of US restrictions on advanced AI capabilities, but in the other direction.

Why it matters for hiring: Many startups and engineering teams are using Chinese open models because they are capable, cheaper, and often strong in coding and agentic tasks. If access becomes restricted, teams will need better model-routing strategies, fallback plans, and cost modelling.

Roles likely to spike:

• Model evaluation engineers

• Open-model platform engineers

• AI infrastructure and routing specialists

• AI governance and geopolitical risk roles

• Enterprise architecture for multi-model systems

Hiring takeaway: “We’ll just use the cheapest open model” is not a strategy. It is a risk posture wearing a hoodie.

5) Meta’s AI image detector fails the crop test

What happened: Reuters found Meta’s AI image detector failed to identify 55% of cropped images created by Meta’s own Muse Image model, despite detecting the originals. Meta also discontinued a related AI image feature after privacy backlash.

Why it matters for hiring: AI trust and safety is still brittle. Watermarking, detection, privacy controls, and synthetic-media governance are becoming real engineering problems, not PR labels added after launch.

Roles likely to spike:

• AI trust and safety engineers

• Detection and watermarking researchers

• Privacy engineers

• Content integrity specialists

• AI product risk managers

Hiring takeaway: If your AI product can generate content, you need people who understand detection, misuse, privacy, and failure modes before users demonstrate them for you in public.

AI Tool of the Week

Findem

What it does: Findem is an AI talent intelligence and sourcing platform built around enriched talent data, relationship intelligence, automated sourcing, candidate rediscovery, talent analytics, and workforce planning.

Who it’s for: Hiring teams that need to move beyond keyword search and find candidates based on deeper attributes, career patterns, skills, companies, academic signals, and hidden talent pools.

Quick pilot idea this week:

• Pick one hard-to-fill AI infrastructure or platform role.

• Define 6 candidate attributes, not just keywords.

• Build a 50-person shortlist in Findem.

• Compare it against a LinkedIn-only Boolean search.

• Audit the top 20 profiles for relevance and diversity of background.

Metrics to track:

• Relevant profiles per hour

• % of candidates not found in your LinkedIn search

• HM approval rate

• Outreach reply rate

• Time-to-first-qualified-shortlist

Why this tool fits this week: The market is fragmenting. If the best hiring teams need to find AI infra, model-routing, security, and platform specialists, keyword-only sourcing is going to miss too many strong people.

Hiring / Interview Insight

Test for “AI operating leverage,” not just AI tool usage

This week’s stories all point in the same direction: companies are not just hiring people who can use AI tools. They are hiring people who can make AI workflows cheaper, safer, more reliable, and easier to govern.

Add a 30-minute “AI operating leverage” station:

Give candidates this scenario:

“You are rolling out an AI agent across a business team. It can access documents, write code, generate customer-facing assets, and trigger internal workflows. How do you decide what it can do, what it cannot do, what gets reviewed by a human, and how success is measured?”

Score for:

• Workflow decomposition

• Access control thinking

• Cost awareness

• Evaluation discipline

• Human escalation paths

• Privacy and data-handling judgement

• Ability to define measurable outcomes

Metrics to track:

• Pass-through rate by seniority

• New-hire 60-day manager satisfaction

• AI workflow error rate

• Cost per successful workflow

• Time saved per team after rollout

Why this matters: “AI-native” is not a personality trait. It is an operating discipline.

Funding Watch

SambaNova | $1B round | $11B valuation

AI chip and computing startup SambaNova raised $1B in a late-stage round.
Likely hires: AI hardware, systems engineering, compiler/runtime engineering, enterprise AI infrastructure.

Norm Ai | $120M Series C | $1.2B valuation

Legal AI startup Norm Ai raised $120M, showing continued demand for AI in regulatory and legal workflows.
Likely hires: legal AI product, compliance workflows, enterprise implementation, applied AI engineers.

Micron | $250B US investment plan through 2035

Micron increased its US investment plan again, committing more than $250B through 2035 as AI memory demand rises.
Likely hires: semiconductor manufacturing, memory systems, supply-chain engineering, advanced packaging.

Meta | C$13B Alberta data centre

Meta announced its first AI data centre in Canada, a 1GW facility in Alberta that can scale to 1.8GW.
Likely hires: data centre operations, power systems, cooling, networking, site reliability, infrastructure programme management.

Nanya | $6B planned 2027 spending

Taiwanese chipmaker Nanya plans $6B in spending in 2027 as memory demand rises with AI workloads.
Likely hires: DRAM, memory engineering, fab operations, chip supply chain, hardware systems.

Quick Bytes

• DeepSeek is reportedly developing its own AI chip, another sign that model companies are trying to control more of the infrastructure stack.

• China may allow select AI firms to buy limited Nvidia H200 chips, but access remains politically sensitive and tightly controlled.

• Big Tech’s AI infrastructure spending is now expected to exceed $700B this year, with debt and equity financing becoming a much bigger part of the story.

• The RAISE Summit in Paris focused heavily on AI cost efficiency, power use, and open/customisable models, which says a lot about where the real pain is now.

• Meta discontinued an AI image feature days after launch following privacy criticism, another reminder that “ship fast” gets expensive when trust is bolted on afterwards.

What to do this week

1) Add “workflow ownership” to AI job specs

Metric: every AI product or platform JD includes ownership of one business workflow, not just model/tool exposure.
Why: ChatGPT Work shows AI is moving into actual work execution, not just answering questions.

2) Build a multi-model talent pool

Target skills: GPT-5.6, Claude, DeepSeek, Kimi, Llama, Qwen, vLLM, routing, evals, inference optimisation.
Metric: 30 qualified profiles identified this week.
Why: China’s model-access risk means companies need people who can work across providers and avoid dependency traps.

3) Add trust-and-safety review to AI product interviews

Metric: every senior AI product or engineering candidate completes a scenario covering privacy, misuse, detection, and rollback.
Why: Meta’s image detection failure is exactly the kind of issue that becomes a hiring problem after launch.

This week’s takeaway is simple: the AI labour market is becoming more technical, more fragmented, and more unforgiving. The best hires will be the ones who can connect AI capability to workflow value, infrastructure cost, governance, and trust. Everyone else will keep writing “AI experience required” and wondering why the shortlist looks like a LinkedIn keyword accident.

That’s all for this week’s Tech Talent Drop - stay informed, and see you next week!