> daily_signal(2026_05_12)

GM laid off hundreds of IT workers, then said it's still hiring for the same department - for AI skills.

PickBits Daily Signal · Tuesday, May 12, 2026

By Mark Pickering · 6 min read · May 12, 2026

// tl;dr

General Motors began global layoffs of roughly 500 to 600 salaried IT workers Monday - and said it is still hiring for the same department, just for AI skills. GM's wanted list: AI-native development, data engineering and analytics, cloud engineering, agent and model development, prompt engineering, "new AI workflows." This is the AI reorg reaching a 116-year-old carmaker's back office, not a software company's.
GitLab opened a voluntary-separation window Monday, denied it was AI cost-cutting, then promised to reinvest the savings into AI. About 2,580 employees, applications due May 18, country footprint shrinking by up to 30 percent, management layers flattening. CEO Bill Staples: "this is not an AI optimisation or cost cutting exercise." Stock down about 7 percent after hours.
OpenAI launched a $10-14 billion enterprise-deployment venture Monday with Bain, Capgemini, and McKinsey signed on as investors, opened GPT-5.5-Cyber to the EU, and Anthropic put the Claude Platform native inside AWS - all the same day. Three channel moves, all the same play: the labs deciding how the AI you'll use at work gets sold to your employer. The consultancies grading those tools now own a slice of one vendor.
Trump named the CEOs flying to Beijing with him this week - Elon Musk, Tim Cook, Larry Fink, David Solomon, Kelly Ortberg. Jensen Huang wasn't invited. The White House framed the trip around "agriculture and commercial aviation, not chips"; Commerce Secretary Howard Lutnick said April 22 the H200 chips Trump cleared for China sale still haven't shipped. The guest list is the chip-export tell.

On Monday, May 11, General Motors told several hundred salaried IT staff their jobs were gone - and said it is still hiring into the same department, just for people whose skills say "AI." The same day, GitLab opened a voluntary-separation window, denied it was AI cost-cutting, and promised to reinvest the savings in AI. OpenAI launched a $10-to-14-billion enterprise-deployment venture with Bain, Capgemini, and McKinsey signed on as investors, opened its GPT-5.5-Cyber model to the EU, and Anthropic put the Claude Platform native inside AWS in 17 regions. And the White House confirmed the CEOs flying with Trump to Beijing this week - without Jensen Huang. AI is being used to rebuild who works where, and on Monday the rebuild showed up in an IT department, not just on a lab's announcement page.

Monday: GM cut its IT department to rehire it for AI, GitLab told its people to apply to leave, the consulting firms grading your company's AI tools bought into OpenAI, and the White House flew to Beijing without the chip CEO.

1. GM laid off hundreds of IT workers, then said it's still hiring for the same department - for AI skills.

On Monday, May 11, General Motors began global layoffs of roughly 500 to 600 salaried IT employees, concentrated in Austin, Texas and Warren, Michigan, per TechCrunch, Bloomberg, and CNBC. GM described it as trimming costs and resetting skills: the company is still hiring into IT, but for AI-native development, data engineering and analytics, cloud engineering, agent and model development, prompt engineering, and "new AI workflows." It is the latest of several white-collar reduction rounds GM has run over the past 18 months as it shifts headcount toward AI.

This is what "AI transformation" looks like when it reaches a 116-year-old carmaker's back office: not AI tools layered on top of the existing team, but the team itself rebuilt - the people whose résumés don't say "AI" out, the people whose résumés do say it in, often into the same boxes on the org chart. GM isn't a software company; it's the kind of employer where IT has been a stable, unglamorous career for decades. The signal for everyone whose company has an IT department: the AI reorg is not "we'll buy some Copilot licenses." It's a headcount question, and the answer is being written one department at a time.

TechCrunch coverage of GM laying off hundreds of IT workers to hire staff with AI skills — techcrunch.com · May 11, 2026

Why this matters: If you work in IT - or your spouse, sibling, or kid does - GM just drew the line of what "AI skills" means in practice: not "can use ChatGPT," but data engineering, cloud, agent and prompt work, and "AI workflow" design. The job description that was safe three years ago is the one being eliminated, often by the same employer that's posting the replacement req. Action this week: if you're in an IT or ops role, take GM's actual list (AI-native dev, data engineering/analytics, cloud engineering, agent/model dev, prompt engineering, AI workflows), pick the one you could credibly claim in six months, and put a concrete learning plan against it now - not at your next review. If you run a team, GM is the benchmark your CFO will cite: check whether your headcount plan has the same shape (fewer of role X, more of role Y, net flat) and whether your people have heard it from you. If you're a hiring manager, "AI skills" on a req is now a real screen - write it as named tools and tasks so you're filtering for the thing GM is filtering for.

https://techcrunch.com/2026/05/11/gm-just-laid-off-hundreds-of-it-workers-to-hire-those-with-stronger-ai-skills/

2. GitLab opened a voluntary-separation window Monday, denied it was AI cost-cutting, then promised to reinvest the savings into AI.

On Monday, May 11, GitLab CEO Bill Staples emailed the company's roughly 2,580 employees (the headcount as of January) telling them GitLab was opening a voluntary-separation window: applications due by May 18, final structure and financial impact disclosed at the company's June 1 earnings. The announcement listed three concrete moves alongside the offer to leave: reducing GitLab's country footprint by up to 30 percent, flattening the management chain by up to three layers, and reorganizing R&D into smaller AI-powered teams. Staples added, per Bloomberg: "this is not an AI optimisation or cost cutting exercise." Then the company promised to "reinvest the vast majority of savings" from the headcount reduction into AI. Shares fell about 7 percent in after-hours trading.

Continuing the labor-displacement arc. The naming is what's new. Cloudflare, PayPal, Coinbase, BILL, and Upwork all explicitly attributed their May-quarter cuts to AI (this daily covered the cohort on May 7 and May 8); Fidelity's 800-position cut announced May 7 went the other way and denied an AI rationale, then said the 3,300 new hires would work on AI-enabled tools. GitLab runs the Fidelity playbook: announce the cuts, deny AI, describe the reinvestment as AI. The denial is load-bearing - it keeps the company off the layoff trackers companies now compete to stay off of, and preserves the option to recall. "Voluntary separation" replaces "involuntary layoff," "transformation" replaces "restructuring," "reinvestment in AI" replaces the saved payroll. If you work in tech, this is the memo template you'll see at your company by the end of Q3.

The Register coverage of GitLab opening a voluntary separation window while pivoting toward AI — theregister.com · May 12, 2026

Why this matters: If you work in tech, the GitLab memo is the template most likely to land in your inbox this quarter: "voluntary separation" not "layoff," "transformation" not "restructuring," "reinvestment in AI" instead of the saved payroll - and the AI denial is doing real work, keeping the company off the trackers and the recall option open. Action this week: open layoffs.fyi and search your employer plus its three closest competitors - the tracker captures the announce-to-effective gap a "voluntary" framing tries to blur. If you're weighing a separation offer, the SaaS floor this quarter is roughly 1.5x your normal severance plus accelerated equity vesting through the bonus period; ask for it in writing. If you set comp or run HR, the GitLab-plus-Fidelity pattern (deny AI, simultaneously hire early-career replacements, call it transformation) is now the public benchmark every employee can read - draft accordingly.

https://www.theregister.com/devops/2026/05/12/gitlab-promises-a-different-kind-of-layoff-as-biz-pivots-toward-ai/5238422

Tomorrow's signal lands here. Subscribe to PickBits Daily Signal if you want the daily that leads with impact, not hype.

3. The labs spent Monday picking distribution channels - and the firms grading your AI tools bought into one of them.

Three moves from the frontier labs landed on Monday, May 11, and they're the same move: deciding which channels they sell through. OpenAI launched The Deployment Company, a majority-OpenAI-owned vehicle that closed $4 billion from 19 founding partners (lead investor TPG; co-leads Advent, Bain Capital, Brookfield) at a reported $10-to-14-billion valuation, and acquired the applied-AI consultancy Tomoro (about 150 "Forward Deployed Engineers") on day one. The part that touches you: three of the biggest enterprise consultancies - Bain & Company, Capgemini, and McKinsey & Company - signed onto the cap table as investors. The "GenAI roadmap" deck your employer paid one of them to write is now produced by a firm that owns a slice of one of the model vendors it's recommending. Denise Dresser, the ex-Slack CEO running the venture, calls the goal helping organizations "build and deploy AI systems they can rely on."

The other two moves: OpenAI gave the EU preview access to GPT-5.5-Cyber, its cyber-defense model, for the European AI Office, member-state governments, and national cyber authorities - while Anthropic kept refusing the same deal for Mythos, the Commission saying its talks with Anthropic are "at a different stage." And AWS and Anthropic made the Claude Platform generally available native inside AWS: the full Claude API, console, and agent tooling under your existing AWS account and bill, across 17 regions - distinct from the Bedrock-wrapped Claude that's been there since 2023. Read together: OpenAI is buying the people who run enterprise rollouts; Anthropic is buying the one-signature-instead-of-two path into companies that already buy AWS, and, separately, OpenAI walked through a regulatory door into the EU that Anthropic so far hasn't. Different bets, same Monday, on how the AI you'll use at work gets sold to your employer.

The Register coverage of OpenAI launching The Deployment Company and acquiring AI consultancy Tomoro — theregister.com · May 11, 2026

Why this matters: If your company has ever paid McKinsey, Bain, or a Big Four firm for an "AI strategy" deck - most large ones have - the firm advising which model to pick now has money riding on the answer. That's a conflict your contracts probably already say has to be disclosed, and probably wasn't written to cover this. Action this week: if you sit anywhere near an active engagement with any of the 19 named investor firms (the full list is on the OpenAI announcement), ask the engagement lead, in writing, whether the firm or any partner holds equity in OpenAI, Anthropic, or any vendor named in the deliverable - and add a cap-table-disclosure line to your next AI-services contract before signing. If you run any production AI that touches EU data or operations, the OpenAI-yes / Anthropic-no split is the cleanest current read on which lab will operate under EU rules; flag the single-vendor version of that as a risk before your next renewal. If you're already on AWS, the native Claude Platform path is live in 17 regions today - worth a look against whatever two-vendor plan you have.

https://www.theregister.com/ai-ml/2026/05/11/openai-buys-ai-consultancy-to-sell-enterprises-on-its-models/5238213

4. Trump named the CEOs flying to Beijing with him this week. Jensen Huang isn't on the list.

The Trump-Xi summit runs May 13 to 15 in Beijing. The White House confirmed the CEO delegation on Monday: Elon Musk (Tesla and xAI), Tim Cook (Apple), Larry Fink (BlackRock), David Solomon (Goldman Sachs), and Kelly Ortberg (Boeing), plus Stephen Schwarzman (Blackstone) and Jane Fraser (Citigroup), per CNBC and Bloomberg. NVIDIA CEO Jensen Huang - the executive who has accompanied Trump on the most foreign trips this year - was not invited. Bloomberg reported the White House framed the trip around "agriculture and commercial aviation, not chips."

The chip-export backstory is the harder constraint. On April 22, Commerce Secretary Howard Lutnick told the Senate Appropriations Committee that the H200 chips Trump cleared for sale to China in December 2025 still hadn't shipped: "we have not sold them any chips as of yet." Beijing hasn't authorized purchases at the volume the export license contemplates, and its preference for Huawei-built domestic AI accelerators is the unsubtle counterpoint. Putting Huang on the Beijing list this week would have made the chip-export deal one of the public deliverables of the summit - a lever the White House has decided not to spend. Bringing Tim Cook and Elon Musk instead lets the deliverables be Apple's manufacturing footprint and Tesla's Shanghai expansion, both of which already exist and need no new export-control change to celebrate.

CNBC coverage of Nvidia CEO Jensen Huang not joining Trump's CEO delegation to Beijing — cnbc.com · May 12, 2026

Why this matters: If you've been waiting on a laptop, phone, GPU, or a car whose price or timeline traces back to US-China trade, this week's summit guest list is the tell: chips are deliberately off the agenda, so don't expect the H200 logjam to clear. Action this week: open your 2026-27 hardware-order pipeline and add a 25 percent surcharge scenario for anything that routes through a Chinese fulfillment node, matching the rate Trump's December H200 deal already imposed - keep that line in the plan, not in the "maybe" column, and cite Lutnick's April 22 testimony ("we have not sold them any chips as of yet") if a board asks why. If you just want to know whether the trade war is thawing, watch the Cook and Fink readouts coming out of Beijing this week, not the chip headlines - that's where any real movement shows up first.

https://www.cnbc.com/2026/05/12/nvidia-ceo-huang-trump-china-trip-chip-sales.html

» What to watch this week

GM's next disclosure - and the first comparable non-tech employer to copy the move. GM's Q2 call (late July) is where the IT-headcount delta and the AI-skills hiring count land in one place. The bigger signal is the first comparable Fortune 500 outside tech (a bank, an insurer, a retailer, a hospital system) to announce the same "cut IT, rehire IT for AI" reset - that's when "GM did it" becomes "everyone is doing it."
GitLab Q1 FY27 earnings on June 1. The voluntary-separation count, the country-footprint reduction, the management-layer flatness, the rebased FY27 operating-expense guide, and the AI-reinvestment dollar figure all land in one disclosure. The number that matters is headcount-per-million-ARR: GitLab's pitch is that ratio falls and stays down.
Bain & Company, Capgemini, and McKinsey conflict-of-interest disclosures. Each firm has a public COI policy requiring partner-level disclosure on engagements where the firm holds equity. Watch for any of the three to update its policy page, send a partner memo, or amend a state-government RFP response naming the OpenAI stake - or for a customer to drop one of them from an AI-rollout engagement over it.
Anthropic's response to the EU "different stage" framing. The Commission named Anthropic publicly, which forces a response. The clean signal is an Anthropic blog post or a Dario Amodei statement; the messy one is a leak from the Brussels meetings. The push-forward trigger fires when Anthropic publicly takes a position.
The Tim Cook and Larry Fink readouts from Beijing. Apple's iPhone manufacturing split between India and China, and BlackRock's China asset-management commitments, are the deliverables most likely to get summit billing. Either announcement reframes the H200 chip silence as a deliberate omission, not an oversight.

Tomorrow's signal lands here.