Where Can Workers Turn? Nowhere. Here’s What to Build Instead.

Foreword: Every time I begin the process of pulling data together for this newsletter, I do so with optimism. Today is no different. The bad news is first. But I think there is hope and I believe we can navigate this acceleration. Our window is closing but not closed. We need to approach work differently than we have in the past. Just as we did with the “Information Age”. And just as people did during the Industrial Revolution. This is systemic change. I purposely chose not to include a top 5 today, even though that is the backbone of Under the Radar. The legal and medical news in the past couple of weeks has me re-thinking once again. What is the true path forward? This newsletter outlines some of my ideas. Please feel free to suggest your own. We have to move forward together to successfully navigate this.

Check out our interactive graphic “The Acceleration Gap” to map this unprecedented shift.

Bottom Line Up Front: Thursday’s employment data confirmed what workers already feel. Displacement is accelerating across every sector we were told was “safe,” and there’s nowhere left to turn. Individual retraining won’t solve structural automation. What works: adaptation through creative, flexible, collaborative solutions. Don’t plan for career paths that exist in 2026 and expect them to be there in 2030. Instead, use your expertise to build alternatives that create the economy we need, not just survive within the one that’s collapsing.

The Numbers Don’t Lie

Thursday’s jobs data:

JOLTs job openings: 6.542 million (expected 7.2M) — that’s 658,000 fewer openings than economists projected
Initial jobless claims: 231,000 (expected 212,000) — more people losing jobs
Continuing claims: 1,844,000 (up from 1,819,000) — staying unemployed longer

Translation: More people losing jobs. Fewer jobs to land in. Staying unemployed longer.

This isn’t a temporary blip. This is what acceleration looks like.

The Convergence: Every “Safe” Career Automating Simultaneously

Financial Services joined the automation wave on February 6, 2026, when Goldman Sachs announced deployment of Anthropic’s Claude to automate accounting and compliance roles. This is the same Anthropic that announced healthcare automation just 26 days earlier. Accounting and compliance aren’t entry-level positions. These require professional credentials, years of regulatory experience, and direct liability for accuracy. Goldman Sachs is among the most risk-averse institutions in global finance. If they’re confident enough to automate compliance work, they’ve tested extensively. The “safe” professional career advice just died again. One AI company, multiple professional sectors, under 30 days.

Manufacturing: Facing intense automation pressure, but workers are fighting back. VW Chattanooga just won union recognition after 2 years of negotiation. Manufacturing employment has declined from 17.3 million jobs in 2000 to around 12.9 million today. But those 12.9 million workers still exist, still matter, and are still organizing. Meanwhile robotics accelerate: Mercedes deploying Apollo humanoids, Unitree shipping 5,500+ robots in 2025, Chinese state coordination pushing “AI+ Manufacturing” at scale. The sector isn’t “gone”, but it is under coordinated assault with workers told to retrain rather than getting systematic support.

Legal: Market confidence crashed this week. Anthropic released a legal plugin for Claude on February 3rd. By Wednesday February 5th, $285 billion had evaporated from legal, software, and data companies. RELX down 15%, Thomson Reuters down 15%, Wolters Kluwer down 10%+. Two days of selling that deepened instead of recovering. Investors repriced these business models as facing fundamental automation threats. Legal work still happens, companies still operate, workers still have jobs. But the market just signaled it expects massive disruption. RELX hit its biggest one-day drop since 1988 and has nearly halved from its February 2025 peak.

Healthcare: Being automated right now. Anthropic’s January 11th announcement targets prior authorization review, claims appeals, medical coding, clinical documentation, chart reviews. Major partners already deploying: Banner Health, Novo Nordisk, Sanofi, Premier (serving 4,400+ healthcare organizations). And Anthropic isn’t alone. Google, Microsoft, OpenAI all have healthcare tools.

Agriculture: Robotics advancing across the entire crop cycle. Carbon Robotics’ LaserWeeder deployed on 100+ farms across 15 countries for automated weeding. John Deere’s autonomous tractors operating without drivers. Harvest automation robots picking fruits and vegetables (companies like Abundant Robotics, FFRobotics). Automated greenhouse systems managing entire growth cycles. Drone technology for crop monitoring, spraying, and field mapping. Milking robots handling dairy operations. Farm labor, already underpaid and undervalued, is now facing coordinated automation across planting, maintenance, and harvest. Workers who feed us getting displaced with minimal safety net.

Construction: Workers are training their own replacements for space deployment. Mercedes deploys Apollo humanoid robots in factories (officially for “labor shortages”), but Apptronik explicitly describes factory work as a “testing ground” for space reliability. China’s Unitree shipped 5,500+ humanoid robots in 2025 at $14,240 each. Meanwhile, SpaceX files for 1 million satellites as orbital data centers. Google’s project Suncatcher, Blue Origin has expressed interest as well.

What percentage of data center construction could robots handle right now? The repetitive tasks – moving materials, assembly work, loading pallets – are exactly what Apollo does in Mercedes factories. The massive data center boom (Oracle’s $7B Saline facility, Meta’s $50B Louisiana campus) creates both demand for robotics AND the controlled environments perfect for training them.

Workers aren’t just being displaced. They’re being used as R&D for orbital deployment. Communities approve 25-year commitments to ground-based infrastructure. Musk plans a million satellites. Construction workers train the robots that will build in space, while the facilities they’re building may be obsolete before completion.

Computer Science: The field is automating itself. If AI can code, debug, architect, test, and deploy – what exactly is the CS career? Starting a four-year CS degree in 2026 means graduating in 2030 when AI has evolved several generations beyond current capabilities. Bootcamps train for entry-level jobs are vanishing faster than programs complete. Some ultra-specialized niches might survive (cryptography, safety-critical systems, AI alignment research), but these require advanced degrees or years of existing experience. You can’t start there.

The honest assessment: Don’t bet your career on traditional computer science. But CS skills remain valuable as tools for other work. Building organizing platforms, creating intelligence tools (what The Open Record does), supporting worker-owned tech co-ops, contributing to open-source public infrastructure. Learn enough to be dangerous. Use it to build alternatives. Don’t expect to make a 30-year career writing code for corporations.

Students graduating 2026-2028 face markets where “safe” careers are automating faster than retraining cycles. Workers were told these fields were refuge after manufacturing displacement. Now professional services are automating simultaneously while communities subsidize infrastructure that may be obsolete before activation.

Meanwhile: Extraction Without Coordination

The same day the legal sector’s market confidence crashed, Fortune reported Meta quietly purchased 1,400 additional acres in Louisiana. Expanding its already-massive Hyperion AI data center to four times the size of Manhattan’s Central Park.

The Investment:

$27-50 billion total (Trump claimed $50B in August 2025)
3,650 total acres when fully built
500 permanent jobs promised (by 2034)
Economically depressed region – Richland Parish, Louisiana (population 20,000)

What’s This Costing Louisiana Taxpayers?

Meta gets a minimum $3.3 billion in tax subsidies. That’s full exemption from all state and local taxes on equipment. Louisiana’s combined tax rate is 9.56%; industry estimates put Graphics Processing Unit (GPU) costs alone at $35 billion. If Meta paid what residents pay, that’s $3.3 billion to state coffers. Instead: $0.

Meanwhile, taxpayers and ratepayers cover $3+ billion for three new gas-fired power plants built exclusively for Meta. Plus road damage (600% surge in accidents), infrastructure strain, water consumption (5 million gallons daily), and an elementary school playground shut down for safety.

Louisiana sold Meta the land for $12 million. Neighboring parcels now sell for $40,000/acre. At market rate, the state gave Meta a $44 million land discount.

Per job math: $6.6 million in tax subsidies + $6 million in public infrastructure = $12.6 million public cost per permanent job created.

And Louisiana enacted these subsidies without producing an official cost estimate. Legislators rewrote a rural broadband bill in weeks to seal the deal, using NDAs and “parliamentary sleight of hand to avoid public scrutiny” (Good Jobs First). Residents won’t know total lost revenue until Meta applies for rebates in future years.

This is extraction. Meta gets billions in subsidies and free infrastructure. Community gets 500 jobs (by 2034), destroyed roads, strained utilities, and decades of lost tax revenue. Workers nationwide lose jobs while Meta builds “superintelligence” with public subsidies and no worker protection.

Fortune’s observation: “expanding so quickly that it may be difficult for local communities to spot or register concerns in real time.”

Breaking (February 6, 2026, 12:27 PM): Nvidia CEO Jensen Huang told CNBC that demand is “sky high,” “massive AI CapEx is appropriate and necessary,” and the “AI build-out will take 7-8 years.” This is the company selling the GPUs. Profiting directly from every data center deal. Telling communities to commit to nearly a decade of infrastructure spending.

Think about what he’s asking: 7-8 years of buildout for technology that evolves every 6-12 months. By the time these facilities are complete, AI will have evolved through 8-16 generations. Communities will be locked into 25-30 year tax commitments for infrastructure Nvidia’s own CEO says needs 7-8 years to build, meaning it’s obsolete before completion.

Louisiana gives Meta $3.3 billion in tax subsidies for equipment? Nvidia profits. Oracle cuts 20,000-30,000 jobs to fund AI expansion? Nvidia profits. Legal sector loses $285 billion in market value? Nvidia profits from the automation tools.

Meanwhile, Federal Reserve Vice Chair Philip Jefferson said the same day: “Job creation has been weaker than we’d like” and “I don’t want to see any further weakening in the labor market.” He’s confirming Thursday’s data (658,000 fewer job openings than expected) while admitting the Fed is worried about further deterioration.

This is the acceleration gap made explicit: The CEO selling the infrastructure wants 7-8 year commitments. The Fed admits jobs are weakening now. Workers are being displaced today. And communities are signing 25-30 year tax deals for technology that will obsolete itself multiple times before the contracts expire.

What Doesn’t Work: Individual Solutions to Structural Problems

The advice workers got after manufacturing displacement:

“Learn to code” → Field automating itself, entry-level vanishing
“Go into healthcare” → Being automated (announced Jan 11)
“Get a law degree” → Market repriced sector (down $285B this week)
“Retrain for the future” → The future automates faster than retraining cycles

Individual career pivots don’t solve structural automation. Going solo as a consultant or coach? AI will undercut you. You’ll compete with every other displaced worker for scraps. It’s a race to the bottom.

This is like telling people to recycle harder while corporations dump unlimited pollution. Individual action is insufficient for structural problems.

What Does Work: Adaptation Through Creative, Collaborative Solutions

The core principle: Don’t plan for career paths that exist in 2026 and expect them to be there in 2030.

Be flexible. Look creatively for solutions. Understand the current model but don’t depend on it.

Your skills from collapsing careers aren’t worthless! They’re the foundation for transition work. But don’t monetize them individually. Join or build collaborative organizations that create alternatives to extraction.

Use What You Know

Computer science background?

Join worker-owned tech co-ops (like Palante Technology Co-op)
Build collaborative career intelligence platforms serving workers collectively
Work for staffing agencies that prioritize worker welfare (push them toward co-op models)
Contribute to open-source projects building public infrastructure
Join tech worker organizing efforts (Tech Workers Coalition)
Use CS skills as tools for organizing, intelligence work, building alternatives—not as sole career bet

Legal background?

Workers’ rights clinics providing collective legal services
Legal aid organizations (not solo practice)
Policy advocacy groups pushing for EU-style worker protections
Community legal education teaching people to negotiate collectively

Healthcare background?

Community health worker programs (collaborative care models)
Mutual aid health networks
Patient advocacy organizations
Health navigation services run as cooperatives

Business/finance background?

Co-op development consulting (IN teams, not solo)
Community development financial institutions
Worker ownership transition specialists (groups like Project Equity)
Social enterprise incubators

Manufacturing/trades background?

Union organizing (VW Chattanooga workers just won after 2 years)
Worker-owned construction/electrical/plumbing companies
Community organizing around data center battles (many states and communities need this work)
Training others in your trade through collective programs

Agriculture background?

Farm co-ops and community-supported agriculture
Agricultural organizing and advocacy
Local food system development
Regenerative agriculture coordination

Why Collaborative Beats Solo

Solo consulting/coaching:

AI automates this (ChatGPT already gives career advice)
Race to the bottom on pricing. Think customer support post internet. This will be similar, but faster.
No collective bargaining power
Competing with every other displaced worker
Extraction model in disguise

Collaborative organizations:

Harder for AI to replace trust networks and coordination
Collective bargaining power with clients
Shared resources, knowledge, risk
Build political power to advocate for policy change
Create alternative economic infrastructure
Your labor benefits the collective, not just yourself

This Work Needs Doing Right Now

The coming crisis creates the demand:

Communities need intelligence to negotiate with data center developers (what PivotIntel does. Helping townships evaluate infrastructure proposals)

Workers need career intelligence to navigate transitions (what Under the Radar does. Tracking automation patterns, identifying opportunities)

Organizers need training to build political pressure for policy change (communities blocking data centers, policy advocates pushing for worker protections)

Co-ops need development support to provide worker ownership alternatives (transitioning businesses from extraction to worker ownership)

Policy advocates need research showing what works (this week’s comparative article proved alternatives exist. Singapore funds retraining with $300-3,000/month, EU requires worker notification and oversight, even China coordinates employment monitoring)

Displaced workers need navigation help (career coaches, benefits specialists, transition coordinators working in teams, not solo)

This work is growing because extraction is accelerating. The worse it gets, the more communities need coordination. The more workers get displaced, the more they need transition support. The faster automation spreads, the more urgently we need alternative economic models.

Real Examples: This Already Works

The Open Record L3C: Not solo journalism. Organizational structure serving multiple audiences (Under the Radar for workers, PivotIntel for communities, investigative articles). Building intelligence infrastructure that scales beyond individual capacity. Social mission + sustainable revenue.

Worker-owned cooperatives: Palante Technology Co-op, Arizmendi Bakery, Cooperative Home Care Associates (1,800+ workers in the Bronx). These create ownership, not just employment.

Community organizations: Townships organizing against data centers. Policy advocates who got EU AI Act passed. Singapore’s systematic retraining (didn’t happen by accident. It required political pressure). VW Chattanooga workers winning union recognition after 2-year fight.

Staffing agencies that actually serve workers: Not perfect, but they understand their business depends on worker success, not just filling corporate slots.

The Bottom Line: Adaptation Is The Way

Be flexible. Look creatively for solutions. Don’t plan on career paths from 2026 existing in 2030.

Check out our interactive graphic “The Acceleration Gap” to map this unprecedented shift.

Thursday’s jobs data (658,000 fewer openings than expected) confirms workers have nowhere left to turn within the extraction economy. Legal market confidence crashed this week. Healthcare automation announced last month. Meta builds $50 billion “superintelligence” facilities with $12.6 million in public subsidies per permanent job created.

Individual retraining is insufficient. Going solo means competing with AI for scraps.

What works: Adaptation through creative, flexible, collaborative solutions.

Use your expertise to build organizations that create alternatives. Worker-owned co-ops. Community intelligence platforms. Policy advocacy groups. Mutual aid networks. Organizations that pool resources, build collective bargaining power, and create political leverage to demand systemic change.

This week I published a comparative analysis of seven countries’ AI employment policies. Singapore funds worker transitions with $300-3,000/month allowances and 90-95% course fee coverage. The EU requires worker notification before AI deployment, mandates human oversight, and imposes €35 million penalties for violations. Even China coordinates employment monitoring with early warning systems.

The United States chose extraction without protection. But alternatives exist and function. Other countries prove systematic adaptation is possible.

Workers can’t individually retrain fast enough. But they can organize collectively for policy change AND build alternative economic structures NOW. That organizing work needs doing. That intelligence work needs doing. That community coordination needs doing. That co-op development needs doing.

These are real careers serving real needs. And they build the alternative economic system rather than just surviving within extraction.

Your skills matter. Use them flexibly. Apply them collaboratively. Adapt creatively.

Don’t optimize your career for the economy that’s collapsing. Build your career around creating the economy we need—and understand that “career” itself might need redefining as we go.

Next Steps

Assess your expertise honestly: What field are you trained in? What do you actually know? What are you good at?
Be flexible about applying it: Your CS degree doesn’t have to mean “software engineer.” Your legal training doesn’t have to mean “lawyer.” Your healthcare experience doesn’t have to mean “nurse.” Think creatively about how those skills transfer.
Find collaborative organizations: Worker co-ops, community organizations, policy advocacy groups, mutual aid networks. If they don’t exist locally, connect with others to build them.
Don’t go solo: Resist the temptation to just “hang out your shingle.” Collaborative structures provide mutual support, shared resources, and collective power that solo work can’t.
Stay informed and adaptive: Read the comparative analysis to see proof that alternatives exist. Follow Under the Radar for weekly intelligence on automation patterns. Follow PivotIntel for community intelligence on infrastructure battles. Adapt as conditions change.
Understand but don’t depend on current models: Learn how things work now. Don’t plan on them working that way in 5 years.

Thursday’s numbers show displacement accelerating with nowhere to turn. But that creates urgent demand for the work of building alternatives.

Adaptation is the way. Be flexible. Look creatively. Build collaboratively.

Sources:

Bureau of Labor Statistics (BLS) JOLTs Report, January 31, 2026
Initial & Continuing Jobless Claims, February 5, 2026
Bloomberg, TechLoy, TradingView, “Legal software stocks,” February 3-6, 2026
Fortune, “Meta’s Hyperion AI data center,” February 4, 2026
Anthropic, “Advancing Claude in healthcare and the life sciences,” January 11, 2026
Good Jobs First, “States Are Opening a Pandora’s Box of Data Centers,” November 2025
Sherwood News, “The power play behind Hyperion,” December 15, 2025
Under the Radar, “Technology Doesn’t Force Displacement: Seven Countries Show Policy Choices Determine Who Pays,” February 4, 2026

Under the Radar publishes every Friday at 8am EST. Career intelligence for workers navigating AI transformation.

The Open Record L3C • theopenrecord.org PivotIntel Weekly • pivotintel.org