- Latestly AI
- Posts
- 25% of White-Collar Jobs Are Already Obsolete, Are You?
25% of White-Collar Jobs Are Already Obsolete, Are You?
$30B funding rounds, federal AI power grabs, and 500MW compute factories: the operators who redesign workflows now will control what’s left.
Top 3 Things in Today’s Latestly AI Edition
1. Anthropic raises $30B at $380B valuation - ARR hits $14B
2. Trump signs executive order to override state AI laws
3. Saudi Arabia + NVIDIA announce 500MW AI factories
AI Foundations Launches February 20
Cohort 1 officially launches on February 20 at 6:00 PM UK time.
The full program is priced at £299, and includes:
Complete AI Foundations 8 week course access with hands-on tasks
One full year of Premium guides and research articles
Ongoing updates as the AI landscape evolves
Payment opens on February 20, 2026 at 6PM UK.
The next edition will include the official enrolment link to begin the course.
Set a reminder now so you don’t miss the launch 👇👇👇
AI STORY OF THE WEEK
How To Make AI Your Strategic Partner?
“Even if the technology doesn’t improve from today, it could wipe out 25% of entry-level white-collar jobs.”
Connor Grennan (Chief AI Architect, NYU Stern) isn’t predicting AGI. He’s pointing at workflow compression. His thesis: we’re misclassifying LLMs.
Most teams treat ChatGPT, Gemini, or Claude like upgraded search engines -deterministic query systems. But LLMs are probabilistic synthesis engines trained to model patterns across massive corpora. Their advantage isn’t perfect answers. It’s scalable transformation of messy inputs into structured outputs repeatedly. That makes them infrastructure for process redesign.
What the 25% Actually Means (Technically)
Entry-level white-collar work clusters around four functions:
Information Compression - summarizing research, synthesizing notes
First-Pass Generation - drafts, slide outlines, reports
Pattern Translation - turning data into narrative
Formatting & Structuring - decks, memos, documentation
LLMs are already strong at task decomposition (breaking ambiguous prompts into sub-steps), iterative refinement across long context windows, perspective simulation (stakeholder modeling, objection mapping), and rapid outline generation.
If 60% of a junior team’s time goes into producing first-pass artifacts, and AI reduces that by half or more, team structure changes. Not because AI understands quality, it doesn’t but because fewer humans are required to supervise higher volumes of AI-assisted output. The bottleneck moves from production to judgment.
The Real Constraint: Cognitive Framing
Grennan highlights something subtle: chat interfaces look like search engines. Our neural pathways default to command → response → exit.
Power users don’t operate that way. They maintain persistent context, run 5–10 iteration cycles, ask for critique and counterarguments, and treat the model as a collaborative reasoning layer. That behavioral delta creates asymmetry inside organizations.
How Operators Should Implement This
Externalize your workflow. Map every step of how work actually gets done including invisible cognitive steps.
Insert AI at transformation layers: option expansion, draft generation, risk modeling, argument stress-testing, process documentation.
Redesign, don’t just accelerate. Speed is linear. Process redesign that scales across a team is multiplicative.
Demonstrate system-level thinking. In interviews or performance reviews, don’t say “I use AI.” Show how you rebuilt a workflow and made it repeatable.
The takeaway isn’t hype. Even if models plateau, their current capabilities are enough to compress entry-level output layers. The shift won’t be driven by model releases. It’ll be driven by operators who understand how to embed probabilistic systems into deterministic business processes and restructure teams around that reality.
GLOBAL AI NEWS HIGHLIGHTS
The Most Important AI Developments Happening Now : Region by Region
Americas
Trump signs executive order preempting state AI laws; directs Attorney General to challenge regulations seen as harming US competitiveness
Google DeepMind launches Gemini Deep Think for research-grade math/science; partners with US DOE on Genesis Mission AI infrastructure Google DeepMind / BNL, Feb 11–12, 2026
Boston Dynamics integrates Google DeepMind foundation models into Atlas humanoid for industrial and human-robot interaction use TechCrunch, Jan 5, 2026
Europe
European Commission issues antitrust objections to Meta over restricting third-party AI assistants on WhatsApp; proposes interim measures European Commission, Feb 9, 2026
Mistral AI commits €1.2B to Swedish data centers, first expansion outside France, opening 2027 CNBC / Bloomberg, Feb 11, 2026
Germany launches Munich AI Factory with NVIDIA, creating one of Europe’s largest AI compute hubs Euronews, Feb 5, 2026
Africa
Nairobi AI Forum allocates 1.5M GPU hours to 130 innovators; launches AI 10B Initiative with AfDB targeting 45M jobs by 2035 Africa Science News / UNDP, Feb 9–10, 2026
Google expands Gemini and NotebookLM access across 7 African countries for education and skills programs Google Africa Blog, Feb 13, 2026
70% of South African adults used AI chatbots, +25pp vs 2023, above global average Africa.com, Feb 12, 2026
Middle East
Saudi Arabia and NVIDIA partner on AI factories up to 500MW; first phase includes 18,000 GB300 Grace Blackwell supercomputers NVIDIA, Feb 3, 2026
80% of GCC organizations include AI in strategy; fewer than one-third have governance or operating models Consultancy-me, Feb 11, 2026
APAC
Zhipu AI releases GLM-5 topping open benchmarks; raises GLM Coding Plan pricing 30% as shares jump 34% Bloomberg, Feb 11, 2026
OpenAI tells US lawmakers DeepSeek may be distilling US frontier models, per memo to Congress Reuters, Feb 13, 2026
ByteDance, Alibaba, DeepSeek preparing upgraded AI model launches ahead of Lunar New Year (Feb 15) Reuters / Rappler, Feb 12, 2026
Oceania
Australian IT spending to hit A$172.3B in 2026 (+8.9% YoY); datacenter growth 22.5% driven by AI/cloud Computer Weekly, Feb 2026
NZ research warns on AI autonomy in emergency decisions; calls for preserving human judgment frameworks University of Auckland / Newsroom, Feb 12, 2026
This Week’s Top 5 AI Startup Fundraising Rounds
Company | Round Size & Date | Focus |
Anthropic | $30B – Feb 12, 2026 | Foundation models for enterprise AI. Claude AI platform reaching $14B annualized revenue with focus on safety and enterprise adoption. |
Cerebras | $1B – Feb 3, 2026 | AI infrastructure and wafer-scale chips. Developing Wafer Scale Engine 3 with 4 trillion transistors to power AI training and inference 20x faster than competing GPUs. |
Apptronik | $520M – Feb 11, 2026 | Humanoid robotics for industrial automation. Apollo humanoid robots deployed in manufacturing and logistics for Mercedes-Benz and GXO, powered by Google DeepMind's Gemini Robotics. |
Goodfire | $150M – Feb 5, 2026 | AI model interpretability and safety. Platform to understand, debug, and modify neural network internals; reduced hallucinations by 50% in large language models. |
Resolve AI | $125M – Feb 4, 2026 | AI-powered site reliability engineering. Autonomous agents for production operations that investigate incidents, identify root causes, and execute remediations across cloud infrastructure. |
Top 10 AI Unicorns - Latest Movements
Company | Latest Valuation | Latest Major Update |
OpenAI | $500B (2025) | ARR reached $20B by late 2025; 800M weekly active users. Exploring $100B funding at $730-830B valuation; confidential IPO filing expected for H2 2026 listing. |
Anthropic | $380B (2026) | Raised $30B Series G on Feb 12, 2026. ARR climbed to $14B; Claude Code hitting $2.5B ARR. Enterprise users represent over half of revenue. |
xAI | $230B (2026) | Raised $20B Series E in Jan 2026; 600M+ monthly active users across X and Grok platforms. Grok 4 launched; revenue projected to exceed $2B in 2026. Merged with SpaceX |
Databricks | $134B (2025) | Raised $5B in Dec 2025; crossed $5.4B ARR with 55%+ YoY growth. Filed confidentially for Q1 2026 IPO. |
Cerebras | $23B (2026) | Raised $1B Series H on Feb 3, 2026; signed $10B+ multi-year deal with OpenAI for compute capacity. Preparing for Q2 2026 public listing. |
Figure AI | $39B (2025) | Raised $1B+ Series C in Sept 2025; Figure 02 humanoid robots deployed at BMW plant. Figure 03 entering household applications testing. |
Safe Superintelligence | $32B (2025) | Raised $2B in April 2025 led by Greenoaks; co-founded by ex-OpenAI chief scientist Ilya Sutskever. No public product yet; focused on safe AGI research. |
Anysphere (Cursor) | $29.3B (2025) | Raised $2.3B Series D in Nov 2025; ARR exceeded $1B. AI coding assistant used by OpenAI, Uber, Spotify with 1M+ daily active developers. |
Scale AI | $29B (2025) | Meta investment valued company at $29B in 2025. Provides training data and model evaluation for frontier AI models; interim CEO named after founder transition. |
Perplexity AI | $20B (2025) | Raised $200M in Sept 2025; ARR reached $148M with 780M monthly queries. Powers Snapchat's AI search; Chrome acquisition discussions active. |
How was this edition? |
We hope you enjoyed this Latestly AI edition.
📧 Got an AI tool for us to review or do you want to collaborate?
Send us a message and let us know!
Was this edition forwarded to you? Sign up here

