Meta Layoffs 2026: 8,000 Cut to Fund $72B AI Push
ALSO: Chrome AI Skills, GPT-5.5 wins, AI compute crunch
Welcome to Guru99 AI Report!
Top Story: AI is evolving at an unprecedented pace, and the impact is real. This week, we explore the latest developments, from Meta’s workforce changes to Google’s new AI capabilities. Don’t miss out on these game-changing shifts.
💼 Meta cuts 8,000 jobs to ramp up its AI investments
Brief Buzz:
Meta is slashing 10% of its workforce — roughly 8,000 jobs — in what Bloomberg framed as a push for efficiency and CNN called an AI-driven reorg. The cuts hit hardest inside Meta Superintelligence Labs, even reaching some hires Zuckerberg personally poached.
- Layoffs help fund Meta’s massive $72B AI capex commitment for 2026, up from $47B in 2025
- Cuts target middle managers, non-agentic researchers, and some transfers from the $14.3B Scale AI deal
- Superintelligence Labs survives but narrows focus to ship-ready “Mythos-tier” projects
- Arrives as Anthropic’s Dario Amodei warns of broader white-collar cuts across the economy
- SemiAnalysis’ Dylan Patel (on Patrick O’Shaughnessy’s podcast) describes one analyst spending $6,000/day on Claude tokens to rebuild what a 100-person team took a decade to make
💡 Why Should You Care?
When one person with AI tools can do the work of 10-to-15 people, the most vulnerable aren’t the AI users — they’re those whose roles AI now handles alone: ticket-shufflers, middle managers, specialists in solved problems. The lesson is blunt: leverage AI, or risk being leveraged out.
📰 Chrome’s AI “Skills” hint at a future where the web works for you
Brief Buzz:
Google just rolled out Skills in Chrome, a new feature that lets users save and reuse their favorite AI prompts with a quick shortcut. Instead of retyping the same instructions every time, you can now summon them in seconds, marking another step in Chrome’s growing AI integration.
- Users can save custom AI prompts and trigger them instantly using “/” or “+” shortcuts in Gemini in Chrome
- Google is launching a ready-to-use Skills library with templates that can be customized, useful for repetitive tasks like comparing product specs or calculating recipe macros
- Skills run on Chrome’s existing security and privacy safeguards and will request confirmation before performing sensitive actions
- The feature is rolling out to signed-in Chrome desktop users and joins similar offerings from Anthropic’s Claude, ChatGPT, and Gemini
💡 Why Should You Care?
Skills could quietly reshape how you interact with the web. Rather than juggling separate AI tools or retyping prompts, your browser becomes a smarter assistant that remembers your routines, whether you’re shopping, meal planning, or researching. It also signals a bigger shift: AI is moving away from standalone chatbots and embedding itself into tools you already use, lowering the barrier for everyday users who’d otherwise skip AI altogether.
🚀 Say Goodbye to Online Shopping — Anthropic’s AI Just Took Over!
Brief Buzz:
Anthropic just ran a fascinating week-long experiment called Project Deal, where 69 employees handed over their classifieds-style trading to Claude AI agents. The result? A real marketplace where AI bots haggled, bargained, and closed deals on actual personal belongings — with some eye-opening findings.
- AI agents struck 186 deals worth over $4,000, trading everything from snowboards to ping-pong balls — all without humans approving anything mid-negotiation.
- Smarter models won bigger: Claude Opus 4.5 sellers earned $2.68 more per item, and Opus buyers paid $2.45 less than people stuck with the weaker Claude Haiku 4.5.
- Users represented by weaker agents didn’t notice they were getting worse deals, rating fairness nearly identically across both models.
- Aggressive negotiation prompts barely moved the needle — model quality mattered far more than strategy, building on Anthropic’s earlier Project Vend research.
- 46% of participants said they’d actually pay for an AI shopping agent like this in real life.
💡 Why Should You Care?
Agent-to-agent commerce is closer than you think. If your future AI shopper is weaker than the one negotiating against you, you’ll likely lose money — and never realize it. As businesses race to deploy these tools, the gap between premium and budget AI agents could quietly widen economic inequality in ways policy hasn’t begun to address.
✅ How to Ace Behavioral Interviews When AI is in the Room
Brief Buzz:
AI-powered cheating tools are quietly reshaping technical interviews, feeding candidates real-time answers that are nearly impossible to detect. As a result, coding rounds are losing their edge, and hiring decisions are shifting in an unexpected direction.
- Invisible AI tools now assist candidates live, bypassing traditional screening methods (State of cheating in interviews)
- Technical interviews are no longer the ultimate filter for hiring
- Behavioral rounds — how you tell your story — are becoming decisive
- Many candidates fail not on coding, but on poor communication of experience
- Experts highlight a growing gap: strong coders, weak storytellers
💡 Why Should You Care?
This shift means technical skills alone won’t land you the job anymore. As AI evens the playing field, your ability to clearly explain your decisions and experiences becomes your real advantage. Communication — not just competence — is now what gets you hired.
🚀 It’s Here! GPT-5.5 Just Crushed Every AI Benchmark
Brief Buzz:
OpenAI just dropped GPT-5.5 (codenamed ‘Spud’), a “worker-class” model built to complete tasks, not just answer questions. Landing exactly seven days after Anthropic’s Opus 4.7, it tops industry benchmarks and powers OpenAI’s much-anticipated agentic superapp vision.
- Beats Opus 4.7 on Terminal-Bench 2.0 (82.7% vs 69.4%) and matches or exceeds human professionals on 84.9% of real-world tasks across 44 occupations
- Built using recursive self-improvement — GPT-5.5 and Codex helped rewrite OpenAI’s own GPU infrastructure code
- Priced at $5/$30 per million input/output tokens, rolling out now in ChatGPT and Codex for Plus, Pro, Business, and Enterprise users
- Excels at long-horizon agentic workflows and even helped discover a novel proof about Ramsey numbers, verified in Lean
- Ships with zero impact on latency despite the upgrades, plus stricter safeguards after being rated “High” on cyber and bio risks
💡 Why Should You Care?
This is the first clean “GPT beats Claude” moment in over a year — and the seven-day turnaround signals a brutal new launch cadence. If you rely on AI agents for real work, expect smarter task automation, sharper competition, and rapidly falling prices as both labs trade blows weekly.
🖥️ AI’s Compute Crunch: When Demand Outpaces Supply
Brief Buzz:
The AI conversation has flipped from “is this a bubble?” to “can we keep up?” A perfect storm of geopolitical conflict, runaway agent usage, and hardware shortages is creating a serious compute crisis — just as frontier labs prepare to roll out their most powerful models yet.
- War in Iran is disrupting supply chains, with helium shortages threatening semiconductor production and rising oil prices squeezing logistics across the tech ecosystem
- AI agents are devouring tokens at staggering rates — per the Wall Street Journal, OpenAI’s token demand jumped from 6 million per minute in October 2025 to 15 billion per minute in March, with OpenAI’s Peter Steinberger noting much of it is wasted on inefficient agent calls
- GPU, memory, and storage are scarce, with the Ornn Compute Price Index showing Nvidia GPU rental costs up 48% in recent months
- Anthropic’s new Mythos model is launching to a small group of partners, officially over cybersecurity concerns — though compute constraints likely play a major role
- OpenAI’s upcoming Spud model is being teed up with a policy paper inviting public debate on jobs and societal impact
💡 Why Should You Care?
If you rely on AI tools, expect slower responses, higher prices, and tighter access as labs ration capacity. Smaller startups risk getting priced out while giants hoard chips, concentrating AI power in fewer hands. The cutting-edge models you’ve been waiting for may stay locked behind partner-only walls for a while.
📸 AI Generator Images: Unfolding Paths
Midjourney Prompt
Vertical magazine advertisement. A pure white fabric is gently unfolded, and beneath it appears a road map with highways, small towns and green forest areas, realistic and elegant. Two hands gently unfolding the fabric. The fabric merges seamlessly into the map, forming a graceful road. On the road: a realistic white Yutong coach bus and a few vivid realistic blue cars, life-size proportions, integrated naturally into the scene, not toy-like. Bright cinematic daylight, premium editorial photography aesthetic, minimalistic yet artistic composition. –ar 2:3 –style raw –v7
Hey! I’m Krishna Rungta
Founder of Guru99.com, Editor-in-chief & Technology Expert
Was this email forwarded to you? Sign up for free here.
