AI Week in Review: May 2026 — Google I/O, Meta's $125B Bet, Nvidia's Record Quarter
From Gemini 3.5 Flash to 8,000 Meta layoffs and a $58.3B Nvidia quarter — the third week of May delivered more AI news than most quarters used to.
Compiled from live news data by NewzAI · May 21, 2026
The AI Boom Is No Longer a Prediction
Nvidia just posted the numbers that make the argument for you. The chipmaker reported $81.6 billion in quarterly revenue and $58.3 billion in profit — a 211% jump year-on-year — cementing its place as the world's most valuable company. CEO Jensen Huang's early bet on GPU-based AI compute has become the infrastructure backbone of an entire industry. Read on NewzAI →
That was just one moment in a week that saw Google remake its search engine, Meta cut 8,000 jobs to fund a $125 billion AI push, Andrej Karpathy jump ship to Anthropic, and the Catholic Church issue its first major statement on AI. Here is everything that happened.
Google I/O 2026: The Biggest AI Blitz in Google's History
Image credit: NDTV
Google's annual developer conference on May 19–20 was, by any measure, the company's most ambitious AI showcase to date. CEO Sundar Pichai unveiled Gemini 3.5 Flash, which Google claims outperforms Gemini 3.1 Pro on most benchmarks and generates 289 output tokens per second — nearly four times faster than OpenAI's GPT-5.5. Read on NewzAI →
The week's biggest consumer product was Gemini Spark, a personal AI agent that runs around the clock across Gmail, Drive, and other Workspace apps — capable of planning events, drafting emails in your writing style, and continuing tasks in the background after you close your laptop. It will roll out in beta for Google AI Ultra subscribers in the US next week. Google also introduced Docs Live, a voice-driven document creation tool powered by Gemini, and Google Antigravity, an AI coding agent already in use by internal teams.
Search itself got its first major interface redesign in 25 years. The new experience accepts images, files, and Chrome tabs directly in the search box, while AI agents can monitor the web continuously and send users automatic updates on chosen topics. A new Universal Cart feature tracks product prices across the internet and surfaces deals automatically.
On hardware, Google unveiled TPU 8t (training) and TPU 8i (inference) chips: 3× the compute of the previous generation, with inference speeds up to 1,500 tokens per second. To deploy them at scale, Google announced a joint AI cloud venture with Blackstone, backed by an initial $5 billion in equity, with total investment potentially reaching $25 billion, targeting 500 megawatts of data-centre capacity by 2027. Read on NewzAI →
And after more than a decade, smart glasses are back. Google revealed Gemini-powered eyewear in two designs — one by Warby Parker, one by Gentle Monster — going on sale in autumn 2026. The glasses are audio-first: Gemini speaks into your ear privately. A version with an in-lens display is in development but not yet ready. Read on NewzAI →
Meta Cuts 8,000 Jobs to Fund a $125 Billion AI Push
Image credit: The Hindu / AFP
On May 20, Meta began its largest company-wide round of job cuts since the 2022–23 "Year of Efficiency" — laying off approximately 8,000 employees, roughly 10% of its global workforce. Termination notices went out in the early morning hours, with Singapore-based workers among the first to be informed.
In a memo posted by Business Insider, CEO Mark Zuckerberg acknowledged the toll: "It's always sad to say good-bye to people who have contributed to our mission and to building this company. I feel the weight of that." He added that Meta does not expect additional company-wide layoffs this year and acknowledged the company had fallen short in communicating with staff.
The restructuring is paired with a decision to move 7,000 employees into AI-related roles and eliminate 6,000 open headcount positions. Meta has forecast capital expenditure of $125–145 billion for 2026 — more than double last year's outlay — with the stated goal of delivering what Zuckerberg calls "personal superintelligence" to users. The cuts follow reports that Meta had been tracking employee mouse movements and keystrokes to generate AI training data, prompting internal protests. Read on NewzAI →
The Chip Race: Nvidia's Dominance and China's Challenge
Nvidia's Q1 2026 results — $81.6B in revenue, profit up 211% year-on-year — are a direct read-out of how much every major tech company is spending on AI infrastructure. Analysts called it "record revenue as AI chip demand surges."
But challengers are emerging. Alibaba unveiled the Zhenwu M890 chip this week: 144GB of GPU memory, designed specifically for agentic AI workloads. It is China's most direct answer yet to Nvidia's dominance, as Beijing accelerates domestic semiconductor development under the pressure of US export controls. Read on NewzAI →
Meanwhile, Dell launched an on-premises agentic AI system built on Nvidia software, claiming enterprises can break even versus cloud costs within three months — a direct pitch to pull workloads away from cloud providers.
Anthropic's Breakout Week
The biggest personnel move of the week: Andrej Karpathy, co-founder of OpenAI and one of the field's most respected researchers, announced he is joining Anthropic's pretraining team. He will build a research team focused on using Claude to accelerate pretraining — a significant signal about where top AI research talent is moving. Read on NewzAI →
Anthropic also expanded Project Glasswing, its controlled AI cybersecurity program, allowing partner organisations to share AI-discovered vulnerability findings with regulators, industry bodies, and the public under responsible-disclosure norms. Read on NewzAI →
In enterprise adoption, Bristol-Myers Squibb announced it is deploying Claude to 30,000 employees across drug research, development, manufacturing, and commercial operations — one of the largest pharma–AI rollouts of the year. Read on NewzAI →
AI and Society: Courts, Unions, Finance, and the Church
A California jury on May 18 unanimously dismissed Elon Musk's lawsuit against OpenAI, ruling the case time-barred under the statute of limitations. The verdict removes a significant obstacle from OpenAI's path toward a public offering. Musk, who had accused Sam Altman of "stealing a charity" by shifting OpenAI from non-profit to for-profit, says he will appeal. Read on NewzAI →
In London, Google DeepMind agreed this week to enter formal talks at Acas — the UK's conciliation service — after several hundred workers at its London headquarters voted to unionise. The workers, represented by the Communications Workers Union and Unite, have raised concerns about DeepMind's AI being used by US and Israeli government defence and intelligence operations. It is the first step toward potential collective bargaining at one of AI's most prominent research labs. Read on NewzAI →
In banking, Standard Chartered announced plans to cut 7,800 back-office roles — 15% of that workforce — by 2030, as AI takes over processes at its operations in India, China, Malaysia, and Poland. CEO Bill Winters described the move as replacing "lower-value human capital" rather than cost-cutting. DBS, HSBC, and Meta have now all made similar announcements in the span of weeks. Read on NewzAI →
And in an unlikely pairing of institutions, the Vatican announced that Pope Leo's first major papal text will be titled "Magnifica Humanitas" (Magnificent Humanity) — an encyclical addressing AI's impact on human dignity, workers' rights, and warfare. Signed on May 15, it will be presented at the Vatican on May 25. Among the attendees: Christopher Olah, co-founder of Anthropic. The Vatican drew the parallel explicitly: Leo XIII responded to the Industrial Revolution in 1891; Leo is responding to the AI revolution now. Read on NewzAI →
What to Watch Next Week
May 25: Pope Leo presents "Magnifica Humanitas" at the Vatican, with Anthropic's Christopher Olah in attendance.
Gemini Spark beta rolls out to Google AI Ultra subscribers in the US — the first real test of whether a 24/7 personal AI agent gains traction with everyday users.
OpenAI IPO timeline — with the Musk lawsuit cleared, watch for formal S-1 or prospectus signals from the company.
Alibaba Zhenwu M890 benchmark results — if independent tests hold up, it would represent the most credible Nvidia alternative to emerge from China yet.
Follow This Story on NewzAI
NewzAI tracks breaking news in real time — summarised from multiple sources so you get the full picture, not just a headline.
Google I/O 2026: Gemini 3.5 Flash, Spark & Smart Glasses →
Meta Cuts 8,000 Jobs in AI Pivot →
Nvidia's $58.3B Profit — the AI Boom in Numbers →