Unicorn glasses, robot SPAC, and a $86 churn trap

Even Realities, a Shenzhen startup founded by ex‑Apple engineers, closed a $150 million pre‑Series B led by Meituan and Tencent, valuing it at $1 billion. The funding backs its privacy‑first, display‑only glasses that skip cameras and rely on proprietary waveguide optics. The round signals investors’ confidence in a camera‑free approach amid a crowded smart‑glasses race.

Agility Robotics is merging with Churchill Capital Corp XI in a $2.5 billion SPAC, targeting a public valuation that could fund production of up to 10,000 Digit units annually. CEO Mark Schwartz warns the company will stay focused on enterprise customers, pushing consumer‑grade robots further into the future.

Alessio Fanelli shows how to hook OpenAI Symphony to Linear, turning a Linear board into a state machine that spawns autonomous coding agents. The setup runs entirely from his phone, and even scouts underpriced Pokémon cards on eBay, proving that AI can manage both software builds and small‑business tasks without manual supervision.

A study of 36 IBM Telco churn analyses shows the common 0.5 classification cut‑off wastes about $86 for every subscriber, or $8.6 million on a 100 k customer base. Missing a churner is roughly 13 × costlier than over‑treating a loyal user, meaning the default threshold is a costly pricing mis‑decision.

Paul Graham argues that raw intelligence is only a baseline for founders. What truly separates the exceptional are curiosity, relentless energy, independent thinking, and an unmeasurable drive. Cultivating these traits, not just IQ, predicts breakthrough ideas and lasting impact.

A new Open Future Forum survey of 421 finance‑room leaders finds that 53% cite proving ROI as the main barrier to more AI investment, while 62% expect measurable returns within six months. One in six already funds AI from headcount budgets, signaling a shift from experimentation to short‑term value extraction.

Xbox will lay off about 3,200 staff and spin off four studios as it pivots away from an over‑extended hardware‑heavy model. The reset shifts focus to Game Pass, open development tools, and a leaner content portfolio, acknowledging margins 3‑10× below peers. The move reshapes the console market and signals a broader industry hardware crisis.

A startup’s hand‑held RxScanner struggled when its cloud AI was 14,000 km away, taking minutes per scan. By shrinking the model to run on an Android phone, the device now authenticates pills instantly without broadband, sparking a shift toward small, edge‑deployed AI in pharma and other sectors where networks are unreliable.

Tech giants are slashing jobs citing AI, even as revenues surge. Microsoft cut 4,800 roles, Oracle 21,000, GitLab 350, and Google trims managers, pushing total 2026 tech layoffs to ~120,000. The pattern shows AI as both growth engine and cost‑cutting justification, reshaping workforce strategies.

Reddit announced upgraded AI defenses that deploy large language models to detect coordinated fake behavior, blocking about 23 million spam views each day and catching roughly 25 000 new spam posts. The rollout has already lowered user spam exposure by 20% and revoked nearly 2 million inauthentic votes daily.

Anthropic’s annualized run‑rate jumped from $9 billion at end‑2025 to $47 billion by mid‑May 2026. At that pace it will eclipse the revenue of every public software company, including Salesforce and Adobe, by year‑end, leaving only Microsoft ahead. The surge redefines how fast AI startups can dominate the software landscape.

Effective July 30 2026 Amazon will stop taking new Mechanical Turk customers, keeping the service alive only for existing requesters and workers. AWS says it will maintain security and availability but will not add features, a sign that low‑cost human micro‑task markets are losing relevance as AI handles more work.

Rauch points out that half of Vercel’s 6 million daily deployments are AI agents, but bundling models and agents creates security and cost issues. Vercel’s new Eve framework and Sandbox let developers keep models separate and enforce data‑access policies, promising tighter control and cheaper, faster production AI workloads.