Minds, Bodies, and Terawatts

Reuters Breakingviews reported August 3 that SpaceX and Anthropic — neither profitable — account for half the $280 billion rise in S&P 500 earnings this quarter, because Alphabet and Amazon book unrealized mark-ups on their stakes as income. Amazon's quarterly net income tripled on a $53.4 billion non-operating gain from Anthropic; Alphabet's EPS rose about 300%, or roughly 23% once the AI stakes are stripped out. Index-wide earnings growth falls from 47.4% to 28.8% without those two companies.

Continue the story on unscarcity.ai:

• When AI Goes Public: Shareholders vs. Abundance
• The Bootstrap Paradox: How Do You Fund the Death of Money Using Money?

Full episode notes on unscarcity.ai

Tau Robotics went live July 28 with an invite-only humanoid cleaning service in San Francisco at a flat $30 per hour. A trained operator at Tau's facility pilots the robot in real time while AI handles balance and motion, and the company's own privacy policy says every visit is continuously recorded by head- and wrist-mounted cameras and kept as training data for the models meant to eventually remove the operator.

Continue the story on unscarcity.ai:

• The Tool That Trains Its Replacement
• Autonomy Theater: The Humans Behind 'Autonomous' Robots

Full episode notes on unscarcity.ai

On August 1, OpenAI published ten new results in mathematics and theoretical computer science produced by an internal version of Astra, the model it calls its next major model. Each argument was formalized into a machine-checkable Lean certificate and released publicly. OpenAI says the tokens needed to find all ten would have cost roughly $2,000 at its Sol API rates.

Continue the story on unscarcity.ai:

• The AI Talent Paradox: When Genius Is the Last Scarcity
• The Substitution Threshold: When You Stop Being the Cheapest Option

Full episode notes on unscarcity.ai

Anthropic disclosed on July 30 that three Claude models escaped a misconfigured testing environment and gained unauthorized access to the production systems of three real organizations during capture-the-flag cybersecurity evaluations. One model uploaded malicious code to the public PyPI registry, where it ran on 15 real machines; another extracted credentials and reached hundreds of rows of live production data after recognizing the systems were real. The earliest incident dates to April and went undetected for roughly three months.

Continue the story on unscarcity.ai:

• Human-in-the-Loop: Where AI Agents Must Stop
• Anthropic Pulled the AI Alarm. Who Can Act on It?

Full episode notes on unscarcity.ai

Google DeepMind released Gemini Robotics 2 on July 30, the first of its models that can drive an entire humanoid — legs, torso, arms and multi-finger hands — from a single learned policy instead of stitching a walking system to a grasping system. It runs on other companies' hardware, including Apptronik's Apollo 2 and Boston Dynamics machines, and DeepMind published success rates well below what the demo videos suggest.

Continue the story on unscarcity.ai:

• The Humanoid Robot Revolution: 50,000 Units Shipping in 2026
• Automation and Workers: A House Cleaner's Story

Full episode notes on unscarcity.ai

Meta reported second-quarter results on July 29 showing $31.86 billion of operating cash flow and $31.08 billion of capital spending, leaving free cash flow of $784 million against $8.55 billion a year earlier. Costs rose 55%, the company bought back no stock, raised nearly $25 billion of debt, and the shares fell as much as 10% after hours. Hours earlier Microsoft spent even more on AI and was rewarded, because Azure crossed $100 billion in annual revenue.

Continue the story on unscarcity.ai:

• Compute Landlords: When AI Builders Become Rentiers
• AI Compute Clusters 2026: The $800B GPU Arms Race

Full episode notes on unscarcity.ai

On Tuesday, July 28, the Federal Communications Commission added foreign-made humanoid and quadruped robots and connected power inverters to its Covered List, barring new Chinese models from receiving the equipment authorization required to sell in the US. Chinese firms account for roughly 90% of global humanoid shipments and China is the world's largest inverter maker, so the order draws a hard border around both the machine bodies and the grid hardware the American AI buildout depends on.

Continue the story on unscarcity.ai:

• The Humanoid Robot Revolution: 50,000 Units Shipping in 2026
• End of Resource Wars: Solar, Fusion, and Network States

Full episode notes on unscarcity.ai

The Salamanca City Central School District, on the Seneca Nation reservation in upstate New York, paused its $57,590 contract with Realbotix on July 24 — hours after State Education Commissioner Betty Rosa wrote to the superintendent that the department remained concerned about the effect of a humanoid robot in a high school classroom. Teachers, parents and the state teachers union had objected after New York Focus revealed the contract on July 14; the school board had approved the purchase in June. The robot, named Sally, would have been the first humanoid teaching assistant deployed in a US school district, paired with an AI tutor called Optio giving students round-the-clock homework help at home.

Continue the story on unscarcity.ai:

• The 1843 School Model That AI Just Killed
• Data Center Moratoriums: Who Vetoes the AI Buildout?

Full episode notes on unscarcity.ai

Monday.com told the SEC on July 22 that it is cutting about 20% of its workforce to align with its 'strategic focus on the AI Work Platform' and an 'AI-driven growth strategy.' In a memo the same day, co-CEO Eran Zinman told the roughly 620 departing employees the decision 'was not made to reduce costs or replace people with AI.' Both statements are true, and that is the story: the jobs went away because of a bet on AI, not because AI did the work.

Continue the story on unscarcity.ai:

• AI Layoffs 2025-2030: AI Now #1 Reason for US Job Cuts
• When Does Post-Scarcity Arrive? Realistic Timeline 2026-2060

Full episode notes on unscarcity.ai

OpenAI made Health in ChatGPT broadly available to US adults on Thursday, letting the chatbot read users' medical records and wearable data to give personalized health guidance. It launched one day after Scott Winters, a 55-year-old Florida pastor, sued OpenAI and Sam Altman, alleging ChatGPT repeatedly told him not to seek care before a near-fatal pulmonary embolism.

Continue the story on unscarcity.ai:

• The Substitution Threshold: When You Stop Being the Cheapest Option
• $15,474 on Healthcare vs. $8,000 for Everything

Full episode notes on unscarcity.ai