Category: Business

• WSJ: Microsoft Profits Jump 31% as Azure Cloud Sales Surpass $100 Billion (Jul. 29, 2026)
  Microsoft reported strong cloud-driven results: revenue rose 18% to $90B, net income grew 31% to $35.8B, and Azure topped $100B in fiscal revenue. 
• NY Times: Microsoft Increases Spending on A.I. as Profit Jumps 31.6% (Jul. 29, 2026)
  Microsoft sharply increased AI and data-center spending, reporting $41 billion in quarterly capex, while profit rose to $35.8 billion, and revenue reached $90 billion. 
• Artificial Analysis: Kimi K3: second only to Fable 5 on AA-Briefcase (Jul. 21, 2026)
  Kimi K3 ranks second to Fable 5 on AA-Briefcase, showing strong analytical performance, weaker presentation, and a 51% rubric pass rate. 
• NY Times: A.I. Companies Are Recruiting Electricians and Carpenters by the Thousands (Jul. 29, 2026)
  Data center construction is drawing electricians with high pay, overtime, and bonuses, creating labor competition and straining other projects. 
• NY Times: A Deluge of A.I. Computing Power Is About to Come Online, Fueling Major Leaps (Jul. 29, 2026)
  The number of A.I. chips in use is roughly doubling every nine months, rising from about 30 million in 2024 to roughly 150 million by 2028, and powering advances in weather forecasting, Alzheimer’s research, and mathematics.
• NY Times: What Is Open-Weights A.I.? (Jul. 28, 2026)
  “Open weights” means publishing a model’s learned numerical parameters so others can inspect and modify its behavior; this is distinct from releasing all underlying code.
• NY Times: What A.I. Is Actually Doing to the Economy (Jul. 27, 2026)
  AI is reshaping the economy, but its true effects are unclear: official data lag, private datasets disagree, and companies may blame AI for layoffs. Adoption is fast, and economists warn it could be transformative. So far, job impacts seem subtle.
• The Chronicle of Higher Education: Jagged Intelligence: Does AI grading make the grade? (Jul. 28, 2026)
  Detecting AI‑generated writing is increasingly difficult: popular detectors (Turnitin, GPTZero) are not fully reliable, can produce false positives, and tend to flag students with low AI literacy.
• NY Times: An A.I. Music F.A.Q.: Can I Remix Madonna? Is This All Legal? (Jul. 24, 2026)
  Music A.I. promises to change how songs are made, shared, and promoted, offering new tools for creators and fans. But it raises copyright, legal, and fraud concerns, and could cut artists’ incomes, dilute credit, or flood listeners with generic songs.

Mark Zuckerberg’s newest manifesto casts him as a defender of the individual against concentrated technological power. He argues that “superintelligence” should be widely distributed, enabling people to create, compete, and retain leverage against institutions. Zuckerberg presents openness, individual empowerment, invention, and balance of power as the principles that should govern AI’s future.

Those principles are appealing, but they’re also hard to separate from Meta’s immediate competitive interests.

Zuckerberg hasn’t always treated decentralization as a civic imperative. In 2017, he presented Facebook as essential social infrastructure: a platform that could build a supportive, safe, informed, civically engaged, and inclusive global community. In his World Economic Forum manifesto, the language was humanitarian, but the vehicle was a closed platform controlled by Facebook. The company would supply the tools, set the rules, measure the outcomes, and decide how billions of people encountered information and one another.

That arrangement was never merely benevolent. As Ben Thompson argued in 2017, Facebook’s power flowed from its control over the social graph and, therefore, over consumption and distribution. The platform’s stated mission to connect the world sat uneasily beside an advertising model that rewarded engagement. Subsequent years made that tension unmistakable: algorithmic systems designed to maximize attention could amplify outrage, misinformation, and division while also maximizing revenue. Zuckerberg’s concern then wasn’t that one company held too much power; it was that Facebook should wield its extraordinary power more responsibly.

Now, however, the AI landscape looks different. Meta has spent aggressively to compete in the race for frontier models, yet its models are nowhere near the leaders. The company recently delayed its Avocado model after tests showed it trailing rivals. Meta was even considering licensing Google’s Gemini while reorganizing its AI efforts. Meanwhile, its enormous infrastructure spending has spooked investors concerned about debt, cash flow, and uncertain returns, even as its core advertising business keeps humming along.

Against that backdrop, Zuckerberg’s attack on centralized AI power strikes me as strategically convenient. Openness can be a genuine public good, especially if it prevents a handful of labs from becoming gatekeepers to an essential technology. But it can also be a useful argument for a company trying to weaken better-positioned rivals, reduce regulatory barriers, and turn access to models into a competitive equalizer. His call for “personalized superintelligence” should be judged not only by its rhetoric but by who controls the data, distribution, infrastructure, and economic gains behind it.

The broader lesson extends beyond Meta. As AI expenditures swell at Microsoft, Oracle, Meta, OpenAI, and other firms, executives may frame their investments as public missions: empowerment, national competitiveness, safety, productivity, or democratization. Those claims may contain truth. But investors, regulators, and users should recognize CEO posturing for what it often is—an attempt to dress up an expensive, uncertain corporate strategy as an unimpeachable public mission.