High Earners: Don't Let Taxes Eat Your Returns
You worked hard for your investments.
A great portfolio can still underperform if your tax strategy is working against you. The smartest investors don't just earn more—they keep more.
Learn proven tax-smart investing strategies from experienced CFPs and CPAs, and discover how to reduce tax drag while building long-term wealth.
Today In Ai
1 AI executives say demand is unlimited, even as enterprise customers get more deliberate about what they spend it on.
Former Intel CEO Pat Gelsinger told CNBC that AI demand is "almost unlimited." Marc Boroditsky, CRO of AI infrastructure firm Nebius, put the same point differently: "there's much more demand than we're able to fill." The nuance in the data is that companies are becoming more careful about which AI deployments they fund, not that overall interest is softening. Demand is holding. The question of which specific applications earn the budget is where the real pressure is building.
2 Tom Blomfield left Y Combinator to join Anthropic's AI compute team, the latest in a string of high-profile hires.
Blomfield announced the move on X, joining a roster that now includes Nobel laureate John Jumper, OpenAI co-founder Andrej Karpathy, and UC Berkeley professor Jelani Nelson. Former Federal Reserve Chairman Ben Bernanke also joined Anthropic's independent oversight board this week. Anthropic is assembling a team that spans frontier AI research, hardware, and institutional credibility in a way that few labs have attempted.
3 TSMC posted its highest-ever monthly revenue in June, closing a quarter at $39.6 billion as AI chip demand stays tight.
The world's largest contract chipmaker reported a 68% revenue jump year over year for June, driven by sustained demand for the leading-edge AI GPUs and CPUs its fabs produce. TSMC is fully sold out of N3, the process node behind this year's top AI silicon, and is building two additional plants in southern Taiwan to address a backlog that is not expected to clear soon. The numbers confirm what chip executives have been saying all year: the demand picture has not changed.
From The Frontier
The economists' statement. More than 200 economists, including 16 Nobel laureates, signed We Must Act Now, a public statement urging governments to steer AI development toward broad human benefit rather than concentrated private gain. The signatories are not fringe voices. Getting 16 Nobel laureates to agree on anything is an unusual event. The statement has arrived as public sentiment on AI has become measurably more anxious.
What the public actually wants. A Verasight survey found that 60% of respondents feel anxious about AI's rise. The specific reforms they want are striking in their specificity: 89% support requiring frontier labs to publicly disclose safety-testing results before a model launches. 81% want the government to have the authority to block potentially dangerous models before release. These are not soft preferences. They are supermajority positions.
The sovereign wealth fund idea. 69% of Verasight respondents also supported forcing AI companies to hand over equity stakes into a fund that would distribute AI's financial gains more broadly. In theory, such a fund would fill three roles: steering AI development at a national level, taking equity positions in new companies, and converting private-sector wealth creation into public revenue to fund social safety nets. It is no longer a fringe idea.
Where things already stand. OpenAI proposed letting the US government take a 5% stake earlier this month. The US government already reviewed both Anthropic and OpenAI's most powerful models before they were allowed back into public access. The gap between what the public is asking for and what is already being built is smaller than most coverage suggests. Whether the next steps are collaborative or adversarial between labs and regulators is the question the next 12 months will answer.
What people are actually watching and sharing
Show us what you built. Sam Altman asked his followers to share what they have made with GPT-5.6. The thread now has 3M views and 3,000 responses, and the range is wide: solo founders shipping SaaS products, developers automating entire workflows, and people building things that would have required a team six months ago. Worth browsing for ideas.
Design terms, finally explained. Name That UI is a visual dictionary of web design terminology that helps you describe what you want when prompting an agent to build something. Try it here (nearly 7,000 bookmarks in two days). The gap between "make it look cleaner" and knowing the specific term for what you mean is exactly where most AI design prompts fall apart.
Testing the image limits. Redditors are pushing AI image generation to its outer edge in a thread with over 3,000 upvotes. The first image is already striking. The comment section escalates from there. Useful if you want to understand the actual boundaries of what current models will and won't produce.
A better model picker. Someone redesigned ChatGPT's model selector to make it more interactive and contextual, and the concept has pulled 1.5M views. It highlights how much friction the current implementation creates and how little it would take to fix it.
Model vs. effort. Anthropic's developer team published a detailed breakdown of when to reach for a stronger model versus when to invest more effort in your prompt. The distinction matters more than most people realize, and this post is the clearest explanation of the trade-off currently available.
Prompt Station
Map every inefficiency in your team's workflow in one prompt
This ChatGPT prompt runs a full productivity audit on any team, evaluating how it manages meetings, communication, task planning, reporting, accountability, decision-making, and collaboration. It surfaces root causes rather than symptoms, recommends specific frameworks where they fit, and delivers a prioritized roadmap organized by impact and implementation difficulty. Paste it in, fill the single placeholder with your team's specifics, and let it ask clarifying questions before producing output.
Act as an experienced productivity consultant and conduct a comprehensive analysis of my team's productivity system to identify inefficiencies, bottlenecks, and opportunities for improvement. Evaluate how the team manages meetings, communication, task planning and execution, reporting, accountability, decision-making, collaboration, and overall workflow efficiency. Assess whether time is being used effectively, where unnecessary complexity or duplication exists, and how current processes impact productivity, focus, and team performance. Based on your analysis, recommend practical improvements, including optimized workflows, communication practices, meeting structures, task management strategies, reporting systems, accountability mechanisms, automation opportunities, and suitable productivity tools. Where appropriate, suggest proven productivity frameworks or methodologies (such as Agile, Kanban, Scrum, OKRs, GTD, or Lean) and explain why they fit the team's needs. Prioritize your recommendations by potential impact and ease of implementation, and provide a clear, actionable roadmap for improving the team's productivity. Team Details: [TEAM INFO].Replace [TEAM INFO] with a description of your team covering: size, function (engineering, marketing, operations, etc.), current tools you use, your biggest time wasters, and one specific outcome you want to improve in the next 90 days. The more concrete the input, the more actionable the framework recommendations. The prompt instructs the model to ask clarifying questions before producing output, so let it ask before you add more detail. The questions it generates are often more diagnostic than the answers.
You’re Invited: Tax-Smart Investing webinar on July 23. Learn the portfolio moves that can help you maximize your after-tax returns — join Range live, and bring your questions for Q&A.