2025 May - What I Have Read
Substack
The greatest muscle you can build is urgency. Decrease the time between having an idea and getting it done. Everything changes – Codie Sanchez
You either chase your one big goal with everything you’ve got, or nothing will happen. Trying to be balanced is what’s wrong with society.
Success in any field comes from IMBALANCE.
Hard work only feels bad if it’s building someone else’s dream, not yours.
The most important thing in your career: Speed.
If you answer emails fast, walk fast, talk fast, get sh*t done fast, you will make a lot of money. No sense of urgency, you won't. Nick Huber
― The Most Successful People I Know Have a Psychopathic Sense of Urgency - Unfiltered by Tim Denning [Link]
UBER: Distribution is The King - Capitalist Letters [Link]
Great business with potential of continuous growth and expansion, and cheap stock price.
Ecosystem is a huge advantage because it creates cross-platform efficiencies.

The concept of network effect was first laid out in 1985, by Carl Shapiro and Michael Katz in their seminal paper “Network Externalities, Competition, and Compatibility.”

According to Russ Harris, author of The Happiness Trap, values are “how we want to be, what we want to stand for, and how we want to relate to the world around us.”
Values are attributes of the person we want to be.
― How Successful People Timebox - Nir Eyal's Substack [Link]
Identify your values -> turn values into time commitments -> create a timeboxed calendar -> track distractions -> reflect and refine weekly. Do remember to schedule fun activities, use flexible categories, and be aware that the goal isn't finishing tasks.
Some key ideas in the article backed by behaviorial science:
- According to The Happiness Trap by Russ Harris (Acceptance and Commitment Therapy - ACT), productivity should align with personal values (e.g., health, relationships, growth) rather than just task completion.
- People often ignore realistic time estimates in favor of optimistic ones, leading to overpacked schedules and missed deadlines.
Especially in the Bay Area, the problem isn’t mediocrity—it’s misdirected excellence. Kids under Chua’s parenting style rarely have a choice in their own extracurriculars from elementary through high school. (I doubt being vice-president of the National Honor Society is a dream to most.) Sure, it can produce a passable overachiever who knows how to get A’s. But to produce someone capable of real vision, high agency, and contrarian thinking, the irony is that that overachiever may be ill-prepared as we approach an era where AI handles rote tasks and the knowledge economy demands more creativity.
― How to Raise High-Agency Kids - Rebecca Wang [Link]
True excellence and future success come from fostering agency—self-directed purpose, curiosity, and ownership—rather than forcing kids to conform to hyper-competitive, checklist-driven achievement cultures (like those common in the Bay Area).
What's the root problem? - misdirected excellence
What's the solution? - Give kids structure (boundaries, values) but autonomy (freedom to pursue interests).
This isn’t about faking confidence. It’s about understanding the low-pressure way to join a group.
Our ability to notice intricate details allows us to ask the specific questions that make others feel truly seen.
In a world where everyone is clamoring to be heard, the ability to observe and truly listen becomes your superpower.
Robert Greene's The 48 Laws of Power completed the picture with "Never Outshine the Master", a lesson teaching the power of blending in rather than disrupting. Don't announce your presence; become part of the scenery, then contribute when appropriate.
― The Spy Trick to Joining Any Conversation (Even If You're Anxious) - AnifragileADHD [Link]
For neurodivergent individuals (ADHD, social anxiety, etc.), socializing isn’t about performing—it’s about strategic observation and gradual integration. This article is backed by psychology and behavioral science.
Small tips:
- Stand inside the group (not on edges) and listen silently at first.
- Linger quietly to blend into the social environment.
- Wait for a group member to naturally include you.
- Ask open-ended questions about others’ interests.
- Sustain conversation with follow-up questions.
Articles and Blogs
Scientists discover quantum computing in the brain - Brighter [Link]
This research bridges quantum physics, biology, and information theory, suggesting that life evolved to exploit quantum mechanics for survival and intelligence. It challenges reductionist views of biology and could redefine our understanding of consciousness, disease, and even the origins of life.
Here are the 19 US AI startups that have raised $100M or more in 2025 - TechCrunch [Link]
Just as “internet” evolved from buzzword to business backbone, AI is following the same playbook.
― In 2025, venture capital can’t pretend everything is fine any more - Pivot to AI [Link]
Venture capital in 2025 is a dying industry clinging to AI as its last hope, with most investment funneled into OpenAI and a few other hyped players while the rest of the startup ecosystem collapses. The sector, which thrived on zero-interest-rate euphoria, now faces a harsh reality: no exits, frozen IPOs, and a market unwilling to fund early-stage ventures. VCs blame Trump’s chaotic tariffs—despite many having supported him—but the real issue is their own inability to adapt to a normal economy. The NVCA report offers no solutions, just desperate optimism, as the industry’s leaders—many of whom lucked into success—flail in ideological fringe movements and pray for a miracle. The only remaining question is whether AI will keep the bubble inflated long enough for them to cash out before it all implodes.
The walled garden cracks: Nadella bets Microsoft’s Copilots—and Azure’s next act—on A2A/MCP interoperability - VentureBeat [Link]
Nadella’s endorsement signals Microsoft’s commitment to open protocols over proprietary ecosystems, aligning with his long-standing advocacy for interoperability (e.g., ONNX, GitHub’s multi-model approach). By backing A2A (agent-to-agent communication) and MCP (model-data context standardization), Microsoft ensures Copilot, Foundry, and Azure AI can seamlessly integrate with third-party AI agents and tools. This move preempts enterprise concerns about vendor lock-in, a criticism of past Microsoft products.
Car Companies Are In A Billion-Dollar Software War, And Everyone's Losing - InsideEVs [Link]
why it's so hard to shift from lagacy automaker to SDV (software designed vehicle) company?
- Cultural shift: Legacy automakers treated software as an afterthought, not a core product. Now, they must adopt a Silicon Valley-like approach.
- Supplier dependence: Traditional automakers rely on suppliers for ECUs, creating a tangled web of software layers.
- Safety vs. agility: They must balance "move fast and break things" with "zero defects or recalls."
- Hybrid challenges: Slowing EV demand means SDV systems must also work with internal-combustion vehicles, complicating power and update logistics.
Legacy automakers must become software companies to survive, but the transition is painfully slow and expensive. The winners will be those who can blend Silicon Valley speed with automotive-grade reliability—something no traditional automaker has fully achieved yet.
8 Reasons Leadership Is Hard And Why Few Are Prepared To Lead - Forbes [Link]
The most inspiring leaders today aren’t just adapting—they’re rewriting the rules. Leadership isn’t a pinnacle; it’s a daily practice of courage and reinvention. The world doesn’t need more bosses; it needs architects of possibility.
Summary:
- The Myth of the Omniscient Leader
Shift: From "knowing it all" to curiosity-driven collaboration.
- Action: Adopt a "Learn It All" mindset (Microsoft’s Satya Nadella
famously replaced "Know It All" with this).
- Tool: Host "No Answers Meetings" where leaders openly discuss unsolved problems, inviting teams to co-create solutions. Example: Google’s "20% Time" empowers employees to explore innovations beyond their core roles, democratizing problem-solving.
- Embracing the Illusion of Control
Shift: From command-and-control to adaptive stewardship.
- Action: Practice "Scenario Planning" (like Shell Oil’s famed
strategy) to prepare for multiple futures, not just one.
- Mindset: View volatility as a laboratory for innovation. Spotify’s
"Fail Fast, Learn Fast" approach rewards experimentation.
- Quote: "The art of leadership is not to control, but to unleash." — Reed Hastings, Netflix.
- The Leadership Pipeline Crisis
Root Cause: Short-term efficiency has gutted long-term talent development.
- Fix: Reverse Mentorship Programs (e.g., GE’s junior employees mentor
execs on digital trends).
- Metric: Track "Readiness Ratios"—how many high-potentials are
prepared for next-level roles?
- Warning: Deloitte’s research shows 89% of executives see "weak leadership benches" as their top threat.
- The Death of Cookie-Cutter Playbooks
New Rule: Context over conformity.
- Action: Build "Modular Strategies"—flexible frameworks adjusted in
real-time (like Amazon’s "Working Backwards" method).
- Tool: Use "Pre-Mortems" (anticipating failures before launch) to stress-test strategies. Example: Blockbuster’s rigid playbook failed, while Netflix’s pivot to streaming embraced uncertainty.
- Respect as a Daily Earned Currency
Key: Authenticity > Authority.
- Action: Practice "Radical Transparency" (like Bridgewater
Associates’ culture of brutal honesty).
- Tool: Replace "All Hands Meetings" with "All Hearts Meetings"—forums for empathy and vulnerability. Example: Edelman’s Trust Barometer shows employees trust "a peer like me" 3x more than CEOs.
- Rebuilding Trust in Judgment
Antidote: Inclusive Decision-Making.
- Action: Form "Shadow Boards" (e.g., Gucci’s millennial council
advising execs).
- Rule: For major decisions, require "Disagree & Commit" (document dissent but align once decided). Example: Patagonia’s CEO involves employees in sustainability bets, building trust through shared stakes.
- Titles vs. Influence
New Power Model: Fluid Hierarchies.
- Action: Adopt "Holacracy Lite" (like Zappos’ role-based authority,
not title-based).
- Symbolic Step: Drop "CEO" for "Chief Enabler" (as some startups do
to signal servant leadership).
- Stat: 72% of Gen Z workers prefer "Project Leaders" over "Managers" (McKinsey, 2024).
- The Reinvention Imperative
Framework: "Learn, Unlearn, Relearn" (Alvin Toffler’s future-proofing mantra).
- Tool: "Skills Gap Heatmaps"—quarterly self-assessments on emerging
competencies (e.g., AI literacy).
- Example: Adobe’s "Kickbox" program gives employees $1,000 to test new ideas, forcing leaders to adapt.
The Path Forward: Leadership as a Dynamic Practice
Your closing question—"Will you be one of them?"—is the call
to action. Leaders who thrive will:
1. Lead with Questions, not answers.
2. Treat Trust as Currency, not a given.
3. Build Antifragile Teams (Nassim Taleb’s concept of growing stronger
through chaos).
4. Measure Success in Learning Cycles, not quarterly profits alone.
Microsoft Follows Competitors Amazon, Meta, and Google in Employee Productivity Crackdown [Link]
The pandemic hiring spree, rising interest rates, and the AI arms race have forced tech giants to abandon the "growth at all costs" mindset. Instead, they’re:
- Prioritizing speed (fewer managers = faster decisions)
- Maximizing output per employee (via stack ranking and attrition policies)
- Investing savings into AI (where Microsoft is battling Google and OpenAI)
Master The Psychology Of Building An Unforgettable Personal Brand - Forbes [Link]
When your brand is rooted in internal conviction, it radiates effortlessly. The right opportunities find you.
"My worth isn’t measured by likes; it’s measured by impact."
"If they don’t buy, it’s not a rejection—it’s a mismatch."
"Outcomes are data, not identity."
"Consistency today compounds into authority tomorrow."
Zero to One: Learning Agentic Patterns - Philschmid [Link]
This guide explores techniques such as prompt chaining, routing, parallelization, reflection, tool integration, planning, and multi-agent collaboration. It features practical code examples for each pattern, enabling the development of efficient, context-aware workflows with Google DeepMind Gemini. Emphasis is placed on structured strategies to enhance task delegation and agent coordination.
Our research shows that by 2030, data centers are projected to require \(\text{\$6.7}\) trillion worldwide to keep pace with the demand for compute power. Data centers equipped to handle AI processing loads are projected to require \(\$5.2\) trillion in capital expenditures, while those powering traditional IT applications are projected to require \(\$1.5\) trillion in capital expenditures (see sidebar “What about non-AI workloads?”). Overall, that’s nearly \(\$7\) trillion in capital outlays needed by 2030—a staggering number by any measure.
To qualify our \(\$5.2\) trillion investment forecast for AI infrastructure, it’s important to note that our analysis likely undercounts the total capital investment needed, as our estimate quantifies capital investment for only three out of five compute power investor archetypes—builders, energizers, and technology developers and designers—that directly finance the infrastructure and foundational technologies necessary for AI growth (see sidebar “Five types of data center investors”). Approximately 15 percent (\(\$0.8\) trillion) of investment will flow to builders for land, materials, and site development. Another 25 percent (\(\$1.3\) trillion) will be allocated to energizers for power generation and transmission, cooling, and electrical equipment. The largest share of investment, 60 percent (\(\$3.1\) trillion), will go to technology developers and designers, which produce chips and computing hardware for data centers. The other two investor archetypes, operators, such as hyperscalers and colocation providers, and AI architects, which build AI models and applications, also invest in compute power, particularly in areas such as AI-driven automation and data center software. But quantifying their compute power investment is challenging because it overlaps with their broader R&D spending.
― The cost of compute: A $7 trillion race to scale data centers - McKinsey [Link]
The Comfortable Life is Killing You - Poetic Outlaws [Link]
Meaning is forged in resistance - Meaning is a byproduct of engagement with resistance. Joy emerges when we meet challenges worthy of our souls. To paraphrase Camus: The struggle itself is enough.
Agentic AI Is Already Changing the Workforce - Harvard Business Review [Link]
Papers and Reports
The power of one: How standout firms grow national productivity - McKinsey Global Institute [Link]
Productivity growth is crucial for economic prosperity. The report suggests that instead of waiting for all firms to improve, targeted support for high-potential firms could accelerate national productivity gains.
Identifying and scaling AI use cases - OpenAI [Link]
OpenAI ads, but useful for pitching GenAI use cases. It offers guidance on identifying and scaling AI use cases within organizations, noting that AI adoption is rapidly increasing and demonstrating significant benefits for early adopters. It emphasizes three key steps for businesses: understanding where AI can add value by focusing on repetitive tasks, skill bottlenecks, and navigating ambiguity; teaching teams fundamental AI use cases like content creation, research, and automation; and prioritizing opportunities using an impact/effort framework to determine which projects to pursue and scale.
ZeroSearch: Incentivize the Search Capability of LLMs without Searching - Alibaba Group [Link]
Traditional RL training requires massive API calls to services like Google Search, costing hundreds of thousands of dollars. ZeroSearch replaces this with a simulated search environment where the LLM itself generates both relevant and irrelevant documents in response to queries.
Real search engines return unpredictable results, complicating training. While ZeroSearch uses curriculum-based rollouts, gradually degrading document quality to teach the model to discern useful information.
It has a cost reduction up to 88% and its performance surpasses real search engines.
AI Global, Global Sector Trends on Generative AI [Link]
Gen AI Traffic Share update - Similarweb @Twitter [Link]
Subdomains and pages only (below).

YouTube and Podcasts
Fed Hesitates on Tariffs, The New Mag 7, Death of VC, Google's Value in a Post-Search World - All-In Podcast [Link]
The Physical Turing Test: Jim Fan on Nvidia's Roadmap for Embodied AI - Sequoia Capital [Link]
This lecture introduces the Physical Turing Test, a new benchmark for robotics. Jim Fan from NVIDIA breaks down why solving this is hard—and what tools researchers are using to make progress.
5 Types of AI Agents: Autonomous Functions & Real-World Applications - IBM Technology [Link]
This lecture covers reflex agents, model-based agents, goal-based systems, utility-based frameworks, and learning agents.
Stanford Webinar - Agentic AI: A Progression of Language Model Usage - Stanford Online [Link]
How to connect AI agents to third-party tools using MCP - Underfitted [Link]
Llamacon 2025 - Conversation with Mark Zuckerberg and Satya Nadella - Meta Developers [Link]
关税大棒下的苹果:一场全球供应链的迁徙风暴 - 硅谷101 [Link]
E191|小而美的机会来了,聊聊这轮AI Agent进化新范式 - 硅谷101 [Link]
Sundar Pichai, CEO of Alphabet | The All-In Interview [Link]
Trump's Big Week: Middle East Trip, China Deal, Pharma EO, "Big, Beautiful Bill" with Ben Shapiro - All-In Podcast [Link]
枪声背后的信任危机:“病不起”的美国人 - 硅谷101 [Link]
Bond crisis looming? GOP abandons DOGE, Google disrupts Search with AI, OpenAI buys Jony Ive's IO - All-In Podcast [Link]
Microsoft Build 2025 | Satya Nadella Opening Keynote - Microsoft [Link]
Exciting new products - copilot studio, foundry local, microsoft discovery, etc!
Google I/O '25 Keynote - Google [Link]
AI mode finally - Smart move to embrace next-gen search. Android XR glass is launching, and Gentle Monster + Warby Parker will be the first eyewear partners. Genimi App has Agent mode is coming. And many more!
NVIDIA CEO Jensen Huang Keynote at COMPUTEX 2025 - NVIDIA [Link]
NVLink Fusion, DGX Spark AI Computer, DGX Station Super Computer, FTX Pro Server, AI Robotics, etc.