The 3 Revolutions That Changed the World

The opening chapter frames the AI boom as the fastest technological shift in human history and sets up the book's investing case. It traces three revolutions: the Industrial Revolution (James Watt's 1765 engine burned 75% less coal; UK GDP per capita rose from £1,750 in 1760 to £3,000 by 1840), the Internet era, and AI. It cites world GDP per capita staying nearly flat for over 1,700 years in the Malthusian Trap, then shows adoption accelerating from the telephone's 75 years to 50 million users down to ChatGPT's single month. OpenAI launched ChatGPT on November 30, 2022; it reached 100 million users in about two months. The chapter also notes NVIDIA's FY2024 revenue of $130.5 billion and every past revolution's bubble (the 1845 Railway Mania, the 2000 dot-com crash that cut the Nasdaq 78%). Six lessons follow, including buying "pick and shovel" infrastructure, respecting bubbles without avoiding them, and acting before feeling ready. The takeaway: AI changes brain-based professions, and the U.S. hosts 7 of the world's top 10 tech companies.

Year	Event	Users (Millions)
1969	ARPANET — the beginning	4 universities
1983	TCP/IP becomes the standard	A few thousand
1989	Tim Berners-Lee invents the World Wide Web	Zero
1993	Mosaic — first graphical browser	~1
1995	Amazon, eBay, Yahoo founded	16
1997	Amazon IPO ($18/share)	75
1998	Google is born	150
2000	Dot-com Bubble peaks, then crashes	~400
2004	Facebook launches; Google IPO ($85/share)	~850
2005	YouTube launches	1,000 (1B)
2007	iPhone launches — the Internet goes mobile	~1,500
2010	Instagram, cloud computing takes off	~2,000
2015	Mobile-first world	~3,200

Year	Event	Impact
1712	Thomas Newcomen builds the first commercial steam engine	Pumps water from coal mines, but guzzles fuel
1765	James Watt adds a separate condenser	Burns 75% less coal — now it's commercially viable
1775	Watt partners with Matthew Boulton to form a company	Mass production of steam engines begins
1780s	Cotton mills powered by steam for the first time	Fabric production speeds up 10–20x
1790s	Cheap steel (Henry Cort's puddling process)	Bridges, railroads, and tall buildings become possible
1804	First steam locomotive (Richard Trevithick)	Goods move 5x faster than by horse
1825	Stockton–Darlington Railway — world's first passenger railway	Humans travel faster than ever in history
1830	Liverpool–Manchester Railway opens	Mass transit begins
1840s	Industrial Revolution spreads to Europe and America	Textile mills in New England, steel in Pennsylvania

The 3 Layers of AI Investing — Infrastructure, Platform, Application

This chapter gives a framework for investing in AI through three layers, using a Gold Rush analogy: selling picks and shovels (Infrastructure, lower risk), running a store in town (Platform), and digging for gold (Application, highest risk). The Infrastructure layer centers on NVIDIA, founded in 1993, whose CUDA platform (launched 2006) and roughly 80% AI-accelerator share produced $215.9 billion in fiscal 2026 revenue; TSMC controls 71% of the foundry market. The Platform layer pits Microsoft, Google, and Amazon, where cloud hit $129 billion per quarter in Q1 2026 and Microsoft's $1 billion OpenAI bet in 2019 grew past $13 billion, with Anthropic valued at about $965 billion versus OpenAI's ~$852 billion. The Application layer warns against thin "wrapper" startups. The chapter recommends ETFs, starting with QQQ at 0.20%, and suggests an allocation of 60% Infrastructure, 25% Platform, 10% Application, and 5% cash. The takeaway: choose the layer that matches your risk tolerance and time.

ETF	Focus	Expense Ratio	Layer	Best For
QQQ 🥇	Nasdaq 100 — AI + all big tech	0.20%	All Layers	Beginners — start here
SMH	Semiconductor (NVIDIA, TSMC, Broadcom)	0.35%	Layer 1	If you believe in infrastructure
SOXX	Semiconductor (broader than SMH)	0.35%	Layer 1	Alternative to SMH
IGV	Software & tech companies	0.41%	Layer 2+3	Platform + Application exposure
BOTZ	Robotics + AI	0.68%	Layer 2+3	AI + robotics theme
AIQ	Broad AI — every layer	0.68%	All Layers	Full AI exposure in one ticker
CHAT	Generative AI	0.75%	Layer 3	Pure-play generative AI
QTUM	Quantum computing + AI	0.40%	Layer 1	Quantum + AI infrastructure

Stock	Role in AI	Why It Matters
Broadcom (AVGO) 🥇	Networking chips for AI data centers	"The nervous system of AI" — every data center needs its chips
Vertiv (VRT)	Power and cooling systems	Every data center needs cooling — and Vertiv is the market leader
Equinix (EQIX)	Data center REIT #1	Owns the physical real estate where AI runs — plus a solid dividend
Digital Realty (DLR)	Data center REIT	Same story as Equinix — pure play on data center demand
AMD (AMD)	GPU/CPU alternative to NVIDIA	The #2 player — gaining ground with MI300/MI400 series
Arm Holdings (ARM)	CPU architecture for AI	Power-efficient chips designed into everything from phones to servers

Deep Dive into Individual AI Stocks — From Global Giants to US Opportunities

This chapter profiles individual AI stocks in four dimensions — business, numbers, risks, and playbook. NVIDIA, born in 1993, posted $215.9 billion in FY2026 revenue and about $193.7 billion in data-center sales, defended by CUDA and roughly 80-90% accelerator share. TSMC, founded in 1987 by Dr. Morris Chang, holds 71% foundry share but carries Taiwan risk. Microsoft, led by Satya Nadella since 2014, runs a $37+ billion AI business; Alphabet launched Gemini after a "Code Red" and hit $20 billion in quarterly Google Cloud revenue; Amazon leads cloud with 28% share and invested in Anthropic. Broadcom supplies networking chips and AMD's MI400 series was projected at ~$7.2 billion in 2026. Beyond the giants, it covers Vertiv, Palantir (~$3.5 billion run rate), and CrowdStrike (~$5 billion). It offers beginner, intermediate, and aggressive sample portfolios and a glossary on CUDA, lock-in, and five-year horizons. The takeaway: match conviction to a 5-year minimum horizon and diversify.

Stock	Layer	Revenue	Growth	Moat	Risk	Rating
NVIDIA 🥇	Infra	$215.9B	+65%	CUDA + HW	Geopolitical	⭐⭐⭐
TSMC 🥈	Infra	>$100B	~30%	71% share	Taiwan risk	⭐⭐⭐
Microsoft	Platform	$82.9B/Q	+18%	Enterprise lock-in	Antitrust	⭐⭐⭐
Google	Platform	$20B/Q (Cloud)	+63%	TPU + Data	Antitrust	⭐⭐
Amazon	Platform	AWS #1	Steady	Ecosystem	Share erosion	⭐⭐
Broadcom	Infra	>$50B	Strong	Networking dominance	P/E	⭐⭐
AMD	Infra	~$25B	Growing	CPU + GPU	Share gap	⭐⭐

Metric	Number
FY2026 Total Revenue	$215.9B (+65% YoY)
FY2026 Data Center Revenue	~$193.7B
Q1 FY2027 (latest)	$81.6B (+20% QoQ, +85% YoY)
AI Accelerator Market Share	~80–90% (vs AMD ~10%)
Next Platform	Vera Rubin (Vera CPU + Rubin GPU) — partners start H2 2026

How to Invest in AI Stocks — Surviving and Profiting

This chapter teaches how to monitor AI stocks without staring at charts, opening with the Gold Rush stories of Samuel Brannan and Levi Strauss to reinforce the infrastructure thesis. It builds an investor calendar around NVIDIA earnings — the "Super Bowl of AI" — noting that in February 2024 NVIDIA reported $18.4 billion in data-center revenue, up 409% year-over-year, and surged 16% the next day, adding about $277 billion in market cap. It then defines five KPIs: AI revenue run rate (contrasting Stability AI's ~$8-11 million 2023 revenue against ~$153 million in expenses), CapEx, NVIDIA gross margin (64.9% in FY2022, ~75% in FY2025, 71.1% in FY2026), adoption (900 million weekly ChatGPT users), and patent counts. Sell signals and green signals follow, along with a 30-minute weekly routine and a review of NVIDIA's past corrections (down 66% in 2022). The takeaway: buy infrastructure, expect corrections, and never go all-in on one theme.

Timing	Event	What to Watch
Mid Feb	NVIDIA Q4 Earnings	Data Center revenue, Gross Margin, Guidance
Mid Apr	NVIDIA Q1 Earnings	YoY growth — accelerating or decelerating?
Mid Aug	NVIDIA Q2 Earnings	Data Center revenue vs. estimates
Mid Nov	NVIDIA Q3 Earnings	Outlook for next year + Vera Rubin updates
Every Quarter	Microsoft, Google, Amazon Earnings	AI Revenue Run Rate, CapEx spending
Every Quarter	Vertiv, Equinix, Broadcom Earnings	Data center buildout trajectory

KPI #3: NVIDIA Gross Margin — The Mother of All AI Metrics

Chart drawn from the table below, using the book’s own figures.

Fiscal Year	Gross Margin	The Story
FY2022	64.9%	Baseline before AI boom
FY2023	56.9%	⚠️ Panic! — Inventory write-down (gaming slump + crypto crash)
FY2024	72.7%	🟢 AI chip demand surged — NVIDIA gained pricing power
FY2025	~75%	👑 Highest in company history
FY2026	71.1%	Vera Rubin ramping — margins slightly normalized

AI Agents — The Real Game Changer of 2026–2030

This chapter argues AI agents — systems that take a goal and act autonomously — are the next phase, distinct from ChatGPT's question-and-answer model. It cites Cognition Labs' Devin (March 2024), which scored about 14% on the SWE-bench benchmark versus under 2% for prior systems, and Anthropic's Claude Code in 2025; by 2026 an estimated 30%+ of software companies use agents for mid-level engineering work. It projects the AI-agent market growing from $7.84 billion in 2025 to $52.62 billion by 2030 (MarketsandMarkets) and notes Goldman Sachs forecasting a 24x rise in token consumption by 2030. Stocks covered include NVIDIA's Vera Rubin platform, TSMC, Broadcom, Vertiv, and Equinix, plus service-software names under pressure: ServiceNow fell ~33% and Palantir ~23% in 2026. It gives real examples at JPMorgan (agents handling ~30% of inbound calls), UnitedHealth, and Walmart, then a 2026-2030 milestone timeline. The takeaway: infrastructure benefits most as agents multiply compute, power, and cooling demand.

Year	Key Milestone
2026	Vera Rubin enters production. AI agents adopted by large enterprises
2027	AI agents become widespread — operational costs drop 30–50%
2028	AI agents begin replacing mid-level management functions
2029	Token consumption approaches +24x (vs. 2026 baseline)
2030	AI agent market >$50B — virtually every Fortune 500 company uses them

ETF	Expense Ratio	Layer	Best For
QQQ 🥇	0.20%	All Layers	Beginners — start here
SMH	0.35%	Layer 1	Believe in infrastructure (chips)
AIQ	0.68%	All Layers	Full AI exposure
BOTZ	0.68%	2+3	AI + Robotics
CHAT	0.75%	Layer 3	Generative AI focus

The AI Platform War — Microsoft vs. Google vs. Amazon

This chapter frames a winner-take-most battle between Microsoft, Google, and Amazon, together worth over $10 trillion and spending more than $550 billion in combined 2026 CapEx ($700 billion including Meta). The cloud market reached $129 billion per quarter in Q1 2026, split among AWS (28-31%), Azure (21-24%), and Google Cloud (12-16%). Microsoft's history runs from buying QDOS for $50,000 in 1980, to the 1998 antitrust case, to Nadella's 2019 $1 billion OpenAI investment that grew past $13 billion and a $37+ billion AI run rate. Google's story covers the Transformer paper (2017), the $500 million DeepMind acquisition, AlphaGo's 2016 win, the "Code Red," and the "Google Graveyard" trust problem; Google Cloud hit $20 billion (+63%) with a $460 billion backlog. Amazon's covers Bezos' regret framework, AWS's 2006 launch, the $170 million Fire Phone failure, and its Anthropic bet (~$965 billion). The takeaway: Microsoft is the safest, Google the cheapest, Amazon the most diversified — but infrastructure wins regardless.

Year	Microsoft	Google	Amazon
1985	DOS creates empire	—	—
1994	Windows 95	—	Bezos launches Amazon
1998	Antitrust trial	Page & Brin found Google	IPO
2002	—	Turns down Yahoo's $3B offer	—
2006	Vista flops	Acquires YouTube	AWS launches
2014	Nadella becomes CEO	Acquires DeepMind $500M	Fire Phone fails
2016	—	AlphaGo beats Lee Sedol	Echo/Alexa launch
2019	Invests $1B in OpenAI	—	—
2022	ChatGPT launches	"Code Red" declared	—
2024	AI Revenue $37B+	Gemini + TPU	Invests $4B in Anthropic
2026	Copilot: 400M users	Google Cloud $20B/Q (+63%)	Anthropic valued $965B

Dimension	Microsoft 🥇	Google 🥈	Amazon 🥉
AI Technology	⭐⭐⭐ (GPT-5.5)	⭐⭐⭐ (Gemini 3 + TPU)	⭐⭐ (Claude via Anthropic)
Enterprise Trust	⭐⭐⭐	⭐⭐ (Google Graveyard)	⭐⭐⭐
Cloud Market Share	21-24%	12-16%	28-31%
AI Revenue Clarity	⭐⭐⭐ $37B+	⭐⭐ Not clearly broken out	⭐⭐ AWS AI services
CapEx 2026	~$190B	~$175-190B	~$200B
Valuation (P/E)	~35x	~25x (cheapest)	~40x
Consumer AI	⭐⭐⭐ ChatGPT integration	⭐⭐⭐ Gemini	⭐ None
Biggest Risk	Antitrust	Enterprise Trust	No consumer AI

How to Analyze AI Stocks Yourself

This chapter supplies a systematic framework for judging AI stocks, warning that most 2024 buyers will lose money by 2030 because transformative technology and profitable investing are different things. It teaches five metrics — price-to-earnings, price-to-sales, the PEG ratio, free-cash-flow yield, and gross margin — each with interpretation ranges, applied to a comparison table of NVIDIA, Microsoft, Alphabet, TSMC, and Amazon. It then presents three mental models: Peter Lynch's PEG ratio and "buy what you know," Warren Buffett's economic moat (ecosystem lock-in, data network effects, scale, switching costs), and Benjamin Graham's margin of safety. Five common traps are dissected, including the "this time is different" fallacy, loving the technology, ignoring capital expenditure, confusing revenue with profit, and overestimating your ability to pick winners. A ten-point pre-purchase checklist follows, closing with Isaac Newton's South Sea Bubble loss of £20,000. The takeaway: use process and discipline, not CNBC or Reddit, and never evaluate an AI stock on P/E alone.

P/S Ratio Range	What It Tells You
1–3	Reasonable / potentially undervalued
3–8	Growth priced in
8–15	Rich — needs excellent growth to justify
15–30	Speculative territory
30+	Pricing in perfection — any miss will hurt

Company	Ticker	P/E (TTM)	P/S	PEG (1yr fwd)	Gross Margin	FCF Yield	AI Moat
NVIDIA	NVDA	~50–60	~25	~1.5	72–75%	~1%	GPUs + CUDA ecosystem
Microsoft	MSFT	~33–37	~11	~2.0	68–70%	~2.5%	OpenAI partnership, Azure AI
Alphabet	GOOGL	~23–27	~6	~1.5	55–58%	~3.5%	DeepMind, TPU, massive data
TSMC	TSM	~20–25	~8	~1.2	50–53%	~3%	Manufacturing monopoly for advanced chips
Amazon	AMZN	~40–50	~3	~2.5	45–48%	~2%	AWS + custom AI chips (Trainium/Inferentia)

The Specific Risks of AI Stocks — Know Before You Lose Money

This chapter catalogs seven structural risks that threaten AI investors. First, AI Winter: the first (1974-1980) followed the Lighthill Report cutting UK funding, and the second (1987-1993) collapsed the expert-systems industry. Second, valuation bubble: the Nifty Fifty saw Polaroid fall 91% and McDonald's 64%, while today NVIDIA traded near 50x and Palantir above 100x earnings. Third, regulation: the EU's GDPR (2018) brought a €1.2 billion fine and became a global template, with the AI Act arriving in 2024. Fourth, geopolitics: TSMC makes roughly 90% of the most advanced chips, and the 2022 CHIPS Act allocated $52.7 billion to reshore production. Fifth, competition: SoftBank's Vision Fund and WeWork's $14 billion write-down show being right about AI doesn't make you right about a stock. Sixth, execution: IBM Watson failed in hospitals and was sold at a loss in 2022. Seventh, concentration. A risk-management matrix and rules follow. The takeaway: no single AI stock should exceed 5% of a portfolio.

Risk Factor	Red Flag (Avoid/Reduce)	Yellow Flag (Monitor)	Green Flag (Acceptable)	How to Mitigate
AI Winter	Company priced on AI breakthroughs that don't exist yet	Revenue depends on AI adoption continuing at current pace	Revenue comes from proven AI products with real customers	Limit exposure to pre-revenue AI companies; favor profitable AI firms
Valuation	P/S > 20, no path to profitability, insider selling	P/S 10–20, founder still buying, improving margins	P/S < 10, positive FCF, reasonable growth assumptions	Use position sizing — no single AI stock > 5% of portfolio
Regulation	Business model depends on regulatory gray area	Compliance costs manageable, legal team in place	Actively working with regulators, diversified revenue	Own companies with strong compliance and government relations
Geopolitics	Entire supply chain through Taiwan or China	Multiple suppliers, some domestic production	Diversified supply chain, CHIPS Act beneficiary	Pair AI chip stocks with software AI stocks as a hedge
Competition	No moat, no data advantage, low switching costs	Some moat (integration, contracts), known competitors	Strong network effects, proprietary data, pricing power	Favor platforms over point solutions in AI
Execution	Product demos well but doesn't work in production	Product works but unit economics unproven	Proven product with expanding margins and retention	Read product reviews and customer interviews, not press releases
Concentration	Portfolio has 50%+ in AI stocks	Portfolio has 20–50% in AI stocks	Portfolio has 5–15% in AI stocks with broad diversification	Rebalance quarterly; set hard limits on single-sector exposure

Company	P/E (late 2024)	What It Priced In
NVIDIA	~50	3–5 more years of hypergrowth
Microsoft	~35	AI adding 10%+ to revenue growth
CrowdStrike	~80	Dominance in AI-powered cybersecurity
Palantir	~100+	AI platform becoming a standard enterprise tool

AI Portfolio Construction for US Investors — From Theory to Practice

This chapter turns theory into buildable portfolios, grounded in Harry Markowitz's 1952 efficient frontier, the Core + Satellite framework, the Layered Approach, and local-plus-global diversification. It presents four models. The Conservative portfolio (VOO 40%, QQQ 20%, VRT 15%, EQIX 15%, cash 10%) targets 6-8% annualized. The Balanced portfolio (NVDA 15%, MSFT 15%, QQQ 20%, SMH 10%, VOO 15%, AVGO 10%, cash 15%) targets 9-12%. The Aggressive portfolio (NVDA 30%, TSM 15%, SMH 15%, AIQ 10%, VRT 10%, PLTR 10%, cash 10%) targets 12-20% with 40-60% drawdowns. The Stealth Wealth portfolio (VOO 25%, QQQ 20%, MSFT 15%, GOOGL 10%, EQIX 10%, BND 10%, GLD 5%, cash 5%) targets 7-10%. It then covers broker selection, account prioritization, DCA versus lump sum, rebalancing, tax considerations, and reflections from Markowitz, Ray Dalio, Cathie Wood, Peter Lynch, Charlie Munger, and Howard Marks. The takeaway: the best portfolio is the one you can hold through the inevitable ups and downs.

Ticker	Name	Weight	Role
VOO	Vanguard S&P 500 ETF	25%	Core equity — broad US market
QQQ	Invesco QQQ Trust	20%	Growth/AI tilt — Nasdaq-100
MSFT	Microsoft	15%	Steady AI leader — enterprise moat, growing dividend
GOOGL	Alphabet (Google)	10%	AI through DeepMind, Google Cloud, massive cash flow
EQIX	Equinix	10%	Data center REIT — real assets, growing dividend
BND	Vanguard Total Bond Market ETF	10%	Fixed income ballast — reduces portfolio volatility
GLD	SPDR Gold Trust	5%	Inflation hedge, tail-risk hedge, uncorrelated asset
Cash	Money market or SGOV	5%	Liquidity buffer

Ticker	Name	Weight	Role
NVDA	Nvidia	15%	AI compute — the engine of the AI revolution
MSFT	Microsoft	15%	Enterprise AI — Azure, Copilot, OpenAI partner
QQQ	Invesco QQQ Trust	20%	Diversified AI/growth exposure
SMH	VanEck Semiconductor ETF	10%	Semiconductor industry broad play
VOO	Vanguard S&P 500 ETF	15%	Core US equity ballast
AVGO	Broadcom	10%	AI networking and custom chips (ASICs)
Cash	Money market or SGOV	15%	Liquidity and drawdown protection

AI in Other Industries — Healthcare, Defense, Finance, and Beyond

This chapter argues AI is reshaping every data-rich industry, so owning only big tech is a concentrated bet, not diversification. In healthcare, it highlights Eli Lilly (founded 1876, ~$800 billion market cap, AI-accelerated drug discovery), Alphabet's DeepMind (AlphaFold mapped over 200 million proteins), Amazon's health infrastructure, the pure-play Recursion, and Hims & Hers. In defense, it covers Palantir (~$4 billion run rate, ~55% government revenue), AeroVironment and its Switchblade drone used in Ukraine, Lockheed Martin, and Northrop Grumman. In fintech it examines SoFi, PayPal, Block, and Robinhood (Gold Card with 1M+ signups). It also surveys agriculture (Deere's autonomous tractors cutting herbicide use ~70%), education (Duolingo), energy (Constellation Energy, GE Vernova), and logistics (Symbotic). Sample sector portfolios and the investment philosophies of Philip Fisher, Cathie Wood, Sir John Templeton, and Peter Lynch close the chapter. The takeaway: diversify across sectors, and remember that even the best company is a bad investment if you overpay for it.

Metric	Detail
Revenue (2026E)	~$4B+ (transaction-based + subscription/Gold)
Funded Customers	~25M+
Gold Subscribers	~3M+ (high-margin recurring revenue)
Assets Under Custody	~$200B+
AI Moat	Behavioral data — Robinhood knows how retail investors trade better than anyone
Key AI Product	Gold Card underwriting + AI-powered trade recommendations

Factor	Traditional Tech	Healthcare AI
Revenue model	Subscriptions, ads	Drug sales, licensing, cost savings
Regulation	Minimal	FDA, HIPAA — massive moats
Sales cycle	Weeks to months	Years — but contracts are sticky
Data moat	Moderate	Extreme — medical data is the most valuable data on earth
Market size	~$5T (tech)	~$12T (global healthcare spend)

How to Buy AI Stocks — A Complete Guide for US Investors

This chapter is a practical, step-by-step guide to buying AI stocks as a US retail investor. It compares brokerages — Fidelity, Charles Schwab, and Vanguard for free trades and fractional shares; Robinhood, M1 Finance, and SoFi Invest for mobile-first, automated investing; and Interactive Brokers for professional and international access. It lays out three purchase methods: buying individual stocks directly (with a table of tickers including NVDA, MSFT, AVGO, TSM as an ADR, VRT, and EQIX), buying ETFs (QQQ at 0.20%, SMH, AIQ), and buying fractional shares. A $10,000 scenario shows first-year costs are almost identical across brokerages, around $27 in ETF fees. It explains dollar-cost averaging, sets up automatic investing per broker, and covers tax-advantaged accounts — Roth IRA ($7,000 limit), 401(k) ($23,500), and the triple-tax-free HSA ($4,300). Five common mistakes and a four-week action plan follow. The takeaway: Fidelity for most people, M1 for automation, and never time the market.

Company	Ticker	Type	Typical Price Range (2025)
NVIDIA	NVDA	US Common Stock	$100–$150
Microsoft	MSFT	US Common Stock	$400–$500
Broadcom	AVGO	US Common Stock	$200–$250
AMD	AMD	US Common Stock	$100–$150
Taiwan Semiconductor	TSM	ADR (NYSE)	$180–$220
ASML Holding	ASML	ADR (Nasdaq)	$700–$900
Vertiv Holdings	VRT	US Common Stock	$100–$140
Equinix	EQIX	US Common Stock (REIT)	$900–$1,100
Arista Networks	ANET	US Common Stock	$400–$500
CrowdStrike	CRWD	US Common Stock	$350–$450

Topic	Key Takeaway
Best brokerage for most people	Fidelity — free trades, fractional shares, automatic investing, excellent all-around
Best for set-and-forget DCA	M1 Finance — pies + auto-deposit = true automation
Cheapest index ETFs	VOO (0.03%), VGT (0.09%), QQQM (0.15%)
Best AI-specific ETF	SMH (0.35%) — concentrated semiconductor exposure
Cost comparison	All major US brokerages: ~$0 commissions, $27/year in ETF fees for a $10K portfolio
Fractional shares	Available at Fidelity, Schwab, Robinhood, M1 — buy $50 of NVIDIA
Best retirement account	Roth IRA — tax-free growth on AI stocks for decades
DCA strategy	Weekly automatic investing into QQQ + VOO + SMH — never time the market
Hidden superpower	HSA — triple tax-free, invest in AI stocks, pay healthcare in retirement tax-free
Biggest mistake	Trading too much — set DCA, check quarterly, ignore the noise

10 Things Every AI Investor Should Remember — Final Words

The closing chapter distills the book into ten lessons, each tied to a major thinker. It opens by citing AI's adoption speed — from the telephone's 75 years to 100 million users down to ChatGPT's two months — and Peter Drucker's warning against yesterday's logic. David Swensen's advice frames layer selection by personality and time horizon. Peter Lynch's PEG ratio is called indispensable, noting NVIDIA traded at a 70+ P/E in 2023 when 100%+ earnings growth produced a PEG of 0.7. Buffett's pick-and-shovel principle points to NVIDIA, Vertiv, TSMC, and Equinix as the bottlenecks. Nassim Taleb anchors the risk discussion, and John Bogle makes the case for dollar-cost averaging over market timing. Harry Markowitz covers diversification, Charlie Munger a 30-minute weekly routine, and Philip Fisher's four sell signals close the practical lessons. Zuckerberg's "start before you're ready" ends the book. The takeaway: build a pick-and-shovel core, diversify, dollar-cost average, and act today rather than waiting.

1. AI Is the Biggest Megatrend of Our Era — Treat It That Way