We have mapped the 14 vectors of disruption. Now we build the fortress. 20 positions to own for the next decade.
Before allocating a single dollar, we must establish the principles that govern this portfolio. These are not suggestions. They are rules. Violating them is the difference between capturing the AI revolution and being destroyed by its volatility.
Nassim Nicholas Taleb introduced the concept of antifragility in his 2012 book: systems that gain from disorder. A robust portfolio resists shocks. An antifragile portfolio actually benefits from them.
How does this apply to AI investing? Consider: if AI adoption is slower than expected, our Energy positions (CEG, VST) still benefit from general electrification trends. If AI triggers geopolitical conflict, our Defense positions (LMT, PANW) surge. If AI causes a broad market crash, our cash buffer (10%) allows us to buy the dip on our highest-conviction names. If AI succeeds beyond expectations, our Compute and Software positions capture exponential upside.
The portfolio does not need AI to succeed to generate returns. It needs AI to continue to be attempted, which is certain, because the economic incentives are overwhelming.
Twenty positions. Six categories. One thesis: the companies that build, power, secure, and physically deploy AI will be the dominant wealth creators of the next decade. Here is the complete allocation.
| Ticker | Company | Category | Weight % | Entry Zone | Thesis | Risk |
|---|---|---|---|---|---|---|
| NVDA | NVIDIA | Compute | 5% | $110-130 | GPU monopoly. Data center revenue 80%+ of total. Blackwell cycle just beginning. | Med |
| TSM | TSMC | Compute | 4% | $160-185 | Fabricates 90%+ of advanced chips. Irreplaceable. Geopolitical risk is the discount. | Med-High |
| AVGO | Broadcom | Compute | 4% | $180-210 | Custom AI accelerators (XPUs) for hyperscalers. VMware integration. Networking dominance. | Med |
| AMD | AMD | Compute | 3% | $100-125 | MI300X gaining share. Credible #2 to NVIDIA. Server CPU dominance with EPYC. | Med-High |
| MSFT | Microsoft | Software | 5% | $380-420 | Copilot monetization across 400M Office users. Azure AI. OpenAI partnership. | Low |
| PLTR | Palantir | Software | 4% | $60-80 | AIP platform. Government + commercial AI operating system. Deep moat in classified data. | Med-High |
| CRM | Salesforce | Software | 3% | $270-310 | Agentforce: AI agents for CRM. 150K enterprise clients as distribution. Margin expansion. | Med |
| CRWD | CrowdStrike | Software | 3% | $300-350 | AI-native cybersecurity. Charlotte AI. More AI = more attack surface = more demand. | Med |
| CEG | Constellation Energy | Energy | 4% | $200-250 | Largest nuclear fleet in US. Microsoft deal for Three Mile Island restart. 24/7 clean baseload. | Med |
| VST | Vistra | Energy | 3% | $100-130 | Nuclear + natural gas fleet. ERCOT exposure. Data center PPAs. 10% free cash flow yield. | Med |
| ETN | Eaton Corp | Energy | 3% | $280-320 | Electrical infrastructure for data centers. Transformers, switchgear, UPS. Multi-year backlog. | Low-Med |
| TSLA | Tesla | Physical | 4% | $250-320 | Optimus humanoid robot. FSD. Dojo training compute. The physical AI play. | High |
| UBER | Uber Technologies | Physical | 3% | $65-80 | Autonomous vehicle deployment platform. Network effect. Mobility + delivery + freight. | Med |
| ISRG | Intuitive Surgical | Physical | 3% | $500-560 | Da Vinci surgical robots. AI-enhanced procedures. Razor/blade model with consumables. | Low-Med |
| LMT | Lockheed Martin | Defense | 3% | $450-520 | AI-enabled autonomous systems. F-35 + hypersonics. NATO spending surge. 2.7% dividend. | Low |
| PANW | Palo Alto Networks | Defense | 3% | $170-200 | Platformization. XSIAM AI security operations. Cortex. Government contracts growing 40%+ YoY. | Med |
| RHI | Robert Half (Short) | Hedge | -2% | Short above $60 | White-collar staffing in structural decline. 85% AI-exposed revenue. See Part 14. | High (short) |
| CHGG | Chegg (Short) | Hedge | -2% | Short any rally | Terminal business model. ChatGPT is free Chegg. Subscriber base in freefall. | High (short) |
| GETY | Getty Images (Short) | Hedge | -1% | Short above $3 | AI image generation makes stock photography obsolete. Library value depreciating. | Med (short) |
| SGOV/Cash | Cash / T-Bills | Cash | 10% | N/A | Dry powder for 30%+ drawdowns. Buy the dip on highest-conviction names. Earns 4-5% in T-bills. | None |
Diversification only works if assets are not perfectly correlated. The beauty of this portfolio is that while every position is connected to the AI thesis, the underlying business drivers are distinct. A chip shortage does not affect cybersecurity revenue. A regulatory crackdown on AI software does not hurt nuclear power generation. The correlation matrix below reveals the portfolio's structural resilience.
Key insight: Energy (CEG, VST, ETN) has only 0.28-0.35 correlation with Software and Compute. This means when tech sells off (which it will, repeatedly), your Energy positions act as a stabilizer. Defense (LMT, PANW) has even lower correlation with Compute at 0.25.
The Hedges bucket (short positions in RHI, CHGG, GETY) shows negative correlation with all long buckets, ranging from -0.10 to -0.55. This is by design. When AI winners rally, AI losers decline, and vice versa. The short book provides genuine portfolio insurance, not just a theoretical hedge.
A portfolio without rebalancing rules is a portfolio that will eventually blow up. Concentration risk, momentum chasing, and loss aversion are the enemies. Here are the rules that keep the portfolio antifragile.
| Trigger | Condition | Action | Frequency |
|---|---|---|---|
| Winner Trim | Any single position exceeds 15% of portfolio | Trim back to 10%. Redeploy proceeds into lagging sectors (Energy, Defense) or cash buffer. | Check monthly |
| Trailing Stop (High-Beta) | TSLA, PLTR, AMD decline 20% from trailing 52-week high | Reduce position by 50%. Re-enter on technical support confirmation or at 30% discount to target. | Weekly monitoring |
| Sector Cap | Any sector exceeds 25% of portfolio | Trim the most overweight name in the sector. Redistribute to underweight sectors. | Quarterly review |
| Cash Deployment | Portfolio drawdown exceeds 15% from peak | Deploy 50% of cash buffer into highest-conviction names (NVDA, MSFT, CEG) at pre-defined entry zones. | Event-triggered |
| Short Cover | Short position rallies 35%+ from entry | Cover immediately. Do not fight the market. Re-evaluate thesis before re-entering. | Daily stop monitoring |
| Tax-Loss Harvest | Position down 10%+ with 30+ day holding period | Sell to realize loss. Replace with correlated ETF (e.g., sell AMD, buy SMH) for 31 days to avoid wash sale. | Year-end + opportunistic |
| Thesis Invalidation | Fundamental change in company or sector outlook | Exit position entirely. Replace with next-best candidate from same category. Document reasoning. | Quarterly review |
No model portfolio is complete without stress-testing. We model three scenarios over a 5-year horizon (2026-2030), each with different assumptions about AI adoption speed, interest rates, and geopolitical stability.
| Metric | Bull Case | Base Case | Bear Case |
|---|---|---|---|
| AI Adoption Rate | Faster than expected. AGI-like capabilities by 2028. | Steady acceleration. Current trajectory continues. | Plateau. Scaling laws hit diminishing returns. |
| Interest Rates (10Y) | 3.5-4.0% (rate cuts support multiples) | 4.0-4.5% (stable) | 5.0-5.5% (inflation resurgence) |
| Geopolitics | De-escalation. US-China tech thaw. | Status quo. Controlled competition. | Taiwan crisis. Full decoupling. |
| Portfolio CAGR | +45% | +22% | -15% then recovery |
| Max Drawdown | -20% (normal correction) | -35% (sector rotation) | -55% (systemic event) |
| $100K Becomes | $640K | $270K | $160K (after recovery) |
| Key Winners | NVDA (+300%), PLTR (+400%), CEG (+200%) | MSFT (+80%), NVDA (+120%), ETN (+90%) | LMT (+40%), SGOV (+25%), Shorts profitable |
| Key Risks | Valuation bubble, then correction | Rotation headwinds, individual misses | Correlated drawdown, forced liquidation |
Hypothetical projections for illustration only. Actual results will differ. Past performance does not predict future results.
The difference between a professional portfolio and a Reddit portfolio is risk management. The thesis can be right and you can still lose money if position sizing, correlation, and drawdown management are ignored. Here are the non-negotiable constraints.
The Kelly Criterion, developed by John Kelly at Bell Labs in 1956, provides a mathematical formula for optimal bet sizing: f* = (bp - q) / b, where f* is the fraction of capital to bet, b is the odds received (payoff ratio), p is the probability of winning, and q is the probability of losing (1-p).
Example: If you believe NVDA has a 65% chance of returning 50% and a 35% chance of declining 25% over the next year, the Kelly fraction is: f* = (2.0 x 0.65 - 0.35) / 2.0 = 47.5%. But this is the theoretical optimum for a single bet. In practice, most professional investors use "Half Kelly" or "Quarter Kelly" (12-25% of the Kelly fraction) because the formula assumes perfect probability estimates, which humans never have.
Practical application: Our 5% weight for NVDA represents approximately Quarter Kelly given our confidence level. This is deliberately conservative. The portfolio can survive being wrong on 3-4 individual positions and still generate strong returns because no single position can cause catastrophic damage.
Understanding how drawdowns compound is essential. A 50% loss requires a 100% gain to break even. The table below illustrates why protecting capital is more important than maximizing returns.
Having a thesis and having a portfolio are two different things. The gap between "I should own NVDA" and actually buying it at the right price, in the right account, with the right tax treatment, is where most retail investors fail. This section bridges that gap.
The execution framework has four pillars: (1) Account structure for tax efficiency, (2) Dollar-cost averaging to reduce timing risk, (3) Order types to ensure price discipline, (4) A rebalancing calendar that forces discipline and removes emotion from decisions.
Tax efficiency can add 1-2% annually to your after-tax returns. Place assets in the right account type:
Do not deploy all capital at once. Use a 6-month DCA schedule to reduce timing risk. Divide total intended allocation into 6 equal monthly tranches. This ensures you buy at various price points and avoids the psychological impact of an immediate drawdown after a lump-sum entry.
Use limit orders only. Never use market orders for positions in this portfolio. Set limit prices at or slightly below the "Entry Zone" specified in the Singularity 20 table. If the price does not reach your limit within the DCA window, skip that month for that name and increase allocation to other names or cash. Patience is an alpha generator.
Over the course of 15 parts, we have mapped the entire AI disruption landscape. Here is what we covered:
The AI revolution will not be evenly distributed. Some companies will capture trillions in value. Others will be destroyed. The opportunity is not to predict the future with certainty; it is to position for the range of outcomes while managing risk. Stay allocated. Stay hedged. Stay curious. The best time to build this portfolio was a year ago. The second best time is today.
End of Series — AI Singularity Trade: How the World Changes Before 2030
Disclaimer: This analysis is for educational and informational purposes only. It does not constitute investment advice, a recommendation, or a solicitation to buy or sell any securities. The "Singularity 20" is a hypothetical model portfolio and should not be construed as personalized investment advice. Short selling involves unlimited risk. Options involve risk and are not suitable for all investors. Past performance does not guarantee future results. All projected returns are hypothetical. Always consult a qualified financial advisor before making investment decisions. The author may hold positions in the securities discussed.