The AGI Portfolio
Everybody Has a Timeline. Few Have a Portfolio to Match.
Metaculus puts 25% probability on AGI by 2029. Sam Altman says 2029. Dario Amodei hints 2026. Demis Hassabis: 50% by decade's end. Forecasters converge on late 2020s.
Yet median portfolios still look built for 2015: 60/40 stocks/bonds. Maybe handful of tech stocks. Almost nobody structured around AGI arriving within one presidential term.
This essay models three scenarios with different arrival dates, shows what each does to sectors and companies, and lets you decide portfolio composition.
Three AGI Timelines, Three Different Worlds
Rather than debate AGI arrival, assign probabilities and build accordingly. Three scenarios based on expert predictions:
Who Wins in Each Scenario
Companies benefiting from AGI span four layers: compute, infrastructure, applications, energy. Each carries different risk by timeline.
Six largest companies are all deep in AI supply chain. Market has priced substantial AI upside. Question: enough, or too much?
| Layer | Company | AGI Bull Case | AGI Bear Case | Key Risk |
|---|---|---|---|---|
| Compute | NVIDIA | Dominant supplier | Custom chips erode | AMD, Google TPUs, in-house ASICs |
| Compute | AMD | Gains share lower | Distant second | Software ecosystem gap |
| Infrastructure | Microsoft | OpenAI integration | $13B bet underperforms | Concentration risk |
| Infrastructure | Alphabet | DeepMind + Cloud | Search disrupted | Ad revenue cannibalization |
| Application | Meta | Open-source moat | Social loses relevance | Metaverse capital drain |
| Application | Tesla | Optimus + FSD | Auto margins compress | Robotics execution |
| Energy | Nuclear Utilities | Data center demand | Regulatory delays | SMR timelines slip |
| Private | Anthropic | Safety-first $380B val | Constant capital need | IPO timing/pricing |
Who Gets Hurt
AGI creates winners and losers. Market hasn't fully priced losers. McKinsey: 30% of U.S. jobs automatable by 2030. Forrester: 10.4 million jobs lost. Conservative: 85% of workers see 10% of tasks handled by AI within five years.
Highest displacement risk: knowledge workers. Financial analysts, legal assistants, customer service, copywriters, mid-level managers making routine decisions.
Companies with large knowledge worker bases but no AI infrastructure face margin compression. Banks with 50,000 customer service reps vs. those with AI agents: different cost structures. Staffing companies, BPOs, consulting firms in blast radius.
AGI Needs Electricity. A Lot of It.
Data center power demand doubles: 49GW (2023) to 96GW (2026). AI accounts for ~40GW growth. By 2035, data centers alone could consume 1,300 TWh annually (more than Japan's total).
Every major tech company signing nuclear agreements. Microsoft, Google, Amazon, Meta all announced nuclear partnerships. 15 new reactors scheduled 2026. China's Linglong One small modular reactor expected commercial operations early 2026.
For AGI portfolio: even unsure which AI company wins, all need power. Nuclear utilities, transmission companies, electrical infrastructure benefit under every scenario.
Three Portfolios for Three Timelines
Each assumes $100,000 starting position reflecting different conviction levels about AGI timing. Not recommendations, frameworks for allocating based on beliefs.
| Allocation | Aggressive (2027-28) | Base Case (2029-31) | Extended (2035+) |
|---|---|---|---|
| Compute (NVIDIA, AMD) | 30% | 20% | 10% |
| AI Labs (MSFT, GOOG) | 25% | 20% | 15% |
| Energy / Nuclear | 15% | 15% | 10% |
| Robotics (Tesla, etc.) | 10% | 10% | 5% |
| IPO Reserve (Cash) | 10% | 10% | 5% |
| Broad Market (SPY) | 5% | 15% | 30% |
| Bonds / Defensive | 5% | 10% | 25% |
Aggressive: 80% AI-adjacent. Risk: overvalued growth stocks if AGI doesn't arrive 2028. Base case: balanced, stays invested if timelines slip. Extended: 55% traditional, treats AI as meaningful tilt.
What the Insiders Are Doing
Leopold Aschenbrenner, OpenAI researcher, left 2024 raising security concerns. Then raised $1.5 billion for hedge fund betting AGI coming soon.
Worth sitting with: person who spent years inside leading AI lab saw enough to bet career and $1.5 billion on aggressive timeline. Whether you agree or not, capital allocation signal is hard to ignore. Smart money is moving.
Five largest hyperscalers committed $197 billion to AI infrastructure in 2024. Google guided $175-185 billion capex for 2026. Not experimental budgets. Companies restructuring entire models around transformative AI assumption.
Four Ways This Goes Wrong
Methodology
If You Believe AGI Is Close
Overweight compute and infrastructure. Hold cash for IPOs. Accept 30-40% correction potential even if thesis right. Set trailing stops. Diversify across supply chain.
If You Think Timelines Are Extended
Keep normal portfolio with AI tilt. Favor strong non-AI businesses also benefiting from AI (Alphabet, Amazon, Microsoft). Avoid pure-play AI at speculative multiples. Be patient.
