By Lynn Räbsamen, CFA | COO, Global Swiss Learning | Advisory Board Member, CFA Institute | Author, Artificial Stupelligence
In a live experiment testing whether artificial intelligence can beat the market it’s built to power, two rival AI models constructed radically different portfolios. The backtest results reveal a stunning winner. Now we’re tracking them for a year to see if silicon beats strategy.
The question isn’t new, but the stakes have never been higher. The technology that promises to revolutionize everything from healthcare to transportation faces its ultimate test:
Can AI beat its own game by investing in itself?
On January 14, 2026, we launched an unprecedented experiment. We gave OpenAI’s ChatGPT and Anthropic’s Claude the identical challenge: construct a portfolio of AI infrastructure stocks positioned to outperform over 1-3 years using a “picks and shovels” strategy. Both models had the same mandate to focus on companies selling the critical infrastructure enabling the AI revolution.
The result? Two completely different portfolios—one diversified across 15 stocks, the other concentrated in just 10. One betting heavily on cloud hyperscalers, the other on pure semiconductor and networking infrastructure. One risk-balanced across the full AI value chain, the other conviction-weighted in bottleneck technologies.
When backtested over the past three years, both portfolios crushed the benchmark by roughly 3x. But Claude’s concentrated infrastructure play outperformed ChatGPT’s diversified approach by a striking 5.28% annually—turning $100,000 into $628,211 versus $519,294, a $108,917 wealth gap.
The question now: Was Claude’s victory luck, or genuine strategic superiority?
Over the next twelve months, we’ll find out.
The Setup: Same Question, Different Answers
Both AI models received an identical prompt. The constraints: 10-20 listed companies, high growth/high volatility tolerance, no geographic restrictions, $100,000 notional allocation.
What emerged were two fundamentally different investment philosophies.
Portfolio allocation comparison reveals distinct strategic differences: ChatGPT diversified across 15 stocks including all major cloud platforms (Alphabet, Meta, Oracle) and speculative plays (C3.ai, Alibaba), while Claude concentrated in 10 pure infrastructure stocks, uniquely including Vertiv (12%), Marvell (6%), and Micron (3%) while completely excluding Alphabet and Meta.
ChatGPT’s Approach: Diversified Value Chain Coverage
ChatGPT constructed a 15-stock portfolio with a pyramid allocation structure, concentrating 55% in the four largest hyperscalers and spreading the remainder across the full AI infrastructure stack.
ChatGPT’s Portfolio:
| Stock | Weight | Rationale |
|---|---|---|
| NVIDIA | 18% | Quintessential AI “pick-and-shovel” with explosive data-center GPU revenue growth and CUDA software moat. |
| Microsoft | 14% | Azure cloud leader accelerating to 40% growth from AI services, OpenAI strategic partnership. |
| Amazon | 12% | AWS re-accelerating on AI workloads, $125B capex, custom Trainium/Inferentia chips scaling. |
| Alphabet | 11% | Google Cloud 34% growth, TPU infrastructure, 9 of top 10 AI research labs as customers. |
| AMD | 7% | MI300 AI accelerators gaining traction, 122% YoY data-center revenue growth. |
| TSMC | 7% | World’s leading foundry manufacturing all major AI chips, CoWoS packaging capacity doubled. |
| ASML | 6% | EUV lithography monopoly with €36B backlog, essential for advanced semiconductor nodes. |
| Broadcom | 5% | Custom AI ASICs for hyperscalers, multi-year Google TPU deal, AI networking leader. |
| Meta | 4% | $60-80B AI capex, Llama open-source leadership, AI-driven engagement improvements. |
| Equinix | 4% | Global data center REIT benefiting from AI-driven capacity expansion needs. |
| Arista Networks | 4% | High-performance networking for cloud data centers, AI cluster fabric specialist. |
| Oracle | 3% | OCI cloud infrastructure growing 55% YoY, Nvidia GPU clusters for AI workloads. |
| Palantir | 2% | AIP platform driving 71% U.S. commercial growth, defense AI contracts. |
| Alibaba | 2% | China cloud market leader with AI model Tongyi Qianwen and regulatory tailwinds. |
| C3.ai | 1% | Pure-play enterprise AI software, positioned as high-upside “moonshot” bet. |
Philosophy: ChatGPT framed its approach as capturing “every major cloud provider” while maintaining exposure across semiconductors, networking, data centers, and software. The model emphasized diversification as risk management, spreading allocations to ensure no single point of failure.
Key thesis: “By providing the ‘picks and shovels’ of the AI revolution, these companies should enjoy robust demand even if certain end-use AI applications or startups falter”.
The portfolio allocated 55% to the top four hyperscalers (NVIDIA, Microsoft, Amazon, Alphabet), betting that platforms building and monetizing AI infrastructure would capture the lion’s share of value. ChatGPT explicitly included Meta, Oracle, and Alphabet despite them being infrastructure consumers rather than pure sellers, reasoning that their AI investments would drive engagement, revenue, and ultimately stock performance.
Claude’s Approach: Concentrated Bottleneck Dominance
Claude took a radically different path: a 10-stock portfolio with equal weighting across four infrastructure leaders at 13% each, and concentrated exposure to networking and power infrastructure.
Claude’s Portfolio:
| Stock | Weight | Rationale |
|---|---|---|
| NVIDIA | 13% | Core AI accelerator supplier with dominant GPU share, capturing most AI training and inference spend. |
| Microsoft | 13% | Hyperscaler with massive AI capex and deep integration of AI across Azure and enterprise products. |
| TSMC | 13% | Critical advanced-node foundry manufacturing nearly all leading AI chips, a central bottleneck in supply. |
| Broadcom | 13% | Key provider of custom AI silicon and high-speed networking chips for hyperscaler data centers. |
| Arista Networks | 12% | Leading cloud networking vendor enabling large-scale AI clusters with high-speed Ethernet fabrics. |
| Vertiv Holdings | 12% | Power and cooling specialist for AI data centers, leveraged to rising density and liquid cooling adoption. |
| ASML | 8% | Near-monopoly supplier of EUV tools required to produce the most advanced AI semiconductors. |
| Amazon | 7% | AWS hyperscaler investing heavily in AI infrastructure and custom chips like Trainium and Inferentia. |
| Marvell | 6% | Designer of custom ASICs and optical networking solutions used in next‑gen AI data center architectures. |
| Micron | 3% | High‑bandwidth memory provider benefiting from AI-driven HBM shortages and strong pricing power. |
Philosophy: Claude organized its portfolio around infrastructure bottlenecks—the scarce, essential technologies where demand vastly exceeds supply. Rather than diversifying across the value chain, Claude concentrated in companies with:
- Non-substitutable technology (TSMC, ASML, Micron HBM)
- Multi-year order backlogs providing revenue visibility
- Pricing power from supply constraints
- Direct exposure to $600B hyperscaler capex cycle
The model’s most distinctive choices were Vertiv Holdings (12%), a power and cooling infrastructure provider not included in ChatGPT’s portfolio, and elevated allocations to TSMC (13%) and Broadcom (13%) versus ChatGPT’s 7% and 5% respectively.
Claude completely excluded seven stocks ChatGPT included: Alphabet, Meta, Oracle, Equinix, Palantir, Alibaba, and C3.ai. The rationale: “Stayed pure to ‘shovels’ thesis—avoided platforms that consume infrastructure versus sell it”.
Claude’s equal-weighting of the top four holdings at 13% each reflected what it called “high conviction without over-concentration,” avoiding ChatGPT’s 18% single-stock exposure to NVIDIA.
The Backtest: Claude Dominates
When both portfolios were backtested over the past three years using Portfolio Visualizer, the results were decisive:
Three-year backtest results show Claude’s concentrated 10-stock portfolio outperformed ChatGPT’s diversified 15-stock portfolio by 5.28% annually, while both AI-constructed portfolios crushed the AIQ benchmark by roughly 3x. Despite similar volatility levels, Claude achieved superior risk-adjusted returns with a Sharpe ratio of 1.23 versus ChatGPT’s 1.11.
Performance Summary (3-Year Backtest):
| Metric | ChatGPT Portfolio | Claude Portfolio | AIQ Benchmark |
|---|---|---|---|
| Final Value ($100K start) | $519,294 | $628,211 | $195,899 |
| Annualized Return (CAGR) | 38.27% | 43.55% | 14.14% |
| Standard Deviation | 31.01% | 31.26% | 20.86% |
| Best Year | 106.22% | 112.92% | 55.39% |
| Worst Year | -38.77% | -34.57% | -36.45% |
| Maximum Drawdown | -43.86% | -41.88% | -39.63% |
| Sharpe Ratio | 1.11 | 1.23 | 0.59 |
| Sortino Ratio | 1.99 | 2.17 | 0.94 |
| Correlation to AIQ | 0.88 | 0.86 | 1.00 |
Key Findings:
1. Claude Won Decisively
Claude’s portfolio generated $628,211 from a $100,000 investment versus ChatGPT’s $519,294—a $108,917 wealth gap (20.9% more final capital). The 43.55% annualized return beat ChatGPT’s 38.27% by 5.28 percentage points annually.
2. Both Crushed the Benchmark
While AIQ returned a respectable 14.14% annually, both AI portfolios delivered roughly 3x the benchmark performance—turning $100K into $196K (AIQ) versus $519K (ChatGPT) and $628K (Claude).
3. Similar Volatility, Better Risk-Adjusted Returns
Despite nearly identical standard deviation (31.01% vs 31.26%), Claude achieved superior risk-adjusted performance with a Sharpe ratio of 1.23 versus ChatGPT’s 1.11. This indicates Claude generated more return per unit of risk.
4. Better Downside Protection
Claude’s worst year was -34.57% versus ChatGPT’s -38.77%, and its maximum drawdown was -41.88% versus -43.86%. The concentrated portfolio paradoxically offered better downside protection despite holding fewer stocks.
5. Lower Benchmark Correlation
Claude’s 0.86 correlation versus ChatGPT’s 0.88 suggests its portfolio was more differentiated from the broader AI sector—a factor that could prove advantageous if sector momentum reverses.
Why Claude Won: The Strategic Differences That Mattered
The backtest results weren’t random luck. Three strategic choices explain Claude’s outperformance:
1. Vertiv Holdings: The Hidden Infrastructure Play (12% weight)
Claude’s single most distinctive pick was Vertiv (VRT), a power and cooling infrastructure provider completely absent from ChatGPT’s portfolio.
As AI workloads exploded, data center thermal management became as critical as compute. NVIDIA’s Blackwell and upcoming Rubin GPU architectures generate unprecedented heat density, requiring liquid-to-chip cooling solutions—exactly Vertiv’s specialty.
Key catalysts Claude identified:
- $9.5 billion order backlog providing multi-year revenue visibility.
- Liquid cooling transition: Air cooling insufficient for 100kW+ racks; Vertiv’s CoolLoop systems address this
- NVIDIA partnership: Co-developed 800V HVDC data centers and liquid cooling systems, with a $5M Department of Energy grant.
- 50%+ EPS growth projected over two years as AI density scales
- Market expansion: Air cooling + heat rejection represents 70% of market ($30B+), liquid cooling 30% with 30% CAGR.
Vertiv’s stock reflects this thesis. The company reported that roughly 80% of sales come from data centers, with management stating “we’re sold out” for AI-related infrastructure through 2026.
ChatGPT’s omission of Vertiv represented a fundamental analytical gap: recognizing that power and cooling are now bottlenecks equal to semiconductors.
2. TSMC Overweight: Manufacturing as the Ultimate Chokepoint (13% vs 7%)
Claude allocated 13% to TSMC versus ChatGPT’s 7%—nearly double the weight. This reflected a crucial insight: TSMC is the foundry for virtually all cutting-edge AI chips.
NVIDIA’s GPUs, AMD’s MI300, Apple’s M-series, Google’s TPUs, Amazon’s Trainium, Microsoft’s Maia—all manufactured by TSMC. The company’s advanced packaging technology (CoWoS) is the physical constraint on AI accelerator supply.
Claude highlighted:
- CoWoS capacity doubled from 35K to 70K wafers/month in 2024-2025, yet still sold out through 2026
- 12x shipment growth in AI-related chips from 2021 to 2024
- Premium pricing power: TSMC announced 5-10% price increases for advanced nodes in 2026
- Geographic risk priced in: Despite Taiwan concentration, no credible alternative exists
By contrast, ChatGPT treated TSMC as one of many infrastructure providers rather than the irreplaceable manufacturing bottleneck.
3. Networking Dominance: Arista + Broadcom = 25% (vs ChatGPT’s 9%)
Claude allocated 25% combined to Arista Networks (12%) and Broadcom (13%), versus ChatGPT’s 9% (4% + 5%). This reflected the thesis that 2026 marks the networking bottleneck era.
As AI clusters scale from thousands to millions of GPUs, networking bandwidth becomes as critical as compute power. Broadcom’s Ethernet switching ASICs and Arista’s cloud networking platforms form the data fabric connecting GPU clusters.
Claude’s thesis:
- Broadcom: $73B AI-related backlog, custom ASIC dominance for hyperscaler TPUs, transition from proprietary to open Ethernet benefits Broadcom
- Arista: $10B revenue target for 2026, 800G/1.6T optical transitions, 42.7% EBITDA margins
- Networking shift: Over 55% of AI infrastructure spending now dedicated to running models in production rather than training—driving networking demand
ChatGPT recognized networking’s importance but under-allocated relative to the bottleneck severity.
4. Memory Shortage: Micron’s Pricing Power (3% weight)
Claude included Micron (MU) at 3%; ChatGPT excluded it entirely.
High-bandwidth memory (HBM) has become the scarcest component in AI infrastructure. Micron’s HBM3E memory stacks are essential for NVIDIA and AMD GPUs, and the company faces unprecedented demand.
Critical datapoints:
- Sold out through 2026: Micron can meet only two-thirds of medium-term client needs
- 3:1 production ratio: Every HBM unit requires forgoing 3 units of standard memory
- 60%+ price surge: HBM spot prices jumped 60% in six months, with forecasts for doubling year-over-year
- $500 module pricing: Up from ~$250 a year ago
- Premium margins: Memory prices at all-time highs driving profitability surge
While 3% seems modest, Micron’s inclusion captured memory as a structural bottleneck—a thesis ChatGPT’s portfolio completely missed.
5. Pure Infrastructure vs Platform Mix
Claude’s most fundamental strategic choice: exclude all cloud platforms that consume rather than sell infrastructure.
| Excluded by Claude | ChatGPT Weight | Claude’s Rationale |
|---|---|---|
| Alphabet | 11% | “Consumes infrastructure for search/ads rather than pure seller” |
| Meta | 4% | “Massive AI capex but not selling shovels to others” |
| Oracle | 3% | “Cloud player with execution risk” |
| Equinix | 4% | “Data center REIT—lower growth than chip/networking plays” |
| Palantir | 2% | “Application layer, not infrastructure” |
| Alibaba | 2% | “China regulatory risk, not pure infrastructure” |
| C3.ai | 1% | “Speculative software play” |
This philosophical purity meant Claude avoided diluting returns with companies whose AI infrastructure spending costs them money (building data centers) rather than making them money (selling chips, networking gear, cooling systems).
ChatGPT’s inclusion of these stocks reflected a “full value chain” philosophy, but the backtest suggests concentration in bottleneck infrastructure beats broad diversification.
The “Picks and Shovels” Thesis: Why It Works
Both portfolios were built on the same historical insight: during the California Gold Rush of 1849, Levi Strauss made a fortune selling blue jeans to miners—a more reliable path to wealth than prospecting for gold.
In the AI context, this translates to a crucial observation: while the world debates whether OpenAI, Anthropic, Google, or some unknown startup will dominate consumer AI, all of them need the same fundamental infrastructure.
The Structural Tailwind: $600 Billion Hyperscaler Capex
Both AI bots built their portfolios around an undeniable trend: hyperscaler capital expenditure exceeding $600 billion in 2026, a 36% increase year-over-year, with approximately $450 billion (75%) directed specifically at AI infrastructure.
Individual company breakdowns:
- Amazon: $125B+ capex, with AWS (Amazon Web Services) adding 3.8 GW capacity in 12 months
- Microsoft: $100B+ capex, doubling Azure AI infrastructure
- Google: $100B+ capex for cloud and TPU (Tensor Processing Unit) deployment
- Meta: $60-80B capex for AI supercomputers and custom chips
- Oracle: ~$20B capex for OCI (Oracle Cloud Infrastructure) AI cloud expansion
This represents capital intensity of 45-57% of revenue—levels historically unprecedented for technology companies. Amazon AWS’s 57% capital intensity, Meta’s 52%, and Microsoft’s 48% reflect an industry in full-scale infrastructure buildout mode.
Goldman Sachs projects that by 2030, global AI infrastructure spending will reach $580 billion annually, surpassing oil supply investment. The IEA forecasts data center capex alone will exceed $580 billion by 2030.
The Three Critical Bottlenecks
Both AI models identified supply-demand imbalances creating monopoly-like pricing power:
1. Advanced Chip Manufacturing
TSMC’s advanced packaging capacity doubled to 70,000 wafers per month yet remains sold out through 2026. NVIDIA alone represents $180 billion in GPU spending in 2026, capturing 90% of AI accelerator demand.
2. High-Bandwidth Memory
Micron’s HBM3E is the scarcest component, with the company stating “we’re sold out for 2026”. The 3:1 production ratio (one HBM unit requires sacrificing three standard memory units) creates structural scarcity. Memory spot prices surged 60% in six months, with forecasts for doubling year-over-year.
3. Networking Infrastructure
Million-GPU clusters require massive networking bandwidth. Broadcom’s $73B AI backlog and Arista’s $10B revenue target reflect networking becoming “as critical as compute”. The shift from training (compute-intensive) to inference (networking-intensive) is accelerating this trend.
Why Infrastructure Beats Applications
BlackRock’s Chief Investment Strategist Ben Powell articulated the investment case at the 2025 Abu Dhabi Finance Week: “Companies providing essential resources—ranging from semiconductor manufacturers to energy suppliers—are likely to enjoy more consistent growth compared to those developing AI models”.
The logic is straightforward:
- Application winners are uncertain: Will OpenAI, Anthropic, Google, or an unknown startup dominate?
- Infrastructure winners are certain: Regardless of which application wins, they all need NVIDIA GPUs, TSMC manufacturing, Broadcom networking, Vertiv cooling, and Micron memory
This reduces binary outcome risk while maintaining exposure to AI’s explosive growth.
Can AI Really Beat the Market?
The timing of this experiment coincides with explosive debate about AI’s investment capabilities.
The Optimistic Case: Stanford’s 600% Outperformance
A Stanford University study published June 2025 found that an AI analyst making stock picks using only public information beat 93% of mutual fund managers over 30 years by an average of 600%.
Between 1990 and 2020, human fund managers generated $2.8 million of alpha per quarter. When AI readjusted their portfolios, it generated $17.1 million per quarter on top of actual returns—a 6x improvement.
“It was stunning,” said researcher Ed deHaan. “AI beat 93% of managers over a 30-year period by an average of 600%”.
The study attributed AI’s advantage to:
- Elimination of behavioral biases (overconfidence, anchoring, loss aversion)
- Processing speed: Analyzing thousands of data points humans can’t track
- Pattern recognition: Identifying correlations across disparate datasets
- Consistency: No “off days” or emotional reactions
The Skeptical Case: Black Swans and Real-World Chaos
Yet skepticism abounds. A 2025 Nature study found that popular deep neural network models for stock chart analysis show only minimal predictive power in real-world noisy markets.
Traders Magazine noted that “when massive market fluctuations occur, operators are forced to switch off the algorithms and allow human traders to take over”. AI excels in stable regimes but fails during regime changes where historical patterns break down.
MarketWatch reported that despite many ETFs and professionals using AI, most still don’t outperform their benchmark. As more firms adopt AI, the mathematical reality of active management ensures “the typical manager will fall short of market performance”.
The consensus: AI works best as a “co-pilot” rather than standalone solution. While AI excels at data processing and pattern recognition, it struggles with:
- Black swan events (pandemics, wars, policy shocks) absent from training data
- Reflexivity: Algorithmic trading itself influences prices, creating feedback loops
- Non-quantifiable factors: Management quality, culture, regulatory changes
- Narrative shifts: When markets re-rate entire sectors (dot-com bust, 2008 crisis)
What Makes This Experiment Different
Unlike academic backtests or proprietary strategies, this experiment has distinguishing features:
1. Full Transparency
Complete portfolio composition, weights, and rationale are public from inception.
2. Real-World Constraints
No rebalancing, no leverage, no short positions—actual conditions individual investors face.
3. Head-to-Head AI Competition
Two state-of-the-art models from rival companies with different training and architectures.
4. Infrastructure Focus
Rather than predicting application winners, both portfolios bet on infrastructure—reducing binary risk.
5. Structural Tailwinds
Launched into a $600B hyperscaler capex cycle with multi-year visibility.
6. Backtested Validation
Unlike typical “launch and pray,” we have three years of historical evidence suggesting the strategies work.
The Year Ahead: What We’re Tracking
Over the next twelve months (January 14, 2026 – January 14, 2027), we’ll monitor several dimensions:
Performance Metrics
- Absolute returns: Growth from $100,000 starting value
- Relative returns: ChatGPT vs Claude vs AIQ benchmark
- Risk-adjusted returns: Sharpe and Sortino ratios
- Drawdown behavior: How portfolios handle corrections
- Correlation dynamics: Do portfolios remain differentiated from benchmark?
Portfolio Evolution (No Rebalancing)
- Concentration drift: Will top holdings dominate even more?
- Winner-loser divergence: Which stocks significantly over/underperform?
- Position sizing validation: Were Claude’s 13% equal-weights superior to ChatGPT’s 18% pyramid?
The Deeper Questions This Experiment Explores
Beyond mere performance, this experiment probes fundamental questions about artificial intelligence and investing:
1. Can AI exhibit foresight or just pattern-matching?
If Claude’s portfolio continues outperforming, is it because the model identified genuine bottlenecks others missed—or because it overfitted to recent data that happens to extrapolate?
2. Does concentration beat diversification in asymmetric markets?
Claude’s 10 stocks beat ChatGPT’s 15, both beat AIQ’s 90. Does this validate “concentrated conviction” in infrastructure, or does it simply reflect a benign backtest period?
3. Is strategic disagreement more valuable than consensus?
The two AIs diverged dramatically—ChatGPT favoring platforms, Claude favoring bottlenecks. Does this diversity of machine perspective offer edge, or confusion?
4. What happens when everyone knows the trade?
The “picks and shovels” AI infrastructure thesis is hardly secret. If backtest performance attracted capital, will forward returns disappoint as valuations expand?
5. Can static portfolios compete with dynamic management?
Real fund managers adjust to new information. These portfolios are frozen. Is “set and forget” viable when the AI landscape evolves rapidly?
The Stakes: More Than Returns
This experiment matters beyond investment performance. It’s a stress test of whether artificial intelligence can systematically generate alpha using publicly available information—a question with profound implications for the $50+ trillion global asset management industry.
If AI can consistently beat markets using public data, active management fees become indefensible.
Why pay 1-2% for human fund managers when algorithms deliver superior risk-adjusted returns at minimal cost?
But if AI merely reflects consensus wrapped in technological sophistication, it becomes another tool—useful for efficiency, not edge.
The fundamental challenge: the Efficient Market Hypothesis, which contends all public information is already priced into stocks. AI models are trained on this same public data—financial reports, news, analyst narratives. If “AI infrastructure bottlenecks” became consensus in 2025 media, Claude and ChatGPT may have simply identified what sophisticated investors already know. If that holds true, their outperformance should fade going forward as the trade becomes fully priced.
Conclusion: Let the Battle Begin
Two artificial intelligence models. Two radically different strategies. One clear backtest winner.
ChatGPT wagered that diversification across the full AI value chain—from semiconductors to cloud platforms to applications—would capture upside while managing risk. Its 15-stock portfolio includes every major hyperscaler and spreads exposure across infrastructure, platforms, and software.
Claude bet that concentration in irreplaceable bottlenecks—manufacturing, networking, power, memory—would generate superior returns. Its 10-stock portfolio excludes cloud platforms entirely, focusing purely on companies selling shovels rather than digging for gold.
The three-year backtest crowned Claude the winner: 43.55% annualized returns versus ChatGPT’s 38.27%, turning $100,000 into $628,211 versus $519,294. Both crushed the 14.14% AIQ benchmark by roughly 3x.
But past performance doesn’t guarantee future results.
The forward year will reveal whether:
✓ Claude’s bottleneck thesis continues as networking, cooling, and memory remain scarce
✓ ChatGPT’s diversification provides better downside protection during corrections
✓ Concentration risk in Claude’s portfolio becomes a liability if few stocks stumble
✓ Platform value in ChatGPT’s Alphabet/Meta/Oracle holdings drives surprise outperformance
✓ AI infrastructure spending sustains, accelerates, or reverses
Most fundamentally, we’ll discover whether artificial intelligence can not merely analyze the market it’s designed to power, but beat it.
The clock starts now. Over the next twelve months, we’ll track every basis point, every drawdown, every catalyst. Because in the end, the market is the ultimate judge—and it doesn’t care about elegant theses, only results.
Follow this series for quarterly updates tracking the battle of the bots. The data will tell the story. The wealth will reveal the winner.
Disclosure: This portfolio is for educational and experimental purposes only and does not constitute investment advice. Past performance does not guarantee future results. Investing in technology stocks involves substantial risk, including loss of principal. Readers should conduct their own research and consult financial advisors before making investment decisions.
For more insights about what AI can or cannot do, check out my book “Artificial Stupelligence: The Hilarious Truth About AI”.
