Dorlexa Score™
A proprietary Quality-Momentum-Bottleneck framework that evaluates AI companies across 6 weighted dimensions and a unique infrastructure bottleneck analysis inspired by Leopold Aschenbrenner's research.
What is Dorlexa Score?
The Dorlexa Score is a proprietary composite rating that evaluates AI-related public equities across multiple weighted dimensions. Each dimension captures a distinct aspect of fundamental quality, competitive positioning, or market dynamics. The final score represents an objective, repeatable assessment designed to surface the highest-conviction opportunities in the AI investment landscape.
Our systematic, quantitative approach combines fundamental analysis with market signals and proprietary competitive intelligence. The model is continuously refined to ensure it captures the factors that drive long-term returns in the AI sector.
Scores are updated weekly using the latest available financial data, analyst estimates, and market signals.
The 6 Dimensions
Each dimension is independently scored and weighted to produce a holistic assessment.
AI Leverage
How deeply is AI embedded in the company's core business? We separate genuine AI infrastructure leaders from companies simply riding the narrative. Companies with deep AI integration — where AI is the product, not just a feature — score highest. This dimension is the foundation of our AI-focused investment thesis.
Growth Quality
Not all growth is created equal. We identify companies with sustainable, high-quality growth rather than unsustainable top-line acceleration. Durable compounders with expanding margins and strong returns on equity are rewarded — flash-in-the-pan revenue spikes are not.
Profitability
Profitability is the hallmark of a durable business. We evaluate whether companies convert revenue into real economic value through superior unit economics, strong free cash flow generation, and consistent margin expansion. Companies that burn cash to grow receive lower scores.
Competitive Moat
We assess market leadership, scale advantages, and whether the company controls a critical position in the AI value chain that is difficult to replicate or bypass. Companies with pricing power, high switching costs, and network effects score highest. This dimension is amplified by our proprietary Bottleneck Analysis.
Valuation
A disciplined check on price ensures the model doesn't chase overextended names. We assess whether current market pricing fairly reflects the company's quality and growth profile. The best AI companies often deserve premium valuations — but excess is penalized.
Momentum & Trend
Price action reflects institutional conviction and capital flow. We capture multi-timeframe trend strength, moving average regimes, and volatility-adjusted momentum. This dimension acts as a confirmation signal — it validates that the market agrees with the fundamental thesis. Our model includes safeguards against momentum traps: stocks with extended rallies but deteriorating fundamentals are dampened.
Scoring Approach
Each dimension is broken down into proprietary sub-factors and deep analysis layers. The exact weightings, formulas, and signal combinations are part of our intellectual property and are continuously refined based on new research and market conditions.
Pro subscribers get full access to per-stock dimension breakdowns, showing exactly how each company scores across all 6 dimensions and where its strengths and weaknesses lie.
Our Edge
Most AI stock analysis focuses on surface-level metrics — revenue growth, price momentum, or analyst ratings in isolation. Dorlexa Score takes a fundamentally different approach by evaluating companies through multiple complementary lenses simultaneously.
Our research focuses on value chain dynamics and competitive positioning. We believe the companies that will generate the most durable returns are those that control critical positions in the AI infrastructure stack — not necessarily the ones building the flashiest applications.
The model rewards companies that demonstrate strength across multiple dimensions rather than excelling in just one area. This multi-factor approach has consistently outperformed single-signal strategies.
The Bottleneck Thesis
Beyond the 6-dimension quantitative model, every stock in our universe is evaluated through the lens of infrastructure bottleneck analysis. Inspired by Leopold Aschenbrenner's research on AI supply chain dynamics, this framework identifies which companies control the critical chokepoints in the AI value chain — the irreplaceable nodes where demand vastly exceeds supply and switching costs are prohibitive.
Why Bottlenecks Matter
The AI buildout is the largest infrastructure deployment in history. But it is not uniform — it flows through a small number of critical chokepoints. Companies that control these bottlenecks capture outsized economics: pricing power, margin expansion, and durable demand regardless of the competitive intensity downstream.
Consider the AI compute stack: every training run and every inference call requires GPUs (NVIDIA), which are manufactured on leading-edge nodes (TSMC), using EUV machines (ASML), designed with EDA tools (Synopsys/Cadence), and connected via high-speed networking (Arista). Remove any one link and the entire chain breaks.
The bottleneck score is not a standalone signal — it is integrated into the Competitive Moat dimension and acts as a qualitative overlay on the quantitative scoring model. A high bottleneck score amplifies a strong quantitative score; it does not rescue a fundamentally weak company.
Monopoly / Irreplaceable
Companies that control a chokepoint with zero viable alternatives. If they stopped producing tomorrow, the entire AI buildout would halt. These are the rarest and most valuable positions in the stack.
High Bottleneck
Companies controlling critical infrastructure with very limited competition — duopolies or oligopolies in their niche. Switching costs are extremely high and capacity constraints persist.
Medium Bottleneck
Companies with strong but not exclusive positions. They control important infrastructure but face viable competition.
Low / No Bottleneck
Companies with competitive advantages but operating in markets with alternatives. They may benefit from AI trends but do not control critical infrastructure.
Specific tier assignments and scores for each stock are available to Pro subscribers.
Model Performance
20 years of model performance (Q1 2006 – Q1 2026) — the Dorlexa Score model vs the S&P 500.
+3,885%
Dorlexa Model Return
+451%
S&P 500 Return (price only, excl. dividends)
+20.2%
Annualized Return
+11.3pp
Annualized Alpha
0.92 / 0.89
Sharpe / Sortino
81%
Annual Win Rate vs S&P
Performance Methodology
Performance was measured over a 20-year historical period (Q1 2006 – Q1 2026) using the same scoring rules applied to current data. The model portfolio was rebalanced quarterly, selecting the 20 highest-ranked stocks from the Dorlexa Score model using equal weighting (5% per position). Momentum signals were calculated using only price data available at each rebalance date. Transaction costs were not modeled.
Data Sources
Financial data is sourced from institutional-grade data providers. Market signals, analyst estimates, and sentiment indicators are aggregated from multiple industry-standard sources. All scoring is updated on a weekly cadence to reflect the latest available information.
See the Model in Action
Explore the full AI model and see how AI stocks score across all dimensions.
Disclaimer
Dorlexa Score is for informational and educational purposes only. It does not constitute financial advice, investment recommendations, or an offer to buy or sell any securities. Past performance does not guarantee future results. The performance figures presented are based on historical data and do not represent live trading performance. Transaction costs, slippage, and market impact were not modeled. Always conduct your own research and consult a licensed financial advisor before making investment decisions.