The "Rule of 40" to be replaced by The "Rule of Data" in 2026

The "Rule of 40", a decade-long standard for SaaS valuation balancing growth and profitability, is rapidly losing its predictive power in the AI era. By 2026, it will largely be superseded by the "Rule of Data" (often operationalised as the Proprietary Data Moat).

The Driver: The commoditisation of foundation models has occurred faster than anticipated.

By late 2024 and throughout 2025, open-source models (e.g., DeepSeek, Llama variants) effectively closed the performance gap with proprietary closed-source models for most enterprise tasks. Consequently, "model advantage" is now transient.

The New Asset Class: Investors have shifted focus to the fuel rather than the engine. In 2026, valuation premiums will be assigned not to companies with the best algorithms, but to those with Proprietary, High-Fidelity and Iterate-able Data, with biological data serving as the gold standard for this new asset class.

I. The Collapse of the "Rule of 40"

For years, the Rule of 40 (Growth % + Profit Margin % ≥ 40) was the North Star for software investing. It assumed that code was the moat and that distribution (growth) was the primary driver of value.

Why it is failing in the AI Era:

Code is no longer a moat: Generative AI has lowered the barrier to entry for coding and feature development. A "feature" that once took a team of 20 engineers six months to build can now be replicated by a competitor in weeks using AI assistants and open-source libraries.

Growth without defensibility is toxic: Investors in 2025 witnessed high-growth AI wrappers implode once foundation models integrated their core features natively. Growth metrics became misleading indicators of long-term value.

Efficiency is table stakes: Profitability (the other half of Rule of 40) is now expected by default due to AI-driven operational efficiencies, making it less of a differentiator.

II. The New Metric: The "Rule of Data"

While the "Rule of 40" was a financial metric, the "Rule of Data" is a strategic defensibility metric. It evaluates the Proprietary Data Moat that prevents an AI model from being commoditised.

Components of the Rule of Data:

Unlike the simple arithmetic of Rule of 40, the Rule of Data is qualitative and consists of three pillars:

i) Pillar
ii) Description
iii) Why it Valuates Higher

1. Exclusivity (The "Un-Scrapeable")

Data that cannot be scraped from the open web (e.g., private patient records, internal financial logs, proprietary sensor readings).

Common Crawl data is free; your internal data is a monopoly.

2. Label Fidelity (The "Truth" Layer)

Raw data is noisy. Data that has been cleaned, annotated, and structured by human experts (RLHF) holds exponential value.

An algorithm is only as good as its ground truth. High-fidelity labels reduce hallucination rates.

3. The Feedback Loop (The "Flywheel")

Does the product generate new proprietary data as customers use it? This creates a compounding moat. The more the product is used, the better the model gets, and the harder it is to displace.
 
III. The "TechBio" Case Study: Biological Data as the Ultimate Moat

The user's observation about proprietary biological data being the "only defensible asset" is strongly supported by 2025 market trends. Biological data is complex, high-dimensional, and heavily regulated, making it impossible for generalist AI models (like GPT-5 or Gemini) to master without access to private repositories.

Key Examples of the Data Moat in Action:

Tempus AI: Their valuation is not driven by their algorithms, but by their massive library of 9M+ patient records and 4M genomic profiles. This multimodal dataset (clinical + genomic) allows them to train precision medicine models that no open-source model can replicate.​

AstraZeneca: Is pivoting from a traditional pharma model to a data-first strategy, building a "Biological Insights Knowledge Graph." They are targeting the analysis of two million genomes by 2026, creating a "data supremacy" strategy where the dataset is the primary asset, and AI is just the tool to mine it.​

Gilead Sciences: Has built a moat around decades of virology and oncology data. By centralising this into a "data mesh," they turn historical trial data into a forward-looking predictive asset. Their strategy explicitly pairs their internal data with partner algorithms, acknowledging that the algorithm is the commodity they can "rent," while the data is the asset they must "own".​

IV. Strategic Implications for 2026

1. The "Acqui-hire" is dead; the "Acqui-data" is born.

M&A strategy will shift. Tech giants will no longer buy startups for their engineering teams or novel model architectures. They will acquire companies solely to ingest their proprietary datasets. We will see traditional enterprises (banks, hospitals, logistics firms) being valued like tech companies because they sit on decades of "un-mined" proprietary data.

2. Vertical AI > Horizontal AI.

General purpose LLMs (Horizontal AI) will race to the bottom on price. Value will accrue to Vertical AI companies that sit on top of a Proprietary Data Moat. A "Legal AI" with access to 50 years of private case files beats a generic GPT-5 every time.

3. Governance as a Growth Driver.

In the Rule of Data, "Data Governance" is no longer a back-office compliance task; it is a core value driver. Clean, structured, and compliant data (FAIR principles: Findable, Accessible, Interoperable, Reusable) commands a premium valuation because it is "AI-ready."

Conclusion:

In 2026, the question to ask is no longer "How fast are you growing?" (Rule of 40). The question is "If I gave your competitor your model architecture and unlimited compute, could they beat you?"

If the answer is yes, you have no moat. If the answer is "No, because they don't have my data," you pass the Rule of Data.

To discuss how Nelson Advisors can help your HealthTech, MedTech, Health AI or Digital Health company, please email [email protected]