# Scoring Plug Power against the thesis

_A walkthrough of the score_against_thesis tool_

**Section:** Net Zero  
**Author:** Keri Waters  
**Published:** 2026-05-19  
**Tags:** Net Zero

---

MECP's thesis is a hundred-year program to return atmospheric CO₂ to pre-industrial levels through engineering rather than sacrifice. The argument rests on a sequenced four-pillar dependency chain: cheap electricity (P1) unlocks green hydrogen and electrification (P2), which together make carbon removal economic (P3), all while we harden infrastructure for the warming already locked in (P4).

A thesis is only useful if you can apply it to specific companies. That's what the `score_against_thesis` tool does. It's part of the MECP MCP server, callable from Claude, ChatGPT, Cursor, or any MCP-compatible AI client. It returns a structured thesis-alignment evaluation against the High-Energy Repair framework.

This piece walks through what the tool actually does, using a company most readers will know: Plug Power. The output below is generated by the tool, then lightly annotated with editorial commentary.

## What the tool returned

I asked the tool to score Plug Power. Here's what came back, structured fields first:

Classification: thesis-core
Suggested pillar: Net Zero Emissions (P2)
Suggested sector: green hydrogen
Aggregate confidence: 88%

Scope note: This is a thesis-alignment evaluation focused on technology and strategy fit. It does not assess capital structure, going-concern risk, customer concentration, or management execution. Those require separate diligence.

Then a business-unit breakdown:

<div class="article-table-wrap" data-accent="cyan">

| Unit | Thesis alignment |
|---|---|
| PEM electrolyzers (Rochester GigaFactory) | Thesis-core (P2 enabler) |
| Green H₂ production plants (GA, TN, LA) | Thesis-core but mid-curve |
| GenDrive forklift fuel cells (Amazon, Walmart, Home Depot) | Thesis-adjacent at best |
| Cryogenic liquefaction + distribution | Thesis-adjacent |

</div>

And a sequencing analysis pointing out the obvious tension: Plug is building Pillar 2 capacity before Pillar 1 has delivered. Their plants run, but the cost stack doesn't pencil against grey hydrogen without the 45V tax credit and bridge subsidies. They are early to the right curve.

That's most of the output. The rest is rationale on each unit and a bottom-line recommendation: "thesis-core with a unit-weighting caveat and a separate viability flag rather than letting the 88% confidence stand uninterrogated."

## What's interesting about this

Three things are worth noticing about what the tool did here.

**It refused to collapse the company into one number.** The 88% headline confidence is technically correct. Plug operates the full electrolyzer-to-deployment stack for green hydrogen, which is the exact non-combustion industrial decarbonization vector P2 names. But Plug isn't one business. It's four, and they don't score the same. The electrolyzer gigafactory is the unit the thesis is actually about. The forklift fuel cell business is something else: a cash-generating niche that exists because batteries are inconvenient on a 24/7 warehouse floor, not because hydrogen is the right decarbonization vector for material handling. The thesis explicitly routes road transport through electrification. Forklift hydrogen survives on warehouse-floor logistics, not on the P2 cascade.

The tool surfaced that split without being asked. It also went further: it noticed that the model was averaging across the units and self-criticized the headline confidence. "A unit-by-unit weighting would land lower. Call it 65 to 70%." That kind of meta-analytical move is what the new schema makes possible.

**It applied the thesis's own sequencing logic.** Plug is on the right curve but early. The thesis names $1.50/kg green hydrogen at sub-2¢/kWh electricity as the unlock for P2 economics. Plug's plants are running at delivered prices of 5 to 8¢/kWh today, which is why every production facility depends on the 45V tax credit and bridge subsidies. The technology bet is thesis-aligned. The timing assumption is what's hard. The tool put both points on the table.

**It refused to confuse thesis fit with viability.** The scope note exists specifically to separate two questions that often get tangled. "Is the technology aimed at the right wedge of the right pillar?" is one question. That's what `score_against_thesis` answers. "Will the equity structure survive long enough to ride the cost curve down?" is a different question, and it requires different inputs (cap table, runway, burn rate, customer concentration, management track record). Plug's going-concern risk is real, but it's orthogonal to thesis fit. The scope note lets the tool name what it does and what it doesn't, so the output doesn't get read as something it isn't.

## How to read the output

For editorial use (populating MECP's ecosystem map, briefing a client, drafting a sector deep-dive), the tool produces three things you can use directly and one thing you have to do yourself.

What you can use directly:

The structured classification with confidence. Goes straight into the ecosystem map as the thesis-alignment field. Plug Power gets tagged thesis-core, with the unit-weighting caveat surfaced in the analyst note.

The business-unit breakdown. This is the part that distinguishes an analyst tool from a classifier. For Plug, it means the ecosystem-map entry should call out that the electrolyzer plus production stack is the half that matters for MECP framing, and the forklift business is a stable niche that doesn't get pulled forward by P1 progress.

The sequencing analysis. Useful for the "where on the curve" framing in any company writeup. For Plug specifically, the answer is "right curve, too early," which is its own analytical category, distinct from "wrong curve" or "right curve, well-timed."

What you have to do yourself:

The viability assessment. The scope note exists to make this responsibility transfer explicit. A company can be thesis-core and not survive. Another can be thesis-adjacent and ride a stable niche profitably for decades. For Plug, the viability question is the one that's been hanging over the company for two years: can the equity structure absorb the time gap between today's costs and the curve-crossing where the unit economics work? That's not a question the thesis can answer. It's a balance-sheet question, a capital-markets question, and an operator-judgment question. Use `score_against_thesis` for the thesis half. Use a different analytical tool, or more realistically a different analyst, for the rest.

## Why we built this

The High-Energy Repair thesis is a framework, and frameworks are useful only if they're applicable. We built a tool that applies it, exposes the application through the Model Context Protocol, and makes it callable from any AI client a reader is already using.

That choice reflects a bet about how analyst platforms work in 2026. The platforms that compound are the ones where the analytical framework is callable. A founder building a green-cement startup can score themselves against the thesis from inside Claude. A journalist working in ChatGPT can pull MECP's view of a sector mid-draft. An investor can run a portfolio through the framework in an evening. The framework becomes more valuable as it gets used, and the data layer underneath gets better with every founder who submits, every analyst who corrects, every sector deep-dive that gets published.

`score_against_thesis` is the first tool that exposes proprietary analytical work as a callable function rather than a static essay. The thesis lives in the book. The application lives in the tool. The two together are the platform.

## Try it yourself

The MECP MCP server is live at [mcp.moreenergycleanplanet.com/mcp](https://mcp.moreenergycleanplanet.com/mcp). Connect it to your preferred AI client and run your own company through the tool. Your own, your portfolio, a name you've been arguing about. The full setup is at [moreenergycleanplanet.com/mcp](https://moreenergycleanplanet.com/mcp).

Or just ask the tool the same question I did:

> Use the More Energy Clean Planet connector to score Plug Power against the High-Energy Repair thesis. Show me the business unit breakdown.

The output will be similar to what's above. The framework is in the tool.