A weighted knowledge base, a deterministic risk engine, a project-finance model and global overlays — a deal earns confidence by closing the right things at the right time, and the model tells you whether it pays.
Every diligence item carries an intrinsic importance and a due-by phase gate. An item's effective severity escalates the longer it stays unresolved past the gate where it should have been locked — so a low-importance point ignored too long can outrank a high one that isn't due yet.
How much the item matters to the deal, 1–5. Four and above is critical.
How front-loaded it is. Lateness against the gate compounds the flag.
Four steps. Everything recomputes live as you work.
Set the jurisdiction, megawatts and phase, choose whether you're developing the site, buying a part-built asset to finish, or acquiring a stabilized asset, and the basis you're developing on: powered land, powered shell, turnkey colo or a full-stack AI cloud, and how you're securing power. The brain ring-fences the diligence and the economics to exactly what you own, and surfaces only the questions that apply.
Answer across 21 workstreams — land and permitting through power, construction, cooling, tenants, legal, financing, jurisdiction and force-majeure — by keyboard if you like, or filter to a single discipline (legal, commercial, financial, technical, ESG, insurance, tax). Drop in decks, IMs and term sheets; they stay local.
The RAG heatmap, DD confidence score, lacuna detector, red-flag register, capital-readiness gates and the economics all recompute on every answer — and as you advance the phase.
Auto-build the diligence questionnaire for the open gaps, a printable risk report, an IC memo with the economics, a one-page acquisition deal summary, per-discipline DD reports (Legal, Commercial, Financial, Technical, ESG, Insurance, Tax) to hand each adviser, a client-ready diligence brief and a prioritised task list — in one click.
No black box. Every score traces back to a weighted answer and a gate.
All 21 workstreams scored red, amber or green — red on any high flag or two critical gaps; green only when complete and clean.
0–100: weighted completeness, less flag penalty, less lateness penalty. Projects start red and earn green.
Isolates the open critical questions — the gaps a sharp analyst would challenge before recommending.
Built-in conditions fire automatically: indicative-only power offers, unsigned ESAs, queue position, FX risk.
A front-loading view — on-gate, due-now, or carrying overdue blockers as the project advances.
Edit any question's text, weight and gate; disable items; add your own; apply deal-type profiles. All changes flow through scoring.
A screening cost model, a project-finance return and an M&A valuation in the same tool — calibrated to 2024–26 benchmarks and tuned to whether you're building it or buying it.
$/MW build, land, grid, on-site generation and compute — by region, redundancy, cooling and build standard, with an honest order-of-magnitude band. Back-solve the rate to a known number.
Levered and after-tax equity IRR, yield-on-cost, DSCR and a year-by-year cash flow — with sculpted, amortising or bullet debt, IDC, an energization-slip drag and an exit at a cap rate.
For a stabilized asset: a going-in/exit cap-rate valuation on in-place NOI — WALT, value-add and a take-out refinance, with debt yield and WAL-vs-WALT for the institutional or ABS lens. The build inputs drop away; a one-page deal summary falls out.
Drag the drivers against the market band and watch the return move live; run 2,000 simulations for a P10–P90 cone and the probability of clearing your hurdle.
Split equity across investors and model the carried-interest waterfall — and size the debt straight from a target DSCR. All of it flows into the IC memo.
Tell the brain whether you're developing or acquiring, what you own and how it's financed; the diligence and the economics follow — automatically, and only for the parts that apply.
Develop a site (build diligence + a project-finance model), buy a part-built asset and finish it (purchase price + the capex to complete), or acquire a stabilized asset (the M&A lens — rent roll, WALT, change-of-control consents, on-supply legality, and a cap-rate valuation). Development-only items go out of scope as the deal type demands; the workspace and portfolio separate the lifecycles.
Filter the questionnaire to a single discipline, then export a standalone Legal, Commercial (CDD), Financial (FDD), Technical (TDD), ESG & Regulatory, Insurance or Tax DD report — itemised findings, a document review schedule and the open-item request list — to hand each adviser their slice of one project.
Powered land, powered shell, wholesale or retail colo, a CPU cloud / IaaS or a full-stack AI cloud (GPU-as-a-Service). The model ring-fences capex, financing and returns to your scope — and switches the questions to match.
Grid, PPA, 24/7 carbon-free, on-site generation or bridging power — each priced into the energy cost, the capex and the energization timeline that drives the return.
Stage capacity over time and finance each phase — even each asset layer within a phase — on its own terms. The capital-call schedule falls out automatically.
Corporate, project finance or bond/ABS — per asset layer — with route-fit readiness and seed → equity → debt gates that show exactly what's still outstanding to bank it.
Pick the projects that make up a slate and Megawatt IQ rolls them up: concentration by jurisdiction, anchor tenant, use case and build standard; the aggregate capital-deployment schedule; a blended, capex-weighted return; and the correlated risks a fund needs surfaced — shared grid exposure, shared tenant, capital sitting in red-rated deals.
Type the jurisdiction and Megawatt IQ lights up the named grid operator and connection process, the permitting regime, water law, FDI screening statute, data-localization rules, incentive program and financing conventions — across 40+ countries and 12 market tiers.
Add a Claude API key and the app reads an uploaded deck or IM and auto-fills what it can evidence; researches the web to fill and verify facts with sources; produces a web-grounded political- and country-risk briefing for the jurisdiction; answers questions about the project; and drafts the IC-memo narrative. Every AI answer is tagged, never overwrites your input, and the key lives in your browser — sent only to Anthropic. Entirely optional; the brain works fully without it.
There is no server to breach. Megawatt IQ is a static app that runs entirely in your browser.
Projects and answers in localStorage; uploaded documents in IndexedDB. Nothing is transmitted by default.
Export any project to JSON to back it up or move machines. You own the data, in a format you control.
Settings → Data → Erase all local data wipes everything from the browser, instantly and completely.
Start a 14-day free trial and put your next teaser through the brain — you won't be charged until it ends.
Start free trial →