Patent Pending · ERCOT Live · PJM Coming 2026

Know when your settlement point is at risk — and what to do about it

GridSignals reads the physical state of the grid — wind, reserves, generation, battery behavior, transmission constraints — and distinguishes system scarcity, local congestion, and node-level basis risk. Then translates them into prepare, hold, curtail, and resume decisions. With receipts.

Many tools are threshold-based or reactive. GridSignals watches the physics that cause price spikes at your settlement point, not the grid average. Minutes of advance operational warning from grid physics, with live and replay evidence depending on mechanism.

30–45
minutes
Advance warning from grid physics before energy prices spike — time to stage your response, not react to it.
410+
GW in queue
ERCOT interconnection requests, 87% data centers. Every new facility needs operational intelligence before its first summer.
5
min cadence
Continuous reassessment of grid state, mechanism classification, and risk level. Every five minutes, aligned with ERCOT's real-time market.
Market-specific intelligence, not one-size-fits-all
Scarcity Detection
Live production primary. Live-event proof accumulating.
Congestion Detection
Live shadow. Validated on multiple real events.
ERCOT 4CP
Shadow live. First post-RTC+B season Summer 2026.
PJM 5CP
Shadow. Summer 2026.
For Operators

Fewer, better-timed curtailment decisions

Know whether to curtail, hold, or wait for resolution. Mechanism context prevents unnecessary downtime from congestion events that resolve in minutes.

For Developers

Flexibility readiness and tenant confidence

Demonstrate operational credibility to enterprise AI tenants and lenders before your site energizes. Grid intelligence as infrastructure, not an afterthought.

For Finance

Receipts, avoided-cost evidence, and governance proof

Every recommendation, action, and outcome documented. The audit trail that CFOs, lenders, and boards need to underwrite operational risk.

The AI infrastructure credibility gap

Enterprise AI customers ask one question that separates the tiers: what happens to my workload when the grid gets tight? Tier 1 answers with hardware. Tiers 2 and 3 need a different answer.

Tier 1

Hyperscalers

Microsoft, Google, AWS

Sovereign-scale infrastructure. They acquire nuclear plants, sign 20-year PPAs, and spend billions on redundant power systems. Grid exposure is a rounding error.

Tier 2

Neoclouds

CoreWeave, Core Scientific, Lambda

Competing for enterprise AI workloads with speed and price. But they operate in deregulated markets where grid volatility is real. They need Tier 1 credibility at Tier 2 economics.

GridSignals target
Tier 3

Regional Operators

Colocation, hosting, hybrid

Adding GPU density and AI tenants. Inheriting grid exposure their customers used to manage. Need operational intelligence they've never had before.

GridSignals target

Not just what's happening — whether it's getting worse, and whether help can reach you

Most tools answer one question. GridSignals answers three — the three questions an operator actually needs before making a curtailment decision.

Question 1

Is something bad coming?

Machine learning models trained on post-RTC+B data detect when grid conditions are deteriorating — reserve compression, AS tightening, wind underperformance, load surprises. Distinguishes system-wide scarcity from localized congestion. 30–45 minutes of lead time.

Question 2

Can the grid handle it?

A proprietary deliverability layer answers the question other tools ignore: even if system-wide reserves look adequate, can marginal generation actually reach your settlement point? When headline reserves mask local risk, deliverability catches it.

Question 3

Is it getting worse?

Rate-of-change signals track how fast the grid's pricing engine is climbing the supply curve. This distinguishes an evening ramp that resolves at $60 from one that escalates to $200 — before the outcome is determined.

The grid average is not your price

Your bill is determined at your settlement point, not the system average. During congestion, these can be $400 apart.

System
System lambda

Scarcity moves the system. When reserves compress grid-wide, everyone's prices rise together.

Hub
Hub price

Congestion moves the hub. Transmission constraints spike one zone while others stay calm.

Node
Your settlement point

Basis moves your node vs the hub. Your meter can diverge from the hub average by $30+ during local constraints.

GridSignals monitors all three levels. Most tools stop at the system. We track the corridor constraints, import paths, and local conditions that determine what you actually pay.

What GridSignals sees that price screens miss

Observed Event — Evening Congestion
System reserves
21,000 MW
Headline looks comfortable
Local hub price
$274
Local constraint binding
Shadow price
$2,424
Constraint severity
System average
$35
Grid-wide: calm
System-wide reserves were 21 GW — comfortable by any system-level measure. But a local transmission constraint drove one hub to $274 with $2,424 shadow prices while the system average sat at $35. A system-level alert tool would show green. GridSignals classified this as local congestion and identified the mechanism, the affected settlement points, and the likely resolution path.

Based on observed ERCOT market data. GridSignals classification from shadow system during live operation. Similar congestion dynamics exist in PJM, SPP, and other nodal markets where transmission constraints create localized price separation.

We watch what moves before price does — at your settlement point

The grid average and your settlement point can be $400 apart during congestion events. We monitor the physical signals specific to your location.

Predictive Scarcity Detection

Detects when system-wide conditions are deteriorating — reserve depletion, AS tightening, and supply-demand imbalances that precede price spikes by 30 to 45 minutes.

Congestion Detection

Localized price spikes from transmission constraints and weather-driven generation shortfalls. Your settlement point can spike $400 while the rest of the grid is calm. We detect these independently.

Disturbance Detection

Rapid detection of sudden grid disturbances — generator trips and rapid imbalances that can't be predicted, only detected fast.

Mechanism Classification

Every alert tells you why — scarcity, congestion, or load surprise. The mechanism determines your response. Scarcity means curtail. Congestion often means wait. Knowing which prevents unnecessary downtime.

One alert. One action. One all-clear.

Every event produces at most three messages: prepare, act, and resume. Each one tells you what's happening physically, not just where the price is. No alert spam.

T − 30 to 45 min
Prepare

Grid physics indicate stress is building

Multiple independent signals are converging at your settlement point. The alert tells you the mechanism — scarcity, congestion, or load-driven — and the expected severity. Time to stage your response.

If conditions escalate
Curtail

Conditions have crossed your threshold

Price has reached your action level with physical confirmation. This is the hard backstop.

Resume

Confirmed safe — not just a price dip

Prices can dip during an ongoing event and rebound minutes later. We hold the all-clear until the underlying stress has genuinely resolved — reserves rebuilding, generation recovering, grid returning to normal.

Software-defined grid credibility

The traditional answer to grid exposure is hardware — heavy UPS systems, switchgear, and battery buffers costing tens of millions per site. For a Neocloud operating hundreds of megawatts across ERCOT, that capital is prohibitive.

GridSignals provides the same operational credibility through software. Instead of buffering grid volatility with hardware, we read the physical precursors and orchestrate the response before the event reaches the facility. The enterprise customer gets timestamped receipts proving their workload was never at risk.

The first answer requires capital. The second requires intelligence. Both deliver credibility. Only one scales at Tier 2 economics.

Easier to finance, easier to insure

Lenders and project finance teams underwrite operational risk. A facility that can demonstrate predictive grid intelligence — with timestamped receipts showing every event was anticipated, every response was documented, and every outcome was measured — is a lower-risk asset than one without.

GridSignals produces the audit trail that CFOs, lenders, and board members need: what happened on the grid, what the system recommended, what action was taken, and what cost was avoided. That documentation turns operational intelligence into a financing advantage.

Your model picks the day. Ours picks the minutes.

Most operators can identify obvious candidate days. The hard part is deciding which candidates are real, which are false alarms, and how tightly to operate around them.

ERCOT 4CP
2 consecutive years
June's official 4CP landed on a Sunday — not a day typically associated with peak demand. Storage behavior suppressed the obvious weekday candidate at settlement.
ERCOT 4CP
1,080 facilities
Now actively managing 4CP. As more participants respond to the same signals, collective response can shift which interval counts at settlement.
ERCOT RTC+B
First summer
Summer 2026 is the first 4CP season under RTC+B. Battery co-optimization every five minutes changes which intervals survive settlement adjustment. Models calibrated on pre-RTC+B data don't account for this.

Settlement-aware ranking

The harder layer is what survives at settlement once storage behavior, forecast changes, and collective response are in the mix. That's the layer we focus on.

Intraday re-ranking

Candidate positions update throughout the day as conditions evolve. When the day tightens or loosens, your ranking and action zone shift with it.

Uncertainty-aware

Optimistic, midpoint, and pessimistic scenarios instead of a single overconfident alert. You see how much uncertainty is in the adjustment.

Auditable receipts

Every candidate ranking, every adjustment, every decision documented. When the season ends, the record shows what happened and why.

The grid changed. Most tools didn't.

ERCOT's RTC+B co-optimization went live in December 2025. Under the new market rules, batteries co-optimize across energy and five ancillary service products every five minutes. This fundamentally changed how prices form, when stress events develop, and which signals lead.

GridSignals is built natively on post-RTC+B data and market structure. Every model, every detection layer, every alert threshold was developed on the current market — not retrofit from pre-RTC+B assumptions.

Meanwhile, the ERCOT interconnection queue has reached 410+ GW, with 87% from data centers. Large flexible loads are increasingly settling at points where localized congestion can spike prices $400+ while the rest of the grid is calm. System-wide monitoring tools don't see these events.

The system improves after every event. When grid conditions produce an outcome our models haven't seen before, the system identifies the gap, incorporates the new evidence, and sharpens detection for the next event. Patent pending.

These dynamics are not unique to Texas. In PJM, rising 5CP capacity charges, increasing battery penetration, and growing data center load create the same need for settlement-point-specific operational intelligence. GridSignals is expanding to PJM in Summer 2026, built on the same physics-first architecture.

Selective deployment per market

We deploy selectively by market and load profile to preserve signal quality and actionability. When many participants respond to the same signal, collective behavior can shift which intervals count. Our architecture is built to manage that uncertainty.

ERCOT Scarcity — Live ERCOT Congestion — Live Shadow
ERCOT 4CP — Shadow Live, Summer 2026 PJM 5CP — Shadow, Summer 2026
Request a site-specific replay

Not just alerts — operational output

Prepare / Curtail / Resume posture

Real-time operational posture with mechanism context. Not just "price is high" but "this is wind-driven congestion on the Panhandle corridor, likely resolves in 15 minutes."

Settlement-point replay pack

Historical analysis of your specific settlement point — what events hit your node, what the system saw, what it would have recommended, and how much lead time was available.

Post-event receipt

After every event: what happened, what was recommended, what action was taken, what cost was avoided or incurred. Timestamped and auditable.

Weekly digest

Synthesis of the week's grid conditions, events, model performance, and system health. What the grid did, what we saw, and what we learned.

Flexibility-readiness memo

For pre-energization sites: settlement-point risk profile, historical exposure analysis, curtailment strategy framework, and operational credibility documentation for lenders and tenants.

4CP / 5CP candidate advisory

Ranked candidate intervals with uncertainty bands, intraday re-ranking, and settlement-aware timing. Receipts after each season showing what happened and why.

Common questions

How do you provide advance warning before prices spike?
Energy prices are the last signal to move during a grid stress event. The physical grid — generation levels, reserve margins, transmission flows, operator actions — shifts first. We monitor these physical precursors and detect patterns of developing stress before they reach the energy market. For system-wide scarcity events, that typically means 30 to 45 minutes of actionable warning. For localized congestion, we track conditions specific to your settlement point. The system continuously learns from every event it observes.
What's the difference between scarcity and congestion?
Scarcity is a grid-wide event — not enough generation to meet demand. Everyone's prices go up. Congestion is a local event — your settlement point spikes because of transmission constraints, while the rest of the grid is calm. We've observed cases where one zone reaches $470 while another is negative $3 at the same moment. Most alerting services watch grid-wide signals and miss congestion entirely. We detect both independently.
Why does the mechanism matter if the price is the same?
Because the correct response is different. During scarcity, the grid is stressed and prices may stay elevated for hours — curtailing makes sense. During congestion, the grid is fine and prices often resolve in minutes as the system re-dispatches. Curtailing for a 15-minute congestion pop costs you production for nothing. Every alert from GridSignals tells you the mechanism.
What is deliverability and why does it matter?
Deliverability answers: even if system-wide reserves look adequate, can the next marginal megawatt actually reach your facility? During congestion events, headline reserves can be 20,000 MW while your settlement point spikes $300+ because power can't flow through a constrained corridor. Our deliverability layer monitors corridor saturation, import flexibility, battery fleet status, and node-level exposure to catch risk that system-wide tools miss entirely.
How do you know when it's safe to resume?
Prices can dip during an ongoing event and rebound minutes later. A brief price drop doesn't mean the stress is over. We track the underlying physical conditions — are reserves rebuilding? Is generation recovering? Are operators still deploying emergency responses? The all-clear only comes when the physical grid has genuinely normalized.
How is this different from existing 4CP management?
Most peak management approaches are strongest at identifying obvious high-demand candidate days from load and weather data. The harder layer — and where outcomes are increasingly decided — is what survives at settlement after storage behavior, intraday forecast changes, and collective curtailment response are accounted for. That settlement-aware layer is what we focus on.
How is 4CP different under ERCOT's new RTC+B rules?
Under RTC+B, batteries co-optimize across energy and five ancillary service products every five minutes. Their charging and discharging behavior directly affects which intervals survive settlement adjustment. Models calibrated on pre-RTC+B data don't account for this. Summer 2026 is the first 4CP season under the new rules.
Why do you deploy selectively?
When many participants respond to the same visible candidate, collective behavior can shift which interval ultimately counts. Deploying selectively by market and load profile ensures our recommendations account for this dynamic rather than contributing to it.
What about the future of 4CP?
Texas Senate Bill 6 requires the Public Utility Commission to review 4CP cost allocation fairness by December 2026. The methodology may change. GridSignals is built with modular architecture — if the rules change, the system adapts.

Built from the ground up

Soren White

Founder & CEO

Designed and built the entire GridSignals platform — data infrastructure, ML models, real-time detection architecture, and autonomous operational systems.

Laurie Fitzmaurice

Head of Energy Markets

Energy markets expertise across ERCOT, PJM, and other deregulated markets. Leads commercial strategy and customer relationships.