Bids are now won on evidence, not hours

CfD: more capacity, sharper prices

AR7a, the onshore and solar half of the seventh round, awarded 4,905 MW of solar to 157 projects and 1,306 MW of onshore wind to 28 projects. Every successful solar project received £65.23/MWh in 2024 prices, against an administrative strike price of £75. Onshore wind cleared at £72.24 against £92. 1 Across AR7 as a whole, government reported 14.7 GW from 201 projects. 2

The results table also shows who is winning. We parsed all 157 solar awards. The median project is 29.9 MW, 84 of the 157 projects are 30 MW or smaller, and 24 sit between 49 and 50 MW. 134 are in England. 1 For a growth-stage IPP this is the relevant competition: many mid-sized projects bidding at one clearing price, where a few pounds per MWh separate a contract from a lost year.

AR8 changes the rules again. The Price and Pot Notice puts solar alone in Pot 1 and onshore wind in Pot 2. It keeps the AR7 strike price caps, gives government sight of anonymised sealed bids in Pots 1–4, and states that budgets will be confirmed only after applications have been assessed, with value for money as a key test. 3 The application window ran from 20 July to 7 August 2026, with Pot 1 delivery years from 2028/29 to 2030/31. 4 NESO's Notice of Auction and the Contract Budget Notice are scheduled for 19 November 2026. 5

Offtake: a thinner, more demanding PPA market

Veyt recorded 27 UK PPAs totalling 1.8 GW in 2025, down 0.5 GW year on year. Across Europe, corporate PPA activity fell 19% while route-to-market PPAs tripled to 15 contracts. 13 Pexapark counted 24 GB PPAs (626 MW) announced by October 2025, a 39% fall. It reports sell-side offers for 10-year pay-as-produced solar of £74–84/MWh, while buyers' best bids sit £5–10 lower, so no transactable price was observed. 14 The two databases count differently, which is why their totals differ. Both point the same way.

Batteries: tolls, floors and a cheaper Capacity Market

Battery revenue is being contracted in new ways. Drax signed a 10-year, CPI-indexed toll for Fidra's 250 MW / 500 MWh West Burton project, subject to final investment decision (FID) by Q3 2026, 15 and a 15-year, non-indexed toll for Zenobē's 200 MW / 800 MWh Coalburn site. 16 In both, the owner keeps construction and availability risk. Gresham House signed a revenue floor with Statkraft and a revenue swap with Nephila Climate and Markel, together securing £35m a year across 789 MW. Modo estimates the floor at about £44k/MW/year. 17

The Capacity Market sends a weaker signal than before. The 2029/30 T-4 auction cleared at £27.10/kW/year, down from £60 the previous year. It awarded 40.1 GW of de-rated capacity, of which batteries took 1.2 GW (3.0%). 12

Grid: from queue to deadline

NESO's reform replaced a first-come queue of over 700 GW, about four times projected 2030 need, with a 283 GW pipeline of generation and storage. 8 Of that, 132 GW is aligned with Clean Power 2030 (Phase 1) and a further 151 GW is needed by 2035 (Phase 2). 9 By 10 June 2026, 713 of the 1,223 projects in the pre-2030 pipeline (58%) had received offers, totalling 37 GW. 10 NESO gives 90 days to sign a Gate 2 transmission offer, 11 while National Grid Electricity Distribution gives 30 days for its Gate 2 variation offers. 20 Phase 2 transmission offers run from early September 2026 to mid-January 2027, 18 and Ofgem has set 31 January 2027 as the deadline for issuing them. 19 Clean Power 2030 still needs solar to roughly triple and batteries to grow about fivefold. 7 The volume is there; access to it is now rationed by readiness and by dates.

There is no public dataset on bid hours in UK renewables, so this section is analysis rather than measurement. It draws on how the competitions above are structured and on Ignis practitioner experience. We see five recurring loss points. Each one is a place where senior time goes into production rather than judgement, or where an error reaches the counterparty.

The pattern is consistent. The scarce people, usually the CEO, the commercial director and one or two senior managers, spend most of their bid time on chasing inputs, drafting and reconciling documents. Judgement, buyer and partner relationships, strategy and negotiation are the work that wins the deal, and they get whatever time is left over. Figure 7 sets out the alternative as an illustrative model: twice the time on what wins the deal, and still fewer hours.

A single-asset developer can run one bid at a time. A fund-backed IPP with 5 to 20 projects cannot. Each project may need a CfD application, a route-to-market or corporate PPA, a battery toll or floor, Capacity Market prequalification and a Gate 2 acceptance, and these deadlines overlap. In 2026 alone, the AR8 window, the T-4 results, a published battery-toll FID deadline and the Gate 2 offer wave all fell within a few months of each other (Figure 8).

Three things break first. Key people: one commercial director holds the price logic for every bid. Consistency: the same project is described differently to DESNZ, a PPA buyer, a toll counterparty and NESO. Learning: when the desk is over capacity, the debrief is the first thing to go. Hiring helps only slowly. A senior bid manager who knows CfD rules, PPA terms and lender requirements takes months to find and longer to bring up to speed.

The design principle is borrowed from good investment committees: separate the preparation from the decision, and keep both on the record. Agents are good at reading long rule books, comparing documents and drafting to a template. People are needed for price, risk appetite, partner terms and anything that commits the company. The eight steps in Figure 9 apply whether the bid is a CfD application, a PPA tender, a toll, a Capacity Market entry or a Gate 2 acceptance.

Under the workflow sits a deliberately boring stack: one evidence library per project, one pricing model per project linked to that library, a template set for each bid type, and an access and audit log. Ignis agents provide the agent layer for this work. The Ignis system is in pilot, and we make no performance claims for it here. The design is meant to be useful on any toolset.

Each term has an owner. P(award) depends on how competitive the price is and how strong the evidence is. Contract margin depends on the strike or contract price, yield, capex and tenor. Bid cost is mostly senior time. P(error) × cost of error is the term most teams leave out. It covers a delivery year the grid date cannot meet, a capex input that is six months out of date, or a PPA offer that contradicts the CfD bid.

Two consequences follow. First, in a CfD round all successful projects in a pot receive the clearing price, 1 so the highest-value pricing work is setting an honest floor from current capex, yield and grid evidence, not shaving the bid. Second, the cheapest way to raise expected value is usually to cut the error term with cross-checks, rather than to add more drafting hours.

The example is a hypothetical co-located project: 60 MW of solar and a 30 MW / 60 MWh battery. The team is considering an AR8 CfD for the solar plus a route-to-market PPA, a toll or floor for the battery, and a Capacity Market entry. Public numbers anchor the case. Everything else is an assumption, labelled in Table 3.

Read across the tables and the order of priorities becomes clear. An agent-supported desk does not make the price decision. It ensures that the price decision is taken early, on current evidence, and that the CfD bid, the PPA offer, the battery term sheet and the Gate 2 acceptance all describe the same project with the same dates.

1. Set a portfolio bid calendar. List every CfD, PPA, toll, Capacity Market and Gate 2 event for the next 12 months against the people available. Decline early where load exceeds capacity.
2. Fix the floor before the bid. Agree each project's walk-away on current capex, yield and grid evidence at the bid or no-bid gate, not in the final week.
3. One project, one evidence library. Every bid draws on the same approved facts. Stale items are flagged automatically.
4. Make cross-checks mandatory. No submission leaves until dates, capacity and prices match the model and every other live submission.
5. Measure the right hours. Track senior time on judgement, relationships and negotiation against time on coordination and production. Aim to double the first while cutting the total. Move production work to agents with named reviewers.

For investors, the outcome is a bid record they can audit: why each bid was entered, at what floor, on which evidence and with whose sign-off. That record supports the board pack and the lender pack at financial close.

Frequently asked questions

How was this analysis done?

Data. Public sources only, listed below and accessed in October 2026. AR7a project sizes were parsed from the DESNZ results table (157 solar rows, which reconcile to the published 4,905.01 MW). Capacity Market prices are from Ofgem's Table 1.

Illustrative modelling. Figure 7 and Table 2, Table 3, Table 4 and Table 5 are illustrative. Bid hours (including the 20/80 conventional split and the 40/30 agent-supported split in Figure 7), event counts, load factor, discount rate and battery de-rating are assumptions based on Ignis practitioner experience, not measurements. The calculations are simple and shown so they can be reproduced.

Limitations. No public dataset covers bid hours or bid costs. The PPA buyer range is approximated from Pexapark's statement that buyers' best bids sit £5–10 below sellers' offers. CfD prices are in 2024 prices and CPI-indexed, while PPA quotes are not, so the two are not directly comparable. Revenue figures exclude indexation, curtailment, negative-price rules and balancing costs.

Not covered. Offshore wind, Ireland, France and Iberia. France's AO10 tender has its own Ignis news note.