Ceres Power closed down 5.02% at 282.60p on 20 October 2026, giving a market value of £604m and leaving the shares about two-thirds below the 871.50p high of the past year. The fall has taken out the spring enthusiasm for fuel cells as power for AI data centres, but it has not made the shares cheap on any measure the accounts support. After the £172.0m of cash and short-term investments held at 30 June, the business is valued at about £438m, nearly ten times the roughly £45m of 2026 revenue the company says is contracted. FY2025 produced a £47.5m loss and only £110,000 of royalties, the recurring income on which the licensing model ultimately depends.
The balance sheet is not the problem. June's £102.6m gross raise removes any near-term funding question, and first-half 2026 revenue rose 8% to £22.8m while the adjusted EBITDA loss narrowed to £6.8m. What matters is conversion. FY2025 revenue of £32.6m fell far short of £50-60m guidance, and the near-term story now rests on a new licensee the company expects to sign in 2026, and on whether Doosan's first export order turns into royalties. The chart shows persistent selling since the June placing, short interest has climbed to 17.55%, and rising gilt yields are making investors less willing to pay today for cash flows several years away.
Ceres Power develops solid oxide fuel cell (SOFC) technology, which turns gas into electricity at high temperature without burning it, and electrolysis technology, which runs the same cell in reverse to make hydrogen. It does not manufacture at scale itself. Partners license the designs, build the stacks and sell the products, paying Ceres licence fees and engineering charges up front and, in time, a royalty on each unit. In FY2025 that last element amounted to £110,000 out of £32.6m of revenue, so the business is still paid almost entirely for access to its technology and not yet for volume.
Doosan Fuel Cell in South Korea is the partner closest to volume. On 6 August 2026 Ceres noted Doosan's contract with Reverion GmbH, worth about 108.7 billion KRW (c£60 million), to supply Ceres-licensed SOFC stacks to power facilities in Germany and Europe, Doosan's first export of the technology. That sum is Doosan's contract value. Ceres's share arrives through royalties as stacks are delivered. The home market looks less secure: Seoul Economic Daily reported on 7 October that Korea has removed a public-building installation mandate for fuel cells, which run on natural gas and emit carbon dioxide.
The wider partner list has changed shape. Denso signed a manufacturing and licence agreement in July 2024, and a strategic partnership with Centrica was announced alongside the FY2025 results in March 2026. Bosch, an earlier manufacturing partner, decided in February 2025 to stop developing SOFC cells and stacks. The pitch to data-centre operators is that gas-fed fuel cells can supply continuous on-site power where grid connections are scarce, then switch to hydrogen later. Gas turbines and engines compete for the same demand.
Ceres has also slimmed down. Headcount fell to 462 at the end of 2025 from 546 a year earlier. RFC Power, a flow-battery business acquired outright in September 2025 for a nominal sum and later flagged as unable to continue as a going concern, was spun out in August 2026 through a £10m Series A led by the British Business Bank, with Centrica and Temasek Trust's C3H. Ceres put in £1m of cash and £1.5m of services in exchange for equity, keeping about 37% of a business valued at about £12m after the round, now renamed Certain Energy.
| 26 Aug 2026 | Value | Change |
|---|---|---|
| Ceres retained stake in RFC Power | c.37% | post-money value c.£12 million |
Sources: Ceres spins out RFC Power RNS, 26 Aug 2026
FY2025 revenue fell to £32.6m from £51.9m. In March 2025 Ceres had reverted to £50-60m guidance while it worked to clarify the correct accounting treatment under IFRS 15, the standard that decides when contract income is booked, and confirmed a gross margin above 75%. The margin came in at 70%, down from 77%, and gross profit fell to £22.7m from £40.2m. In a licensing business the timing of revenue recognition can move a year's result substantially, because a single licence fee may be booked at signing, spread over the contract or tied to milestones. No restatement of prior accounts has been quantified, so the evidence points to a forecasting failure, not to misstated figures.
Losses widened as revenue fell. The statutory operating loss was £47.6m against £31.3m, and the adjusted EBITDA loss (earnings before interest, tax, depreciation and amortisation, excluding one-off items) grew to £32.5m from £22.3m. Operating costs did come down, to £70.1m from £74.3m, including £3.4m of exceptional costs. Finance income of £4.1m on the cash pile cushioned the pre-tax loss of £46.3m.
The first half of 2026 looks better. Revenue rose 8% to £22.8m and the adjusted EBITDA loss narrowed to £6.8m. A large part of that revenue came from one-off licence income, which flatters a single half and says little about recurring earnings. The test for the second half is whether costs stay at the lower level and whether any of the income starts to repeat.
Cash outflow halved in 2025. Operating cash flow was negative £20.1m against negative £35.9m, and total cash outflow was £19.2m against £37.5m, leaving net cash of £83.3m at the year end. Capital spending fell to £1.9m from £6.7m, which is what an asset-light model should look like. The smaller cash burn than the £47.5m loss suggests that non-cash charges and working capital helped, and it still falls well short of self-funding.
The June placing changed the picture. Ceres raised £102.6m gross, and 18,000,000 new shares were admitted on 12 June 2026, lifting the share count by about 9% to 213,832,469 by 30 September. Cash and short-term investments reached £172.0m at 30 June, about 80p for each share in issue. Debt is negligible: £2.41m of borrowings and £3.15m of leases at the end of 2025.
The accounts carry no material uncertainty over going concern. The directors stress-tested their projections against scenarios including a slower intake of new licensee partners. At the FY2025 rate of outflow, £172.0m covers several years, though that is a static calculation and spending plans can change. The split between cash and short-term investments, and the reconciliation of licence revenue to receivables and cash received, are the details that will show whether first-half revenue is arriving as money or building up as amounts owed. Ceres pays no dividend and has no defined-benefit pension scheme.
The September interim report put 2026 contracted revenue at about £45m and said Ceres was confident of signing another licensee this year. Figures on the company's website, which it presents as indicative, show FY2026 revenue of £60m and EBITDA of £1m. The gap between the two suggests the higher figure depends on a new licence being signed and booked before the year end. A licence that slips into 2027, or that arrives with fees staged over several years, would leave the full year closer to the contracted base.
The guidance record argues for treating pipeline language as a probability. In January 2025 Ceres expected £55-60m of revenue for the year, cut that to £50-60m in March, and delivered £32.6m. The more recent pattern, with lower costs and a contract-based revenue figure, is easier to check. The research database lists 18 November as a possible results date, with low confidence. Interim results were published on 23 September, so the firmer markers are any licence announcement, a year-end trading update and the FY2026 results.
The backdrop is less forgiving than in the spring. The two-year gilt yield rose 0.42 points in three months to 4.90% at 16 October, while Bank Rate held at 3.75% and CPIH inflation was 3.3% in August. Brent crude reached $110.55 on 20 October. Ceres does not produce oil, so the main effect of dearer energy is the risk of tighter policy for longer, which lowers the value investors place on profits several years away. The 28 October Budget and the Bank of England's November decision are the next tests for rates.
| Figure | Value | Change |
|---|---|---|
| FY2026 revenue (company website, indicative)FY2026e | £60m | |
| FY2026 EBITDA (company website, indicative)FY2026e | £1m |
At 282.60p on 20 October, Ceres is valued at £604m. Taking off the £172.0m of June cash and short-term investments and adding back roughly £5.6m of borrowings and leases gives an enterprise value (the value of the operating business, net of cash and debt) of about £438m. That is nearly ten times the roughly £45m of contracted 2026 revenue, and the market value is about 18 times FY2025 revenue. Losses rule out any earnings multiple.
Breakeven arithmetic shows how far the business has to travel. At FY2025's £70.1m of operating costs and a 70% gross margin, revenue would need to reach about £100m to cover costs. Annualising the lower first-half 2026 cost base brings that to about £88m. Both figures are roughly double what is contracted for 2026, and they assume the margin holds as the mix moves from licence fees to royalties.
The 52-week range runs from 206.00p to 871.50p, and the shares closed at 591p on 12 June, the day the new placing shares were admitted. Investors who bought in the placing are sitting on losses of around half, which tends to create sellers whenever the price recovers. The year-to-date performance figures in the data conflict, so the claim that the shares are still up on the year is a weak anchor. Measured from the 206.00p low, they remain well above where they started the rally.
The shares have fallen hard and on volume. To the 19 October close of 295p they were down 28.1% over a month and 20.3% in a week, and the 50-day moving average crossed below the 200-day on 10 September, a pattern chartists call a death cross. The price sits below every major average, 31% under the 200-day. On-balance volume, a running tally that adds volume on up days and subtracts it on down days, has fallen almost continuously from about -9 million in mid-June to -42 million, which points to distribution (persistent selling) since the placing.
Momentum readings are stretched. The 14-day relative strength index was 29.7 at the 19 October close, below the 30 level usually read as oversold, which raises the odds of a bounce without signalling a turn. There is no bullish divergence yet, and the latest swing high on 23 September and 2 October failed to exceed the one before. Daily moves average about 10% of the price, so swings of a few per cent mean little.
The nearest support at 290.7p has been touched seven times, most recently in July. The 20 October trade at 282.60p, in a day range of 278.20p to 295.44p, puts that level under test. A decisive close below it on rising volume would open the way to the next shelf at 207.5p, close to the December 2025 low. On the upside, resistance sits at 348p to 361p, with an unfilled gap from 15 October between 354.39p and 368.98p.
Positioning adds to the volatility. Disclosed short positions reached 17.55% of shares on 15 October across ten holders. The October holding notices complicate the picture: JPMorgan's 5.34% and Barclays' 5.37% are held mostly through financial instruments, and the 5.10852% notified by funds linked to Christian Leone sits almost wholly in cash equity swaps. Such positions can reflect broker hedging or long exposure as well as bets against the shares, so they do not all point one way. BlackRock lifted its holding to 6.78% in September.
The case for starts with the balance sheet and the price. Ceres holds £172.0m of liquidity, about 80p a share, and almost no debt, so it can keep developing the technology and pursue partners without returning to shareholders soon. First-half 2026 revenue rose 8% and the adjusted EBITDA loss shrank to £6.8m after a year of cost cuts. Doosan's c£60m Reverion order is the first export of Ceres-licensed stacks and the first sign of demand outside Korea. A new licensee, particularly one aimed at data-centre or Western power markets with meaningful upfront cash, would show the model can grow beyond its current partners. With short interest at 17.55% and the shares two-thirds below their high, good news could move the price sharply.
The case against rests on conversion and credibility. Royalties were £110,000 in FY2025, FY2025 revenue fell almost £20m short of the low end of guidance, and the margin missed too. Contracted 2026 revenue of about £45m is half what is needed to break even on the current cost base, and an enterprise value of about £438m already prices in years of improvement. The partner base has thinned, with Bosch gone and Korea's public-building mandate removed, while gas turbines compete for data-centre demand. June's cash came from shareholders, and a future raise would come at a far lower price than 12 June's 591p close if the licence timetable slips. Selling has been persistent since the placing and higher gilt yields weigh on shares whose value lies years ahead.
The evidence that would separate the two cases is specific. First is whether the new licensee is signed by the end of 2026, and on what cash terms, since a headline with back-loaded fees would do little for this year's numbers. Second is whether FY2026 revenue clears the £45m contracted base and how it compares with the £60m shown on the company website. Third is royalty income in the FY2026 accounts against FY2025's £110,000, alongside licence revenue turning into cash rather than receivables. Then come the next deliveries under the Reverion contract and any further Doosan exports. Rates matter too, starting with the 28 October Budget and the Bank of England's November decision.
| Figure | Value | Change | |
|---|---|---|---|
| RevenueFY2025 | £32.6m | −37% from £51.9m; guidance had been £50-60m | |
| Adjusted EBITDAFY2025 | -£32.5m | +46% from -£22.3m | |
| Operating profitFY2025 | -£47.6m | +52% from -£31.3m | |
| Operating costsFY2025 | £70.1m | −5.7% from £74.3m | |
| Net profitFY2025 | -£47.5m | +68% from -£28.3m | |
| RevenueH1 2026 | £22.8m | +8% | |
| Adjusted EBITDAH1 2026 | -£6.8m | ||
| FY2025 revenue guidanceGiven 11 Mar 2025 | £50-60 million | ||
Sources: Final results RNS, 26 Mar 2026; Interim Results RNS, 23 Sep 2026; Guidance, 11 Mar 2025
| Figure | Value | Change | |
|---|---|---|---|
| Net cash31 Dec 2025 | £83.3m | −18% from £102m | |
| Net cash from operating activitiesFY2025 | -£20.1m | −44% from -£35.9m | |
| Cash outflowFY2025 | £19.2m | −49% from £37.5m | |
| Leases31 Dec 2025 | £3.15m | ||
| Shares in issue30 Sep 2026 | 213,832,469 | ||
| New shares admitted12 Jun 2026 | 18,000,000 | ||
Sources: Final results RNS, 26 Mar 2026; Debt, maturities and covenants, 2025-12-31; Total Voting Rights RNS, 30 Sep 2026; Total Voting Rights RNS, 30 Jun 2026
| FY2025 revenue guidance in January 2025Given 29 Jan 2025 | £55-60 million |
| 2-year gilt yield16 Oct 2026 | 4.90% | +0.42 points over three months |
| Bank Rate19 Oct 2026 | 3.75% | +0.00 points over three months |
| Brent crude20 Oct 2026 | $110.55/bbl | +18% +over three months |
Sources: Company website, key financial figures (indicative); Guidance, 29 Jan 2025; UK economy and markets, 16 Oct 2026; UK economy and markets, 19 Oct 2026; UK economy and markets, 20 Oct 2026
| 20 Oct 2026 | Value | Change |
|---|---|---|
| Share price | 282.60p | −5% on the day |
Sources: Market data, 20 Oct 2026
| Figure | Value | Change |
|---|---|---|
| Day range20 Oct 2026 | 278.20p - 295.44p | |
| JPMorgan Chase & Co. voting rights15 Oct 2026 | 5.335648% | 5.143303% through financial instruments |
| Barclays PLC voting rights14 Oct 2026 | 5.37% | 4.80% through financial instruments |
| Thebes funds voting rights9 Oct 2026 | 5.10852% | 5.08138% through cash equity swaps |
| BlackRock, Inc. voting rights22 Sep 2026 | 6.78% | from 5.85% |
Sources: Market data, 20 Oct 2026; Holding(s) in Company RNS, 20 Oct 2026; Holding(s) in Company RNS, 19 Oct 2026; Holding(s) in Company RNS, 15 Oct 2026; Holding(s) in Company RNS, 24 Sep 2026
A stronger reading of the same evidence starts from what the market value now implies. Strip out cash worth about 80p a share and the operating business is valued at about £438m, for intellectual property that Doosan, Denso and Centrica have chosen to work with and that has just produced its first export order in Europe. The FY2025 miss was a timing failure in a lumpy licensing business, and no restatement has been quantified. The first half of 2026 shows costs falling and losses narrowing, while the £45m contracted base already sits well above FY2025 revenue. On this view, the comparison with contracted revenue understates the value of a royalty stream that barely exists today, because the multiple falls fast once partner volumes start. The persistent selling since June can be read as placees and momentum holders exiting after a sentiment-driven spike, and much of the reported short interest sits in swaps that may hedge long positions. A single cash-backed licence would then meet a shareholder base that has already sold, which is how sharply higher prices could follow a modest piece of news. The weakness in this reading is that it needs the licence, the royalties and a friendlier rates backdrop to arrive within the same window, and the company's guidance record gives little reason to assume the timing.
