Wall Street rarely puts a triple-digit return forecast on a company with a long history of being unprofitable. Goldman Sachs just did exactly that.The bank named U.K. fuel cell developer Ceres Power (CWR) as a top European pick for August, forecasting a 168% gain over the next 12 months. That would nearly triple the money of anyone buying today.Ceres shares fell more than 31% in July as investors pulled back from artificial intelligence spending. Goldman sees that drop as a buying window rather than a warning.For investors, Goldman thinks Ceres will help solve one of AI’s biggest problems: where the electricity comes from. Whether that outlook holds depends on partners, timelines, and cash the company does not yet have.Why Goldman Sachs sees 168% upside in Ceres Power stockGoldman analyst Michele Della Vigna reiterated his Buy rating on Ceres with a 930p price target on July 4, according to The Globe and Mail. Ceres closed around 383p on Aug. 5, which is where that 168% figure comes from.The stock has been volatile all year. It trades up more than 168% year to date, yet it sits well below its 52-week high of 872.50p after the July sell-off.More AI Power Stocks:Morgan Stanley says Bloom can withstand an Oracle project delayAnother AI power darling just hit a 52-week lowMorgan Stanley strongly resets GE Vernova stock targetDella Vigna was among the first major analysts to connect fuel cells to AI power demand, and he has raised his Ceres target repeatedly through 2026.Goldman also placed Ceres near the top of its August European Conviction List, the bank’s roster of highest-confidence Buy calls, CNBC reported. Of all the names on that list, Ceres carries the largest forecasted gain.What Ceres Power actually does, and why AI needs itCeres does not make electricity. It designs solid oxide fuel cell technology and licenses it to manufacturers that build the actual systems.A solid oxide fuel cell converts fuel such as natural gas or hydrogen into electricity through a chemical reaction rather than combustion. That means power generated on-site, quickly, without waiting years for a grid connection.That speed is the whole point for AI. Data centers running AI models consume enormous amounts of electricity, and utilities cannot build new grid capacity fast enough. Goldman expects AI to push global data center power demand up by about 160% by 2030, Investing.com reported.Ceres earns money through licensing and royalties instead of factories, an approach analysts call “asset-light.” Goldman expects that model to drive gross margins above 80% as royalty revenue scales.
AI data centers are straining power grids, and Goldman Sachs believes fuel cells could help fill the gap.Bloomberg / Getty Images
The partners who decide whether the Ceres thesis worksSince Ceres licenses its technology, its revenue depends heavily on what its partners do. South Korea’s Doosan Fuel Cell and Taiwan’s Delta Electronics are both developing and mass-producing fuel cell systems built on Ceres technology, aimed squarely at data centers.That structure creates a specific risk. If Doosan or Delta slow their production plans, Ceres royalty income slows with them, regardless of how strong AI demand looks on paper.Ceres and Delta have targeted the start of commercial production by the end of 2026. That timeline is the single clearest milestone investors can watch to judge whether the Goldman thesis is on track.The risks behind the 168% Ceres Power forecastThe bull case is real, but so is the reason Ceres is called a controversial pick. The company loses money and is expected to keep losing it for the next few years.Goldman’s analysts have predicted negative earnings per share through 2027, reflecting near-term revenue weakness, even as they forecast strong long-term growth.Related: JPMorgan resets oil price target for rest of 2026That combination defines the bet. Investors are paying today for royalty income that mostly arrives between 2028 and 2030.Here is what has to go right for the 168% target to pay off:Doosan and Delta must hit their data-center production timelines.Commercial output must begin on schedule by late 2026.Royalty revenue must scale fast enough to lift margins toward the 80%-plus range Goldman models.AI-driven power demand must stay strong through the end of the decade.If any of those slip, the target date and the return both move further out.Ceres Power stock vs. the broader AI power tradeCeres is not the only way to invest in AI’s electricity problem, and comparing it to peers shows how much extra risk sits inside the Goldman call.Ceres also has a U.S. rival, Bloom Energy (BE), which already supplies fuel cells to data centers through partners like Oracle. That means Bloom has real revenue today. Ceres does not, at least not yet at scale.Larger power names such as GE Vernova (GEV) and Constellation Energy (CEG) offer the same AI demand theme with real profits behind them, which makes them steadier but far less explosive.Ceres offers the largest forecasted gain on Goldman’s European list, and it also carries the least proven earnings.What investors should take from the Goldman callGoldman’s 168% forecast is a genuine high-conviction call from an analyst who called the fuel cell and AI connection early. It is not a guarantee.For investors comfortable with risk, Ceres offers direct exposure to the 2026 through 2030 data center rollout, and the July drop has lowered the entry price. Position sizing is very important here, because a pre-profit stock can fall as fast as it rose.For more cautious investors, the smarter move may be to wait for proof. If Doosan and Delta begin commercial production on schedule by late 2026, the thesis gets a concrete data point that today’s price does not yet reflect.Related: Goldman Sachs sees an oil glut coming, but don’t expect much relief at the pump