Bloom Energy Confronts the Scandium Question as Earnings Near

Bloom Energy reports Q2 earnings amid short-seller allegations of China supply-chain dependency; stock up 883% on AI data center demand but faces scandium sourcing scrutiny.

Bloom Energy’s next earnings report, due July 28 after the market close, arrives under an unusual cloud. The fuel-cell maker’s stock has more than octupled over the past twelve months, powered by surging demand from AI data centers. But a short-seller attack has cast doubt on whether the company can secure the raw materials needed to sustain that growth — and next week’s quarterly update will be the first chance for management to address the criticism head-on.

The controversy began on July 8, when Hunterbrook Capital published a report titled “Bloom’s Big Lie,” accusing the company of downplaying its reliance on Chinese suppliers for scandium oxide, a key component in its solid-oxide fuel cells. Shares cratered roughly 12% that day. Bloom Energy fired back, calling the allegations “false and misleading” and pointing to its audited SEC filings. By July 14, the stock had recovered some ground, climbing 4.2% to $243.40, though it still posted a weekly loss of 4.27% and a monthly decline of 10.13%.

Hunterbrook’s central argument is a supply-chain math problem. Bloom aims to scale its annual installation capacity from roughly 1 gigawatt today to 5 gigawatts — a target the short seller calls “physically and commercially unachievable.” The report calculates that hitting 5 GW would require about 220 metric tons of scandium oxide per year, against a total estimated global annual production of just 240 tons. The short seller also released follow-up research, including photos from a Bloom supplier conference in May 2026 that it says show an ongoing relationship with a Chinese supplier — contradicting CEO KR Sridhar’s public statements that the company has “no China supply chain.”

Wall Street, however, has largely looked past the allegations. Clear Street lifted its price target from $250 to $290 while keeping a “Hold” rating, citing expected revenue growth of 84% in the current fiscal year and 67% the following year. Susquehanna also raised its target, and RBC Capital maintained its “Outperform” rating. The bullish view rests on the same narrative that has driven the stock’s 883% gain over the past twelve months and 155% year-to-date increase: AI data centers’ insatiable need for electricity.

Bloom’s fuel cells deliver constant, high-density power — exactly what hyperscalers require for around-the-clock computing. Brookfield recently expanded its AI infrastructure financing commitment from $5 billion to $25 billion, with a broader fund targeting up to $100 billion. The company is widely seen as a primary beneficiary of this buildout, especially as grid constraints push data center operators toward on-site generation.

Yet the enthusiasm is not universal. The stock’s relative strength index sits at 43.1, signaling neutral momentum after retreating from overbought territory. It trades between its 50-day moving average near $243.45 and its 100-day average around $192.84, with annualized 30-day volatility above 111%. More tellingly, insiders have sold $59.8 million worth of shares over the past three months without a single insider purchase — a pattern that often gives retail investors pause.

The geopolitical backdrop adds another layer. The U.S. Department of Defense has flagged critical vulnerabilities in the supply of minerals where China dominates the global market, noting that 78% of American weapons systems are exposed to potential disruptions. Canada has launched a C$2 billion Critical Minerals Accelerator to boost domestic production. Scandium sits squarely within that category, putting Bloom’s sourcing practices under a microscope that extends beyond the short-seller report.

The July 28 earnings call will test whether Bloom can provide concrete details on its supply chain — and whether those details reassure investors that the growth story is on solid ground. Until then, the stock occupies an uncomfortable middle ground: buoyed by a transformative AI tailwind, but weighed down by questions that a single quarterly filing may not fully resolve.