ChangXin Memory Forecasts Record Profit as AI-Driven DRAM Boom Accelerates
TREE NEWS reports: ChangXin Memory Technologies has issued a preliminary earnings forecast projecting net profit attributable to shareholders of 131 billion to 132.5 billion yuan for the first three quarters of 2026, a dramatic reversal from a loss of 5.28 billion yuan in the same period a year earlier. The swing represents a profit improvement of more than 100 billion yuan, marking a historic turnaround for China’s leading domestic memory chipmaker. Revenue for the period is expected to reach approximately 265 billion to 270 billion yuan, up more than sevenfold from 32.08 billion yuan a year ago — an increase of over 230 billion yuan. On a per-day basis, the implied profit run-rate equates to roughly 480 million yuan daily.
AI Compute Demand Is the Core Driver
The company attributed the surge to rapidly growing AI compute demand, which has driven up global DRAM industry conditions since the second half of 2025. Products are in short supply, prices have risen sharply, and the company’s capacity utilization has remained at high levels, allowing economies of scale to materialize. ChangXin also cited steady progress in capacity construction, product research and development, and process technology upgrades, which have improved output per wafer and overall utilization. Product mix optimization and deeper supply-chain collaboration further strengthened profitability.
Market Implications: Semiconductors, Memory Pricing and the AI Trade
The results underscore how the AI infrastructure buildout is reshaping the memory market. DRAM has shifted from a cyclical commodity business to a supply-constrained growth engine, with pricing power resting firmly with producers. For equity investors, the read-through extends across several fronts:
- Memory and semiconductor equities: Strong DRAM pricing supports earnings for global memory producers and their equipment suppliers. ChangXin’s scale-up may intensify competition in the medium term, but near-term industry pricing remains supportive.
- AI infrastructure names: Sustained demand for high-bandwidth memory and server DRAM suggests AI capex is not slowing. This reinforces the bull case for compute hardware, data-center suppliers, and the broader AI trade.
- Downstream cost pressure: Rising DRAM prices raise input costs for device makers, module assemblers, and consumer electronics firms, potentially compressing margins for buyers of memory.
- China’s semiconductor self-sufficiency: ChangXin’s performance highlights the scale of China’s domestic memory ambitions, a theme relevant to export-control policy, supply-chain diversification, and long-term competitive dynamics.
Buyback Signals Management Confidence
Chairman Zhu Yiming has proposed a share repurchase of A-shares worth between 1 billion and 2 billion yuan, funded by internal or self-raised capital, with a maximum price of 150% of the 30-day average trading price preceding board approval. The proposal, disclosed alongside the earnings forecast, is typically read by the market as a signal that management views current valuation and future prospects favorably. While the buyback size is modest relative to the profit scale, its timing — paired with a blockbuster earnings beat — amplifies its signaling value.
Key Takeaways for Investors
- AI-driven DRAM demand has produced an extraordinary earnings inflection, validating the memory upcycle thesis.
- Watch memory pricing trends and capacity utilization as leading indicators for the sustainability of this cycle.
- Consider second-order effects: beneficiaries in equipment and AI hardware versus margin pressure for downstream device makers.
- The buyback proposal, though limited in size, reinforces management’s confidence and may support near-term sentiment.
- Monitor regulatory and export-control developments, as China’s memory self-sufficiency push carries geopolitical implications.
The scale of the turnaround — from a 5.28 billion yuan loss to over 130 billion yuan in profit in a single year — illustrates how forcefully the AI cycle is reshaping semiconductor economics. For investors, the question is no longer whether the DRAM upcycle is real, but how long supply constraints and pricing power can persist.
