DRAM is now pricier than compute silicon
The seemingly immutable economics of the semiconductor race are taking a surprising turn, signaling a significant shift in how value is being extracted from cutting-edge silicon. A new analysis suggests that the relationship between memory manufacturing and advanced logic processing is flipping, challenging long-held assumptions about which sector drives the most lucrative economics in the foundry world.
Specifically, the data points toward a startling finding: DRAM chip makers may currently be achieving higher revenue per unit of silicon area from their wafers than TSMC earns on its demanding N3 (3nm) logic process.
This observation represents a notable reversal of the historical dynamic that characterized the industry. For decades, the economics of leading-edge compute manufacturing—driven by complex logic nodes—have often set the benchmark for profitability. However, this latest insight suggests that the highly specialized, high-density memory sector is now outpacing the profit generation of the most advanced logic fabrication.
What this shift implies is that the current value being generated in the memory space is more potent on a per-area basis than the complexity of the newest compute processes. It reframes the strategic importance of memory, positioning it not just as a necessary component, but as a powerful economic engine in the relentless pursuit of technological advancement.
This reversal isn’t just an academic curiosity; it has profound implications for supply chain strategies, capital allocation, and the future balance of power within the semiconductor ecosystem. As the industry continues to push the boundaries of nanometer technology, understanding this new economic relationship is crucial for predicting where the next wave of market value will be created.