The semiconductor industry as a whole is experiencing a crisis-level shortage of DRAM. This one fixation is wreaking havoc on data centers and tech companies all over the world. We commend Samsung Electronics for taking bold steps to respond to the crisis. They are planning to open a new advanced manufacturing plant in Pyeongtaek, South Korea by 2028. This effort coincides with a surge in demand for DRAM chips, especially ones catered to artificial intelligence (AI) data centers.
Beyond just expanding manufacturing capabilities through their historic investment, Samsung has stepped up in the realm of research and development too. In 2024, its researchers inked one scientific win after the other, including proving the production of a 16-high stack with hybrid bonding technology. This technological breakthrough more than doubles the efficiency of production. On top of that, it sets the stage for higher capacity as Samsung claims stacking as many as 20 dies is possible.
These big developments come at a time when global demand for DRAM has never been higher. At the same time, there are nearly 2,000 new data centers planned or currently under construction, which is putting more pressure on the need for new DRAM production. The road ahead is filled with challenges. One of the largest hurdles is the gigantic cost of building new fabrication plants, sometimes well over $15 billion.
Challenges in Expanding Production Capacity
The investment landscape for semiconductor manufacturing has become more risk-averse. Further, major manufacturers are hesitant to increase capabilities because of the large financial burden that comes with building new fabs. Supply chain expert, Thomas Coughlin, shares a critical point of understanding. He illustrates that the opportunities to expand—i.e., capital—always seem to be available at a firm’s disposal during economic booms. This reluctance is complicated by the reality that it currently takes a new fab 18 months or more to stand up.
In 2023, Samsung made a huge cut of 50 percent. This courageous action was taken to prevent further inflation and lower prices below cost of production. This strategic decision shines a light on the precarious dance that semiconductor multinationals must perform between satiating high demand and worrying about profitability. Combine that with the lack of investment in new production capacity through 2024 and most of 2025, and you really have a recipe for disaster.
“In general, economists find that prices come down much more slowly and reluctantly than they go up. DRAM today is unlikely to be an exception to this general observation, especially given the insatiable demand for compute.” – Mina Kim
Samsung’s production strategy reflects a broader trend in the industry as firms grapple with the ongoing supply chain disruptions that began during the COVID-19 pandemic. The chip supply panic of that period has had enduring effects on how companies have decided to expand and invest.
Innovations in DRAM Technology
While these challenges are great, innovators continue to be the strongest determinant in pushing through supply barriers. Mina Kim emphasizes that there are two primary ways to address the current DRAM shortages: through technological advancements or by building more fabs. Samsung’s recent breakthroughs in hybrid bonding technology may give it a leg up here.
Its newest product, the B300, is powered by 8 high-bandwidth memory (HBM) chips. Every chip consists of a 12-high stack of DRAM dies, yielding phenomenal performance. These types of innovations can deliver performance improvements both today and within the pipeline, helping memory technology avoid future pitfalls and ensuring a bright trajectory moving forward.
What’s even more groundbreaking, as Samsung’s researchers explained to us, is the HBM4 standard’s support for 16 stacked DRAM dies. It’s a big jump in memory architecture progression! Improvements have been made in both aspects since the initial launch of their API. Reaching 20 dies per stack paves the way for thrilling opportunities to address the new demands of advanced AI applications.
“Relief will come from a combination of incremental capacity expansions by existing DRAM leaders, yield improvements in advanced packaging, and a broader diversification of supply chains.” – Shawn DuBravac
>Moreover, as NVIDIA’s data center revenue soars from under $1 billion in late 2019 to an estimated $51 billion by next October, demand will continue to grow. This increasing demand augments the urgency on DRAM manufacturers to develop new production technologies and ramp up their capacities.
The Long Road Ahead
Looking towards the future, industry leaders anticipate that meaningful relief from current shortages may not materialize until Samsung’s new facility becomes operational in 2028. Intel CEO Lip-Bu Tan puts it succinctly, “There’s no relief until 2028. This timeline serves to highlight the high stakes involved as well as the urgent need for both speed in response and strategic vision across the semiconductor ecosystem.
Industry insiders contend that even if successfully constructed, new fabs wouldn’t come close to providing a global supply and demand equilibrium. Obstacles notwithstanding, this expansion has the opportunity to maximize capacity by 20 percent. This depends almost entirely on the basis of firms getting past their reticence to invest, in the face of continued pervading economic uncertainty.
The semiconductor industry stands at a crossroads. As companies like Samsung push forward with ambitious plans and innovative technologies, the fate of DRAM production hinges on balancing immediate demands with sustainable growth strategies.

