Samsung Sees Memory Chip Shortages Worsening Through 2028 as AI Demand Surges

Samsung Forecasts Extended Supply Constraints Despite Record Profit Growth

Samsung Electronics warned on Thursday that memory chip shortages will intensify and persist through 2028, delivering the stark assessment even as the world’s largest memory chipmaker reported a more than 250-fold jump in chip division profit. The announcement counters growing investor concerns that aggressive AI spending by major technology firms may decelerate and curb growth across the semiconductor sector.

The company’s shares surged as much as 8% in early trading before retreating to close down 1.1% on Thursday. The volatile session reflected mixed investor sentiment about the sustainability of AI infrastructure buildout and competitive pressures from China that could threaten future chip earnings.

Jaejune Kim, executive vice president of Samsung’s memory business, told analysts during an earnings call that “The supply shortage in 2027 is expected to worsen compared to this year, and it is expected to continue in 2028.” The forecast underscores the structural mismatch between exploding AI demand and the semiconductor industry’s constrained production capacity.

Long-Term Supply Agreements Lock in Major Data Center Customers

Samsung revealed it has signed long-term supply agreements with the top five global data center firms and is nearing deals with five other large companies, though the electronics giant declined to disclose their names. These strategic partnerships mark a shift in how memory chips reach AI infrastructure developers.

Kim explained that these long-term deals would last at least five years and account for 60% to 70% of its total capacity for the longer term. The agreements include upfront payments and floor pricing aimed at hedging the risks of its capital investments, providing Samsung with revenue stability while guaranteeing supply for critical customers.

The company noted that frontier artificial intelligence model developers have recently approached Samsung directly to express intent to purchase memory and request multi-year supply contracts. This represents a significant change in the AI value chain, as memory chips are typically supplied to AI chip developers like Nvidia, with cloud service providers then building large-scale AI servers that model developers use.

Ryu Young-ho, a senior analyst at NH Investment & Securities, described the outlook as encouraging, stating “Management’s commentary on the conference call was better than expected, and it was one of the more reassuring calls we’ve heard in quite some time.”

Record Semiconductor Profits Offset by Mobile Division Losses

Samsung’s semiconductor division posted an operating profit of 89.2 trillion won ($61.7 billion) in the second quarter, representing a dramatic recovery from a year earlier. This second-quarter profit surpassed Samsung’s combined earnings over the past three years, marking a sharp turnaround as the company races to catch up with Korean rival SK Hynix in supplying high bandwidth memory (HBM) chips used in AI processors.

However, the surge in chip prices created a paradox within Samsung’s business portfolio. The company’s mobile division reported a 700 billion won loss, its first quarter in the red, as soaring memory costs squeezed profitability in smartphone manufacturing.

Josh Gilbert, an analyst at eToro, captured the dilemma facing the conglomerate, noting “The chips enriching one side of Samsung are now hurting the other, leaving the group more exposed than ever to memory pricing and the durability of hyperscaler demand.” The comment highlights how Samsung’s diversified business model creates both opportunities and vulnerabilities in the current market environment.

HBM Revenue Set to Triple Amid AI Processor Demand

Samsung, which counts Nvidia and Advanced Micro Devices among its HBM customers, expects HBM4 revenue to more than triple in the third quarter. This explosive growth trajectory would help bring the company’s HBM market share closer to that of SK Hynix, which currently leads the high-performance memory segment critical to AI applications.

The company’s bullish view on long-term supply constraints stems from the extended lead times required for capacity expansion. Samsung explained that considering the more than three and a half years required from building a new fabrication facility to producing wafers, expanding supply through capacity additions will inevitably take considerable time.

The company emphasized that significant supply increases cannot be expected until 2028, as the physical infrastructure required for semiconductor manufacturing creates inherent delays in responding to demand surges. This multi-year constraint provides Samsung with pricing power while simultaneously creating supply security concerns for technology firms dependent on advanced memory chips.

Market Context and Competitive Pressures

Samsung’s optimistic long-term outlook follows a sharp slide in chip stocks in recent months driven by investor concerns about funding for the AI infrastructure buildout. Questions about whether technology giants can sustain current investment levels and competitive threats from Chinese semiconductor manufacturers have created uncertainty in the sector.

As the entire AI value chain grapples with supply shortages, AI model developers-the end customers-are now pursuing direct supply contracts with Samsung Electronics to expand their AI servers independently. This disintermediation reflects the strategic importance of securing guaranteed memory supply for companies betting their futures on artificial intelligence capabilities.

Samsung’s strategy of expanding long-term agreements provides the company with predictable revenue streams and capacity utilization while offering customers supply certainty in a constrained market. The upfront payments and floor pricing mechanisms protect Samsung’s massive capital investments in fabrication facilities while giving buyers assurance that they can access the memory chips essential to their AI ambitions.

The semiconductor giant’s forecast that supply shortages will intensify in 2027 compared to 2026 and persist into 2028 signals that the current market dynamics favoring memory producers will likely continue for years. This extended timeline reflects both the surging demand from AI applications and the physical constraints of semiconductor manufacturing capacity expansion.