Middle Eastern sovereign AI funds have officially entered the memory procurement market, reshaping the supply-demand landscape and accelerating price increases. Concurrently, the recent release of the Kimi K3 model, a hot topic in AI circles, has been interpreted by a Korean securities firm as a medium-term positive for AI hardware, rather than the demand shock feared by the market.

According to a research report released by Meritz Securities on July 19, sovereign AI investment entities from Saudi Arabia and other Middle Eastern nations have recently begun negotiations with South Korean memory manufacturers such as Samsung Electronics Co., Ltd. and SK Hynix Inc. regarding medium- to long-term procurement plans, inquiring about supply capacity and volume for the coming years. The report forecasts that the sequential increase in server DRAM contract prices for the third quarter of 2026 will exceed the market's previous expectation of approximately 15%. Suppliers who adopted more lenient pricing strategies in the second quarter are expected to implement particularly significant price hikes in the latter half of the year.

Meanwhile, clear signals of price increases have already emerged in the spot market. The spot price for 64GB DDR5 server DRAM has surged sharply since mid-July, with recent quotes ranging between $3,100 and $3,400. This represents an increase of about 146% compared to the contract price of around $1,380 at the end of June. The price surge is especially pronounced for high-end products with a bus speed of 6400 Mbps.

Amidst the dramatic shift in the memory market's supply-demand dynamics, the investment strategy team at Meritz Securities, in a separate report released the same day, characterized the launch of the Kimi K3 model by Moonshot AI as a medium-term positive for AI hardware demand. They advised investors to buy shares in AI component companies such as semiconductors and memory, while reducing holdings in large-cap tech stocks like Alphabet Inc. (Google) and Microsoft Corporation. Meritz argues that the market's judgment that "Kimi K3 will reduce AI hardware demand" is based on two major misconceptions. Instead, K3 is expected to bring incremental demand for hardware manufacturers from a medium-term perspective.

Entry of Middle Eastern Sovereign AI Funds: A Single Buyer Can Shift the Supply-Demand Balance

Analysts at Meritz Securities noted in the report that the core logic behind Middle Eastern sovereign AI investment lies in national security and data sovereignty—countries desire to build data centers within their own territories and operate AI systems using their own languages and data autonomously. As these investment plans become more concrete, the supply tightness expected for 2027 is materializing earlier.

The current structural changes in the memory market mean that the entry of a single large buyer can have a powerful impact on prices. The entities building AI data centers are highly concentrated, meaning the entry of a single procurement party can instantly change the market demand fulfillment rate by 3 to 5 percentage points. The demand fulfillment rate, which is the ratio of actual supply to market-required volume, can trigger significant price jumps with only minor fluctuations. Currently, large customers account for over 70% of memory manufacturers' revenue, indicating that pricing power is highly concentrated among a small number of buyers.

This is fundamentally different from the dispersed procurement structure of the past smartphone and PC era. Back then, numerous manufacturers made small-volume purchases, and order changes from a single client had limited impact on the overall market. Today, the centralized procurement model of AI data centers means that each new sovereign-level buyer entering the market can become a decisive variable for prices.

Supply Shortage Extends to End Products; Volume, Not Price, is the Core Issue

Meritz points out that the impact of the memory shortage has spread from the data center side to consumer electronics end products. Both Apple Inc. and Chinese smartphone manufacturers are unable to procure sufficient memory ahead of the peak season in the second half of the year, leading to production shortfalls in their fourth-quarter plans.

The analyst emphasizes that the essential problem facing the market is "volume allocation" rather than a simple "price" issue—"In short, there isn't enough supply; everyone is scrambling to buy." Addressing cyclical concerns circulating in the market, he believes this judgment is somewhat off the mark. The current tightness is rooted in structural supply insufficiency, not short-term demand fluctuations.

The report also notes that some suppliers adopted relatively flexible, low-price strategies in the second quarter to accommodate client needs. This implies these manufacturers have greater room to raise contract prices in the third and fourth quarters, and their pricing elasticity will be more pronounced.

Kimi K3 Launch Interpreted as Positive for AI Hardware Demand

The investment strategy team analyst at Meritz Securities, in the report released the same day, characterized the launch of the Kimi K3 model as a medium-term positive for AI hardware demand. The recommendation was to invest in AI component companies like semiconductors and memory, while reducing exposure to large-cap tech stocks such as Alphabet Inc. and Microsoft Corporation.

Meritz clearly distinguishes this event from the impact of DeepSeek last year. The core narrative of DeepSeek was completing AI training at an extremely low cost of $6 million. In contrast, Kimi K3 has not disclosed its training cost or GPU usage. Furthermore, the official statement explicitly indicates that running K3 requires a large cluster of at least 64 high-performance chips, which is not a model solvable by low-cost hardware. In terms of usage cost, according to data from AI performance evaluator Artificial Analysis, K3's cost per task processing is $0.95, placing it in the same order of magnitude as GPT-5.6 Sol ($1.04) and Claude Fable 5 ($2.75), but a full order of magnitude higher than DeepSeek V4 Pro ($0.04).

The analysis further states that the logic that "K3 will reduce AI hardware demand" is based on two misconceptions. First, AI infrastructure investment is not solely led by large technology companies. Taking NVIDIA Corporation's data center chip sales data as an example, the share of large cloud vendors is declining, while the proportion from mid-to-small AI cloud service providers and ordinary enterprises continues to rise. Second, the inference services for Chinese AI models actually run extensively on U.S. hardware. Inference service providers like Fireworks AI and Together AI provide services to enterprise clients based on American chips.

The analyst concludes, "Whether closed-source or open-source, all AI business participants will ultimately need to use hardware companies, which will gain medium-term incremental demand from the K3 release."