AI triggers an 886% surge in DRAM prices; Meta's massive acquisition of startup Manus creates a butterfly effect; NVIDIA bets $5 billion on Intel stocks
The drastic fluctuations in the global memory market are triggering a chain reaction. As of December 2025, the spot price of DDR4 8Gb DRAM has skyrocketed to $14.10, an increase of 886% year-on-year, setting a record high. Behind this price surge lies the dual pressure of supply-demand imbalance and technological transformation.
On one hand, the explosive demand for AI computing power has driven the three major DRAM manufacturers (Samsung, SK Hynix, and Micron) to shift production capacity toward High Bandwidth Memory (HBM), leading to a sharp reduction in traditional DDR4 supply. On the other hand, the surging demand for large-capacity SSDs in data centers, combined with the industry trend of "HDD to SSD transition", has further exacerbated the shortage of memory chips. Japanese PC manufacturer Mouse Computer was forced to suspend sales due to out-of-control memory costs, while Sharp subsidiary Dynabook stated bluntly that "price increases have exceeded the limit of affordability".
The transmission effect across the industrial chain is becoming evident: NAND flash giants such as Kioxia are shifting to the data center market, restricting PC supply; IDC data indicates that leading manufacturers like Lenovo and HP may need to raise prices by 10%-20% to maintain profitability. More notably, memory cost pressures on consumer electronics products like the iPhone may force Apple to reevaluate its pricing strategy, signaling an impending industry-wide price hike.
Meta's recent two major transactions have revealed its in-depth strategy in the fields of AI and metaverse. On December 30th, Meta announced the acquisition of Singapore-based AI startup Butterfly Effect for "billions of dollars". Its core product, Manus, is an AI agent designed for small and medium-sized enterprises, offering services such as resume screening and itinerary planning, with an annualized revenue of $125 million. This acquisition is not only Meta's third-largest acquisition in history (after WhatsApp and Scale AI) but also marks a significant breakthrough in its AI commercialization path.
Almost simultaneously, Meta upgraded its Interaction SDK and launched a "hand-first" interaction solution, including an improved teleportation gesture, natural climbing mode, and physics engine movement system. Although foreign media RoadtoVR noted that users still prefer traditional controller operations, Meta is clearly betting on the future—by lowering barriers for developers, it aims to accelerate the construction of an immersive ecosystem centered on gesture interaction.
These two events reflect Meta's dual objectives: in the short term, to quickly monetize AI investments through Manus's subscription model; in the long term, to reshape the user experience of XR devices through interaction innovation. Zuckerberg's "All in AI" strategy is shifting from technological investment to ecosystem harvesting.
The PC industry crisis triggered by DRAM price hikes has exposed the vulnerability of hardware manufacturers. Japanese market data shows that memory chips account for 30% of PC costs, and the depreciation of the yen has further amplified the premium pressure on imported components. Faced with the dilemma of "raising prices to protect profits" or "sacrificing market share for growth", manufacturers are exploring new paths.
Lenovo has adopted a "medium-to-long-term price adjustment + cost management" strategy, attempting to absorb costs through scale effects; HP has remained silent, observing competitors' moves. More disruptive is Kioxia's strategic shift—abandoning the consumer market to focus on data center and enterprise-grade SSDs. Although this "abandoning the lesser to preserve the greater" strategy has exacerbated PC supply shortages, it highlights the new logic in the memory industry: "whoever wins the data center wins the world".
At the same time, tech giants like Apple and Google are bypassing intermediaries to purchase directly from DRAM manufacturers, revealing the restructuring of supply chain power dynamics. As "hardcore players" gain control over pricing power, the survival space for small and medium-sized manufacturers may be further squeezed.
Despite undercurrents in the industry, the capital market presents a mixed picture. On December 29th, US-listed large-cap tech stocks generally fell in pre-market trading, with NVIDIA and Tesla dropping over 1%, and Meta and Apple following suit. This stands in stark contrast to the optimistic sentiment triggered by NVIDIA's investment in Intel, reflecting market divergence on the outlook for the technology industry.
In the short term, memory price hikes and geopolitical risks have increased uncertainty. Omdia analysts point out that the PC price hike cycle may extend into 2026, dragging down consumer electronics demand; while Taiwanese manufacturers are struggling to increase production, supply chain recovery will take time. However, from a long-term perspective, the AI-driven computing power revolution remains the core theme—the surge in HBM demand, strong shipments of AI servers, and the commercialization of AI agents are all reshaping the industry value chain.
Meta's acquisitions, NVIDIA's contrarian investment, and the strategic shifts of memory giants together outline a "fire and ice" landscape of the technology industry: the old order is teetering under cost pressure, while a new ecosystem is accelerating its formation catalyzed by AI and interaction innovation. In this transformation, only players that master core technologies and flexibly adjust their strategies can navigate the cycle and win a ticket to the next decade.
