# Пов'язані статті щодо HBM

Центр новин HTX надає останні статті та поглиблений аналіз на тему "HBM", що охоплює ринкові тренди, оновлення проєктів, технологічні розробки та регуляторну політику в криптоіндустрії.

Will the Storage Industry Break the "Death Cycle" This Time?

Samsung, SK Hynix, and Micron posted record quarterly earnings, yet their stock prices fell recently due to market fears of a recurring "boom-bust" cycle. However, this expansion is focused on AI chips like HBM, not general production. Crucially, all three have signed unprecedented 3-5 year supply contracts with major clients (e.g., top data centers), featuring price floors, take-or-pay clauses, and substantial prepayments. These "SCAs" aim to lock in high-margin revenue and smooth out future volatility. The shift is driven by a severe, long-term supply shortage forecasted until at least 2028. Key constraints are EUV lithography tool delivery times (over 2.5 years) and the fact that producing HBM consumes roughly three times the capacity of standard DRAM. While quarterly price increases are slowing, analysts expect strong profits through 2027. AI server demand is squeezing consumer electronics supply, pushing prices so high that even Nvidia is reportedly considering product redesigns with less memory. Post-2028, new capacity and Chinese competitor CXMT could ease supply pressure. However, the new long-term contract structure is expected to make industry profits less cyclical than in the past, with a significant portion of revenue protected from market downturns. New demand from automotive (ADAS), CXL technology, and high-bandwidth flash (HBF) may also help absorb future supply.

marsbit5 год тому

Will the Storage Industry Break the "Death Cycle" This Time?

marsbit5 год тому

AMD Buys Taalas: Hardware AI Manages Without the Scarce HBM Memory

AMD has agreed to acquire Toronto-based startup Taalas, which addresses a major bottleneck in AI inference: the need to constantly transfer a neural network's model weights from memory to the processor for each token generated. Taalas's chips eliminate this operation by embedding the model weights directly into the transistors themselves. This data transfer is what currently limits inference speed and has made high-bandwidth memory (HBM) a critically scarce resource. Taalas, founded in 2023, has developed application-specific integrated circuits (ASICs). Its first test chip, fabricated on TSMC's 6nm process, reportedly ran Meta's Llama 3.1 8B model at speeds 48 times faster than Nvidia GPUs. The architecture features a mask ROM section for permanently stored weights and SRAM for adaptable components. AMD plans to integrate these chips into its Helios racks alongside its Instinct accelerators. However, this approach comes with a significant trade-off: each chip is permanently hardwired for a single model. Switching models requires a partial chip redesign, a process taking about two months even with Taalas's accelerated method. This limits its applicability to stable, widely-used models. The acquisition highlights a broader challenge in the semiconductor industry: the current memory shortage. HBM is sold out through 2026, and DRAM prices have surged. Yet, Taalas's technology demonstrates that this memory bottleneck is an engineering challenge, not an absolute physical limit. The industry is actively working on solutions, from Nvidia's model compression to Samsung's zHBM and new high-speed flash memory standards, all aimed at reducing reliance on scarce HBM. From an investment perspective, the deal challenges the assumption that AI-driven memory demand will keep prices permanently high. It serves as a reminder that memory has historically been a cyclical business, and current high prices are funding the very innovations designed to reduce future demand.

cryptonews.ru16 год тому

AMD Buys Taalas: Hardware AI Manages Without the Scarce HBM Memory

cryptonews.ru16 год тому

AMD acquires Taalas: hardware AI manages without scarce HBM memory

AMD has agreed to acquire Toronto-based startup Taalas, which tackles a key bottleneck in AI inference: the constant need to transfer model weights from memory to the processor for each generated token. Taalas's chips eliminate this operation by permanently embedding the model weights into the transistors themselves. This data transfer is what currently limits inference speed and has made high-bandwidth memory (HBM) a scarce commodity. Taalas's first test chip, fabricated on TSMC's 6nm process, reportedly generated tokens for Meta's Llama 3.1 8B model at speeds 48 times faster than comparable Nvidia GPUs. Its architecture features a mask ROM section for fixed weights and SRAM for adaptable components. However, this design comes with a significant trade-off: each chip is permanently dedicated to a single model. Switching models requires a partial redesign and fabrication, a process taking about two months. While the acquisition is seen as part of AMD's rivalry with Nvidia in inference, its broader implication lies in challenging the assumption of a permanent HBM memory shortage. The AI memory market is currently booming, with HBM supply sold out through 2026. Yet, Taalas's technology demonstrates that the memory bottleneck is an engineering challenge, not an absolute physical constraint. This aligns with industry-wide efforts from companies like Nvidia (through model compression) and memory makers like Samsung and SK hynix (developing new packaging and storage technologies) to reduce dependency on scarce HBM. AMD's move suggests that the current high prices for memory, driven by AI demand, may not be sustainable. It highlights a growing engineering push against the premise of perpetual memory scarcity, reminding investors that memory has historically been a cyclical business.

cryptonews.ru21 год тому

AMD acquires Taalas: hardware AI manages without scarce HBM memory

cryptonews.ru21 год тому

All Metrics Smashing Records, Yet Stock Prices Plunge Across the Board

Memory giants like Western Digital (WDC) and SanDisk (SNDK) reported blockbuster earnings in the summer of 2026, featuring毛利率 exceeding 80%, massive customer prepayments, and long-term supply agreements. Despite this seemingly perfect performance, their stocks plummeted post-earnings (WDC down 13%, SNDK down 7%), along with peers like Micron. The collapse highlights a core market rule: "good" isn't enough; results must beat already sky-high expectations. With valuations at peak "perfect asset" levels, even slightly conservative forward guidance triggered a sell-off. The market saw "peak performance" as a signal to exit. Beneath the stellar numbers, four反常 trends emerged: 1. **Financialized Pricing:** Customers provide百亿级 in upfront "interest-free deposits" to secure future capacity. 2. **Reversed Cost Curve:** Advanced DRAM (HBM4, DDR6) costs are rising per bit due to complex packaging, breaking Moore's Law. 3. **AI vs. Consumer Split:** Data center storage demand soars (+103% for SanDisk), while consumer electronics demand weakens under high costs. 4. **HDD Revival:** Hard drives, now used for AI agent context caching, see毛利率 near 55-57%. Underlying隐忧 persist. Soaring capital expenditure (CapEx) by SK Hynix and Micron risks future oversupply. Revenue growth is increasingly driven by price hikes, not surging shipment volumes (bit growth), making profits vulnerable to any price correction. In conclusion, while AI has created a long-term growth narrative, transforming storage into "strategic infrastructure," the market's violent reaction signals that peak valuations and expectations have left no safety margin. The周期 hasn't disappeared; it's merely wearing an AI disguise.

marsbit08/07 10:21

All Metrics Smashing Records, Yet Stock Prices Plunge Across the Board

marsbit08/07 10:21

With 80% of HBM Capacity but Financially Savaged: South Korea's Foundry Labor, Finally Becomes America's Fattened Slaughter Lamb

Over the past month, the South Korean stock market experienced a dramatic crash, with the KOSPI index plunging from historic highs in June to levels equating to a loss of one year's GDP. This triggered massive forced liquidations of leveraged retail investor accounts. The crisis is paradoxical given South Korea's dominant position in the high-bandwidth memory (HBM) market, with SK Hynix and Samsung collectively holding nearly 80% of global production capacity—a critical component for AI systems like NVIDIA's. The article argues that despite this manufacturing monopoly, South Korea lacks pricing power and control over demand. HBM production is entirely dependent on orders from US AI giants like NVIDIA, Google, and Meta, making Korean firms effectively high-end subcontractors. The Korean stock market's AI boom was largely a leveraged bet on these two companies, tying national retail investor fortunes directly to US capital expenditure cycles. The piece suggests this crash represents a strategic financial harvest by Wall Street. It outlines a pattern: US capital first built positions and fueled a speculative bubble through bullish narratives, attracting massive Korean retail investment, often via leveraged ETFs. Once散户 were heavily invested, a series of targeted negative events unfolded—including US antitrust lawsuits, patent investigations, and political pressure for greater profit-sharing and onshoring of production to the US. This combination precipitated the crash, allowing capital to exit at highs and potentially re-enter at lows, transferring wealth from Korean散户 to foreign investors. The article draws a parallel to the US dismantling of Japan's semiconductor dominance in the 1980s, noting the current method utilizes financial markets rather than just trade policy. The core lesson is that without controlling downstream demand, technical standards, and having a robust domestic market, even a manufacturing leader remains vulnerable. South Korea's predicament is framed as the inherent risk of a "subcontractor" model in global tech: when you work for the boss (the US tech ecosystem), you cannot be allowed to earn more than the boss. True industrial security requires mastering the entire value chain.

marsbit08/06 07:58

With 80% of HBM Capacity but Financially Savaged: South Korea's Foundry Labor, Finally Becomes America's Fattened Slaughter Lamb

marsbit08/06 07:58

Changxin Rejects Apple's Price Pressure, Prices Not Lower Than Samsung and SK Hynix, Apple Loses Pricing Power

Apple recently attempted to negotiate lower-priced DRAM procurement deals with China's CXMT (ChangXin Memory Technologies) compared to its agreements with Samsung and SK Hynix but was rejected. CXMT stated its prices would not be lower, and could even be higher, than those of the South Korean suppliers. This refusal is attributed to CXMT's production capacity being largely secured by long-term contracts with major domestic clients like Huawei, Xiaomi, OPPO, Vivo, and Chinese internet giants. Consequently, CXMT feels no pressure to meet Apple's stringent terms. The backdrop is a significant surge in memory prices driven by the AI boom. As Samsung and SK Hynix shift more production capacity towards high-margin High Bandwidth Memory (HBM) for AI servers, the supply of conventional DRAM has tightened, causing prices to skyrocket. This has drastically increased the bill-of-materials cost for devices like iPhones, pressuring Apple's profits. Apple's traditional strategy of leveraging multiple suppliers for price competition has weakened, as memory makers prioritize more profitable AI-related orders. CXMT's confidence stems from achieving technological parity. Its DDR5 and LPDDR5X products have reached mass-production yields above 90%, closely matching Samsung's performance. With technical gaps closed, CXMT no longer competes solely on low prices. Furthermore, Chinese companies are wary of the risks associated with over-reliance on Apple's supply chain, citing cases like OFILM and Wingtech, which suffered severe losses after being removed from or impacted by US sanctions. This event signals a shift in the semiconductor industry's power dynamics. The AI-driven demand has transformed the market from buyer-centric to supplier-centric, where control over scarce, advanced production capacity grants pricing power. For Chinese semiconductor firms, the episode marks a transition from being low-cost alternatives to becoming equal suppliers with their own pricing authority, backed by domestic demand and technological advancement. Apple's loss of leverage with a mainland supplier underscores this changing era.

marsbit08/06 00:41

Changxin Rejects Apple's Price Pressure, Prices Not Lower Than Samsung and SK Hynix, Apple Loses Pricing Power

marsbit08/06 00:41

Goldman Sachs Details Korea Memory "8 Key Focus Points": Valuation, Long-Term Agreements, Inventory, CXMT Impact, Buybacks, etc.

Goldman Sachs maintains a "Buy" rating on Samsung Electronics and SK Hynix, stating their recent sharp share price corrections reflect excessive market pessimism unsupported by fundamentals. The report addresses eight key market concerns: 1. **HBM Pricing:** GS forecasts HBM blended average prices for Samsung and SK Hynix to nearly double by 2027 due to persistent tight supply and a need to re-establish price premiums over standard DRAM. 2. **Long-Term Agreements (LTAs):** LTA terms are shifting in favor of suppliers, featuring longer terms, higher coverage targets (60-70% of capacity), "floor-price" protections, and significant upfront customer deposits. 3. **Inventory:** While module maker inventory is elevated, this represents a small portion of the overall market. Supplier and end-customer inventories remain healthy at low levels. 4. **NAND Supply:** Fears of a NAND oversupply are overblown. Strong enterprise SSD demand is expected to offset consumer weakness, with the supply-demand gap widening through 2027. 5. **Shareholder Returns:** GS expects both companies' actual dividends to exceed current market consensus. Share buybacks would be well-received, especially for SK Hynix to counter ADR-related dilution. 6. **SK Hynix ADR Premium:** The ADR's ~30% premium over the local share is likely persistent due to conversion constraints but aids long-term valuation re-rating by improving global investor access. 7. **SK Hynix Q2 '26 Results:** The earnings miss was due to one-time factors like slower HBM4 mix transition. Q3 is expected to see a strong recovery with higher volumes and better pricing/mix. 8. **CXMT Impact:** The competitive threat from China's CXMT is limited to the domestic market due to its technology lag and different product mix, posing minimal risk to the global supply-demand balance. GS concludes the current low valuations (approx. 3.5x 2027 P/E) imply an unrealistic lack of profit sustainability.

marsbit08/05 06:41

Goldman Sachs Details Korea Memory "8 Key Focus Points": Valuation, Long-Term Agreements, Inventory, CXMT Impact, Buybacks, etc.

marsbit08/05 06:41

Despite the sell-off, Goldman Sachs remains bullish on Samsung and SK Hynix. Here's why.

Goldman Sachs maintains "Buy" ratings on Samsung Electronics and SK Hynix despite recent stock declines. Its bullish view centers on three core arguments. Firstly, it expects HBM (High Bandwidth Memory) pricing to re-establish a premium over conventional DRAM by 2027, with a projected blended ASP of around $2.9/Gb. This is driven by tight supply-demand dynamics, increasing manufacturing complexity for newer HBM generations, and the need to restore its historical price premium. Secondly, the volatility of the memory cycle is expected to moderate due to widespread adoption of 3-5 year Long-Term Agreements (LTAs) with key customers. These contracts, covering a significant portion of planned capacity, feature mechanisms like price floors, prepayments, and penalties, reducing supplier risk and improving earnings visibility. Thirdly, inventory levels remain low at key suppliers and major customers, providing a buffer against a sharp downturn. Furthermore, robust demand from enterprise SSDs for AI servers is seen offsetting weakness in consumer segments like smartphones and PCs, preventing NAND markets from slipping into oversupply in the near term. While risks exist—such as potential weaker AI demand or aggressive capacity expansion—Goldman Sachs believes the combination of HBM repricing, LTAs, and low inventory underpins a more stable earnings outlook for the leading Korean memory makers.

marsbit08/05 04:11

Despite the sell-off, Goldman Sachs remains bullish on Samsung and SK Hynix. Here's why.

marsbit08/05 04:11

活动图片