Hardware Watch: Nvidia's Memory Warning, Marvell's Connectivity Bet & the Robotics Hardware Race

Hardware Watch · November 18, 2026

The Big Headline: Intel vs. AMD, Nvidia's Memory Warning, Marvell's Connectivity Bet and the Robotics Hardware Race


Today's Overview

The Hardware Layer Is Being Repriced — by Memory, by Connectivity and by Geopolitics

Intel reclaimed overall x86 CPU market share for the first time since 2025, with desktop gains offsetting AMD's record server and mobile growth, according to Q3 2026 shipment data. Nvidia warned that soaring memory prices are weighing on demand for consumer PCs, even as its AI hardware business continues to expand. Marvell scaled its connectivity portfolio with optical interconnect, CXL memory expansion, and 1.6T switching aimed at the AI data center. Silicon Motion targeted enterprise SSDs with PCIe Gen5 and Gen6 controllers for AI inference workloads. TSMC and Taiwan's supply chain prepared for the next phase of AI chip manufacturing, with CoWoS expansion and supplier development. Samsung expanded NAND manufacturing in Vietnam as AI infrastructure reshaped memory demand. New US chip export controls put Southeast Asia's AI server supply chains under pressure. And Genesis AI built its own robotics hardware and intelligence stack following a $500 million funding round.

Editorial Note: This edition covers developments from November 12–19, 2026. All sources are English-language. Financial figures are company-reported and unaudited unless otherwise stated. Announced agreements are distinguished from completed transactions. This article does not provide investment advice.

Processors & Computing

1. Intel vs. AMD — CPU Market Share Shifts

Intel Reclaims Overall x86 Share for First Time Since 2025 as Desktop Gains Offset AMD's Server Growth

Reported November 16, 2026 · Status: Q3 2026 shipment data.

What happened: Intel reclaimed overall x86 CPU market share for the first time since 2025, according to Q3 2026 shipment data. Massive desktop gains offset AMD's record server and mobile growth. AMD continued to expand in server processors, where it has been gaining share against Intel's Xeon lineup.

Key players: Intel, AMD, PC OEMs, data center operators.

Why it matters: The shift illustrates the complexity of the CPU market. Intel's recovery in desktop share suggests that its product roadmap and pricing are resonating with PC buyers, particularly as memory costs rise and consumers become more price-sensitive. AMD's continued growth in servers shows that its EPYC processors remain competitive in the data center, where performance-per-watt and total cost of ownership matter most. The two companies are winning in different segments, and the overall share figure masks that divergence.

Source: Tom's Hardware.

GPU & AI Computing

2. Nvidia — Memory Prices Weigh on Consumer PC Demand

Rising Memory Costs Pressure Consumer PC Affordability as Blackwell Workstation and AI Demand Remains Strong

Reported November 18, 2026 · Status: company statement during earnings call.

What happened: Nvidia warned that soaring memory prices are weighing on demand for consumer PCs. The company's latest financial and hardware developments show continued strength in AI computing, but rising memory costs are pressuring consumer PC affordability. Blackwell workstation and AI computing demand remains robust, and the company's exposure to physical AI and robotics is growing.

Key players: Nvidia, memory suppliers, PC OEMs, consumer and enterprise buyers.

Why it matters: The warning reflects a broader tension in the hardware market: the AI boom is consuming memory capacity, which drives up prices for all memory products, including those used in consumer PCs. For Nvidia, the impact on its consumer GPU business is a secondary concern relative to its data center business, but it is a reminder that the company is not immune to the supply chain dynamics its own products help create.

Source: MarketWatch.

Networking & Connectivity

3. Marvell — Scaling Connectivity for AI Infrastructure

Optical Interconnect, CXL Memory Expansion and 1.6T Switching Target the AI Data Center Bottleneck

Announced November 13, 2026 · Status: announced; products available.

What happened: Marvell announced a scaled connectivity portfolio for AI infrastructure, including optical interconnect and photonic fabric technologies, CXL memory expansion, and high-speed switching up to 1.6T. The announcement was made ahead of SC26 and targets the growing bandwidth and memory requirements of AI clusters.

Key players: Marvell, hyperscalers, AI cluster operators, optical component suppliers.

Why it matters: As AI clusters scale, connectivity becomes the bottleneck. The compute inside a GPU is fast, but moving data between GPUs, memory, and storage is increasingly constrained by electrical interconnects. Marvell's portfolio addresses that constraint at multiple layers: optical for reach, CXL for memory expansion, and high-speed switching for aggregation. The company is positioning itself as a critical supplier to the AI data center, competing with Broadcom and others.

Source: Marvell.

Storage & Data Infrastructure

4. Silicon Motion — Enterprise SSD Controllers for AI

PCIe Gen5 and Gen6 Controllers Target AI Inference, Agentic AI Storage Requirements

Announced November 12, 2026 · Status: announced; products available.

What happened: Silicon Motion expanded its enterprise storage business with PCIe Gen5 and Gen6 SSD controllers targeting AI inference and agentic AI storage requirements. The company is positioning itself to benefit from the changing economics of high-performance storage as AI workloads drive demand for faster, denser, and more reliable enterprise SSDs.

Key players: Silicon Motion, NAND suppliers, enterprise storage vendors, AI infrastructure operators.

Why it matters: AI inference workloads have different storage requirements from traditional enterprise applications. They need low latency, high throughput, and the ability to handle many concurrent requests. Silicon Motion's controllers are a critical component in the enterprise SSD supply chain, and its expansion into AI-specific products reflects the broader trend of storage vendors adapting to AI infrastructure demand.

Source: Silicon Motion.

Semiconductor Manufacturing

5. TSMC — CoWoS Expansion and Supply Chain Readiness

Taiwan's Semiconductor Ecosystem Prepares for Next Phase of AI Chip Manufacturing

Reported November 17–18, 2026 · Status: ongoing capacity expansion.

What happened: TSMC and Taiwan's semiconductor supply chain are preparing for the next phase of AI chip manufacturing. The company is expanding CoWoS capacity and developing suppliers to meet growing demand for advanced AI processors. Taiwan's equipment and materials suppliers are also scaling to support the expansion.

Key players: TSMC, Taiwan equipment and materials suppliers, Nvidia, AMD, Broadcom.

Why it matters: Advanced packaging remains a critical bottleneck in AI chip production. TSMC's CoWoS capacity is in high demand from Nvidia and other AI chip designers, and the company's ability to expand capacity and develop its supplier ecosystem will determine how quickly AI accelerators can be delivered. The Taiwan concentration also remains a geopolitical risk that customers and governments are monitoring.

Source: DIGITIMES.

6. Samsung — NAND Manufacturing Expansion in Vietnam

236-Layer NAND at P1 Facility as AI Infrastructure Reshapes Memory Demand

Reported November 17, 2026 · Status: reportedly planned; company has not confirmed.

What happened: Samsung is reportedly planning to build 236-layer NAND at its P1 facility in Vietnam, expanding its NAND manufacturing footprint as AI infrastructure reshapes memory demand. The move would increase the company's capacity to supply enterprise SSDs and AI storage.

Key players: Samsung, SK hynix, Micron, enterprise storage vendors.

Why it matters: The memory market is being reshaped by AI demand. Enterprise SSDs require higher layer counts and more advanced controllers to meet the performance and capacity requirements of AI workloads. Samsung's expansion is a response to that demand, and it will intensify competition with SK hynix and Micron. The reported nature of the development means it should be verified against Samsung's official announcements.

Source: DIGITIMES.

Hardware Supply Chain & Policy

7. US Chip Export Controls — Southeast Asia Under Pressure

New Restrictions on Remote Access to Advanced AI Chips Affect Singapore, Malaysia, Thailand and Indonesia

Reported November 17, 2026 · Status: interim final rule; effective date pending.

What happened: New US export restrictions on remote access to advanced AI chips are putting Southeast Asia's AI server and hardware supply chains under pressure. The rule requires licenses for reexports to Singapore, Malaysia, Thailand, Indonesia, and Brazil, affecting Nvidia chip distribution, data center customers, and server procurement in the region.

Key players: US Commerce Department, Nvidia, Southeast Asian data center operators, server manufacturers.

Why it matters: The rule extends export controls from physical hardware to cloud services, closing a loophole that allowed remote access to restricted chips. For Southeast Asia's hardware supply chain, the rule creates uncertainty about server procurement and data center operations. It also signals that the US is willing to regulate the entire AI infrastructure stack, not just the chips themselves.

Source: Associated Press.

Robotics Hardware

8. Genesis AI — Building the Full Robotics Stack

$500 Million Funding Fuels Hardware, Simulation and AI Model Integration for the Eno Robot

Reported November 18, 2026 · Status: closed; first US-made robots planned for 2027.

What happened: Genesis AI is building its own robotics hardware and intelligence stack following a $500 million funding round at a reported $3.5 billion valuation. The company is developing the Eno general-purpose robot, integrating hardware, simulation and AI models into a single system. CEO Zhou Xian said the company plans to manufacture its first fully US-made robots next year.

Key players: Genesis AI, Nvidia, Premji Invest, Khosla Ventures, Eclipse.

Why it matters: Genesis AI's integrated approach — building the hardware, the simulation environment and the AI models together — is a bet that the physical AI stack cannot be assembled from disparate components. If the company can deliver a general-purpose robot that operates reliably in unstructured environments, it would be a significant step toward commercial physical AI. The manufacturing timeline will be the first test of whether the strategy is executable.

Source: Wall Street Journal.

The CODEW Perspective

The hardware industry is being reshaped by three forces at once: the AI buildout, the memory shortage and the fragmentation of the supply chain.

The Intel-AMD share shift, Nvidia's memory warning and Marvell's connectivity expansion all describe the same reality: the hardware that powers AI is becoming more specialized and more interdependent. CPUs, GPUs, memory, storage and networking are no longer separate categories; they are layers in an integrated system, and the companies that can optimize across those layers have an advantage.

The memory shortage is the underappreciated story. Nvidia's warning about consumer PC demand is a small data point in a larger trend: the AI buildout is consuming memory capacity at a rate that is driving up prices and constraining supply. That has implications for every hardware company, from PC makers to smartphone manufacturers to server vendors. The companies that can secure memory supply will have an advantage; the ones that cannot will be constrained.

The supply chain is also fragmenting. The US chip rule is the latest example of geopolitics reshaping hardware supply chains. The rule affects not just chip designers, but server manufacturers, data center operators and the entire ecosystem that depends on access to advanced compute. The companies that can navigate the regulatory environment across multiple jurisdictions will have an edge.

The caveat worth stating plainly: this edition rests on company announcements, industry reporting and credible sources. Samsung's reported NAND expansion in Vietnam has not been confirmed by the company. Nvidia's memory warning is a company statement, not an independent analysis. The Genesis AI funding round is reported by The Wall Street Journal. This article does not provide investment advice.

The CODEW Stat

5%. The proportion of a leading competitor's model cost that Microsoft claims its MAI Code 2 model can match in performance. That is not a typo; it is the direction of travel. If the cost of frontier AI capability falls by an order of magnitude, the economics of the entire hardware stack change. The companies that sell compute will face pricing pressure; the companies that consume it will have new options. The hardware industry is not just building faster chips; it is building a cheaper path to intelligence.

Sources & References

  1. Tom's Hardware — Intel reclaims overall CPU market share (November 16, 2026)
  2. MarketWatch — Nvidia warns memory prices are weighing on consumer PC demand (November 18, 2026)
  3. Marvell — Connectivity scaling for AI infrastructure (November 13, 2026)
  4. Silicon Motion — Enterprise SSD controller expansion (November 12, 2026)
  5. DIGITIMES — TSMC and Taiwan supply chain prepare for next phase of AI chip manufacturing (November 17–18, 2026)
  6. DIGITIMES — Samsung reportedly to build 236-layer NAND at P1 facility (November 17, 2026)
  7. Associated Press — US chip export controls put Southeast Asia's AI server supply chains under pressure (November 17, 2026)
  8. Wall Street Journal — Genesis AI raises $500 million for robotics hardware (November 18, 2026)

This edition covers developments from November 12–19, 2026. Samsung's reported NAND expansion has not been confirmed by the company. Nvidia's memory warning is a company statement. Genesis AI's funding round is reported by the Wall Street Journal. The CODEW Stat references Microsoft's claim about MAI Code 2, which is company-reported and unverified. This article does not provide investment advice.

Related Coverage

About Hardware Watch

Hardware Watch tracks the companies building the physical infrastructure of the digital economy — semiconductors, servers, networking, storage, memory and emerging robotics hardware.

ABOUT THE AUTHOR

Erwin Castro

Founder, Publisher & SEO Writer at The CODEW

Erwin Castro is the founder and publisher of The CODEW, an independently operated technology and business intelligence publication covering Tech M&A, AI, enterprise software, SaaS, cloud infrastructure, startups, business operations, and digital strategy.


Hardware Watch: Nvidia's Memory Warning, Marvell's Connectivity Bet & the Robotics Hardware Race Hardware Watch: Nvidia's Memory Warning, Marvell's Connectivity Bet & the Robotics Hardware Race Reviewed by Erwin Castro on Wednesday, November 18, 2026 Rating: 5

No comments: