The Show's Defining Moment: Jensen Huang and Marvell

The single most impactful event was not on the official keynote calendar. During Marvell CEO Matt Murphy’s presentation, Nvidia CEO Jensen Huang walked on stage and made a pronouncement that moved markets instantly .

Huang told the audience that Marvell is “the next trillion-dollar company” . The market reacted violently: Marvell’s stock surged 32.5% in a single trading day, adding roughly $40 billion in market capitalization . The company’s market cap had been under $200 billion before the statement .

Huang’s confidence rested on a deepening strategic partnership. Marvell’s custom ASICs and optical connectivity silicon are essential data-center infrastructure for AI workloads. Nvidia had already placed a $2 billion strategic investment in Marvell months earlier . Marvell’s AI data-center revenue now accounts for roughly 76% of its business .

Intel's Keynote: Agentic AI and the 18A Process

Intel CEO Lip-Bu Tan used the COMPUTEX stage to reposition the chipmaker as a systems company for the agentic AI era . Tan argued that the rise of agentic AI—autonomous software agents performing complex, multi-step tasks—will drive a CPU renaissance, not just GPU demand .

To back that thesis, Intel formally launched the Xeon 6+ processor family, codenamed “Clearwater Forest” . It is Intel’s first data-center CPU built on its Intel 18A process node .

Key specifications include :

• Up to 288 Darkmont efficiency cores per socket
• 576 MB of L3 cache
• 12-channel DDR5-8000 memory support
• 96 PCIe Gen 5 lanes and 64 CXL 2.0 lanes
• Thermal Design Power (TDP) ranging from 300W to 450W

Tan confirmed that Intel 18A manufacturing is “now at full scale” . The node is so supply-constrained that Intel teams are making daily allocation decisions between customers . A single liquid-cooled rack of Xeon 6+ processors can pack 36,864 cores in 32U of rack space at roughly 100 kW, potentially hosting up to 150,000 AI agents per rack .

Intel also announced rack-scale design partnerships with SambaNova and Foxconn, signaling a push toward pre-integrated AI infrastructure .

The Full Keynote Lineup

Beyond Intel and Marvell, the show’s executive keynote roster underscored the event’s AI focus:

• Qualcomm CEO Cristiano Amon delivered a keynote centered on on-device AI and connectivity.
• NXP CEO Kurt Sievers focused on edge intelligence and secure processing for industrial IoT .

Edge AI Moves from Concept to Reality

The exhibition floor made one trend clearer than any keynote slide: AI is escaping the data center and landing in factories, vehicles, and industrial equipment.

Innodisk demonstrated its five-layer edge AI ecosystem at the show. The framework integrates compute, memory, storage, sensing/communication, and software layers into a single production-ready stack . Innodisk’s on-premises AI platform, the AccelBrain running on APEX-X200, promises 100% data sovereignty by running LLMs entirely locally . For industrial environments, the company demonstrated heavy-machinery safety vision systems powered by Intel, Qualcomm, and NVIDIA Blackwell architectures .

MICROIP debuted its new “AI Vehicle System Business Group,” announcing dual AI software platforms called AIVO and XEdgAI . The automotive focus includes an adaptive system with 360-degree multi-channel AI awareness, front-vehicle and traffic-light recognition, and rear blind-spot monitoring . The company described its approach as a “triple safety protection” concept .

Exhibition Zones and Four Strategic Themes

The show organized itself around four high-level themes to help attendees navigate the massive floor plan:

• AI Computing – data-center infrastructure, server hardware, and chip design
• Robotics & Intelligent Mobility – autonomous systems, smart vehicles, and humanoid robotics
• Next-Generation Technology – advanced communications, immersive tech, and quantum-adjacent R&D
• Sustainable Innovation – green energy solutions and energy-efficient computing

Three new dedicated zones appeared for the first time: an AI Robotics Zone, a TechXperience Hall, and an E-Paper Pavilion .

The Taiwan AI Supply Chain's Winning Streak

Taiwan’s server ODMs entered the show with strong momentum. Leading manufacturers, including Foxconn, Quanta (through QCT), Wiwynn, Inventec, Gigabyte, Asus, and ASRock, reported record Q1 financial results .

Hyperscaler orders for Nvidia’s next-generation GB300 rack systems were accelerating into the second half of 2026 . The industry is also tracking a new category called “Physical AI” — intelligent systems that interact with the physical world — as a fresh growth lane beyond traditional training and inference workloads .

By the time the show concluded on June 5, organizers declared COMPUTEX 2026 the largest and most AI-focused edition in its history . The record turnout of over 110,000 buyers and the expanded international footprint positioned Taiwan as the unchallenged hub of the global AI supply chain .

The Show's Defining Moment: Jensen Huang and Marvell

The single most impactful event was not on the official keynote calendar. During Marvell CEO Matt Murphy’s presentation, Nvidia CEO Jensen Huang walked on stage and made a pronouncement that moved markets instantly .

Huang told the audience that Marvell is “the next trillion-dollar company” . The market reacted violently: Marvell’s stock surged 32.5% in a single trading day, adding roughly $40 billion in market capitalization . The company’s market cap had been under $200 billion before the statement .

Huang’s confidence rested on a deepening strategic partnership. Marvell’s custom ASICs and optical connectivity silicon are essential data-center infrastructure for AI workloads. Nvidia had already placed a $2 billion strategic investment in Marvell months earlier . Marvell’s AI data-center revenue now accounts for roughly 76% of its business .

Intel's Keynote: Agentic AI and the 18A Process

Intel CEO Lip-Bu Tan used the COMPUTEX stage to reposition the chipmaker as a systems company for the agentic AI era . Tan argued that the rise of agentic AI—autonomous software agents performing complex, multi-step tasks—will drive a CPU renaissance, not just GPU demand .

To back that thesis, Intel formally launched the Xeon 6+ processor family, codenamed “Clearwater Forest” . It is Intel’s first data-center CPU built on its Intel 18A process node .

Key specifications include :

• Up to 288 Darkmont efficiency cores per socket
• 576 MB of L3 cache
• 12-channel DDR5-8000 memory support
• 96 PCIe Gen 5 lanes and 64 CXL 2.0 lanes
• Thermal Design Power (TDP) ranging from 300W to 450W

Tan confirmed that Intel 18A manufacturing is “now at full scale” . The node is so supply-constrained that Intel teams are making daily allocation decisions between customers . A single liquid-cooled rack of Xeon 6+ processors can pack 36,864 cores in 32U of rack space at roughly 100 kW, potentially hosting up to 150,000 AI agents per rack .

Intel also announced rack-scale design partnerships with SambaNova and Foxconn, signaling a push toward pre-integrated AI infrastructure .

The Full Keynote Lineup

Beyond Intel and Marvell, the show’s executive keynote roster underscored the event’s AI focus:

• Qualcomm CEO Cristiano Amon delivered a keynote centered on on-device AI and connectivity.
• NXP CEO Kurt Sievers focused on edge intelligence and secure processing for industrial IoT .

Edge AI Moves from Concept to Reality

The exhibition floor made one trend clearer than any keynote slide: AI is escaping the data center and landing in factories, vehicles, and industrial equipment.

Innodisk demonstrated its five-layer edge AI ecosystem at the show. The framework integrates compute, memory, storage, sensing/communication, and software layers into a single production-ready stack . Innodisk’s on-premises AI platform, the AccelBrain running on APEX-X200, promises 100% data sovereignty by running LLMs entirely locally . For industrial environments, the company demonstrated heavy-machinery safety vision systems powered by Intel, Qualcomm, and NVIDIA Blackwell architectures .

MICROIP debuted its new “AI Vehicle System Business Group,” announcing dual AI software platforms called AIVO and XEdgAI . The automotive focus includes an adaptive system with 360-degree multi-channel AI awareness, front-vehicle and traffic-light recognition, and rear blind-spot monitoring . The company described its approach as a “triple safety protection” concept .

Exhibition Zones and Four Strategic Themes

The show organized itself around four high-level themes to help attendees navigate the massive floor plan:

• AI Computing – data-center infrastructure, server hardware, and chip design
• Robotics & Intelligent Mobility – autonomous systems, smart vehicles, and humanoid robotics
• Next-Generation Technology – advanced communications, immersive tech, and quantum-adjacent R&D
• Sustainable Innovation – green energy solutions and energy-efficient computing

Three new dedicated zones appeared for the first time: an AI Robotics Zone, a TechXperience Hall, and an E-Paper Pavilion .

The Taiwan AI Supply Chain's Winning Streak

Taiwan’s server ODMs entered the show with strong momentum. Leading manufacturers, including Foxconn, Quanta (through QCT), Wiwynn, Inventec, Gigabyte, Asus, and ASRock, reported record Q1 financial results .

Hyperscaler orders for Nvidia’s next-generation GB300 rack systems were accelerating into the second half of 2026 . The industry is also tracking a new category called “Physical AI” — intelligent systems that interact with the physical world — as a fresh growth lane beyond traditional training and inference workloads .

By the time the show concluded on June 5, organizers declared COMPUTEX 2026 the largest and most AI-focused edition in its history . The record turnout of over 110,000 buyers and the expanded international footprint positioned Taiwan as the unchallenged hub of the global AI supply chain .