NVIDIA Corp. CEO Jensen Huang landed Saturday in Taiwan ahead of the Computex 2026 trade show, smack dab in the middle of a high-stakes geopolitical storm involving record corporate earnings, tight supply chains, and a major international hardware smuggling scandal.
Upon landing at Taipei’s Songshan Airport, Huang immediately called on long-time server manufacturing partner Super Micro Computer Inc. to aggressively tighten its export compliance controls. The remarks follow a string of major legal crackdowns in both the U.S. and Taiwan targeting the illicit diversion of NVIDIA-powered hardware to mainland China.
Huang’s rare public admonition of a major partner comes just days after Taiwanese prosecutors launched the island’s first formal crackdown on illicit AI hardware exports. On Thursday, a Taiwanese court granted a request to detain three individuals accused of submitting fraudulent shipping declarations to smuggle Super Micro servers equipped with restricted NVIDIA artificial intelligence (AI) chips to China, Hong Kong, and Macau.
The Taiwan case mirrors a massive, multibillion-dollar U.S. federal prosecution unsealed in March. In that case, federal prosecutors charged Supermicro co-founder and former Senior Vice President Yih-Shyan “Wally” Liaw, alongside two others, with conspiring to smuggle roughly $2.5 billion worth of NVIDIA-equipped servers to China via Southeast Asian shell companies. Liaw has pleaded not guilty. Super Micro, which is not named as a defendant, stated it is cooperating fully with investigators and has placed the implicated employees on administrative leave.
Responding to the widening scandal, Huang told reporters that NVIDIA remains “rigorous” in enforcing compliance.
“Ultimately Super Micro has to run their own company,” Huang said. “We insist our partners are compliant. We hope that they will enhance and improve their regulation compliance and prevent that from happening in the future.”
Super Micro responded in an email statement, promising to further strengthen its global trade compliance program via pre- and post-shipment verifications, adding that recent events “underscore the need for industry-wide solutions to safeguard supply chains.”
The compliance controversy stands in stark contrast to NVIDIA’s explosive financial performance. Just days prior to Huang’s arrival, NVIDIA posted a record Q1 fiscal 2027 revenue of $81.62 billion, an 85% year-over-year surge driven by an astronomical $75.2 billion in data center revenue.
Astonishingly, NVIDIA guided its Q2 fiscal 2027 revenue even higher to $91 billion, factoring in zero data center compute revenue from China. U.S. export controls enacted in 2022 have effectively collapsed NVIDIA’s Chinese market share from 95% to near zero, opening a vacuum that domestic rival Huawei has rapidly filled with its Ascend 950PR chip.
While a December 2025 policy framework theoretically created a licensed pathway for NVIDIA to sell its H200 chips to 10 approved Chinese tech giants (including Alibaba and ByteDance) under a 25% U.S. government revenue share, not a single H200 has been delivered. Beijing has reportedly instructed domestic firms to prioritize Huawei alternatives, leaving the deal entirely frozen.
Despite the current stagnation, Huang confirmed to reporters that China remains included in his projected $200 billion total addressable market for NVIDIA’s upcoming Vera CPU line.
“The Chinese market is very important. It’s very large, of course,” Huang said, deeming the complete concession of the market strategically problematic.
With the Chinese market effectively blocked, Huang’s trip to Taipei is focused on ensuring the global supply chain can keep pace with rampant Western demand. The CEO is scheduled to meet with TSMC Chairman and CEO C.C. Wei to lock down critical production commitments for NVIDIA’s next-generation Vera Rubin platform.
Billed by Huang as “probably the largest product launch in the history of Taiwan,” the hyper-complex, six-chip architecture represents NVIDIA’s first foray into the global CPU market. Designed to power agentic AI, a single Vera Rubin NVL72 configuration connects 36 Vera CPUs and 72 Rubin GPUs, utilizes nearly 2 million parts, and requires the coordination of roughly 150 Taiwanese ecosystem partners.
NVIDIA claims the new platform delivers 3.5 times the training performance and five times the inference performance of its current Blackwell predecessor, while cutting inference costs to a fraction of previous generation expenses.
However, manufacturing the platform relies entirely on TSMC’s highly coveted Chip-on-Wafer-on-Substrate (CoWoS) advanced packaging technology. While TSMC is executing a historic capacity expansion, up to 140,000 wafers per month by the end of 2026, capacity remains entirely sold out. NVIDIA has reportedly pre-committed to more than half of TSMC’s CoWoS allocation through 2027, severely tightening the supply bottleneck for chief competitors like AMD Inc.
As Huang prepares for his Computex keynote, he is not the only tech titan on the ground in Taipei. AMD CEO Lisa Su has also arrived in Taiwan, signaling an aggressive push to challenge NVIDIA’s AI dominance.
AMD has successfully achieved mass production of the industry’s first 2nm High-Performance Computing (HPC) product at TSMC, codenamed EPYC Venice. As the AI landscape shifts toward agentic workflows that require massive CPU orchestration alongside GPU compute, AMD is positioning its upcoming Computex announcements to directly counter both Intel Corp. and NVIDIA.
SHARE THIS STORY
Originally published by Techstrong.IT. Republished with attribution.
