On June 23, the A-share AI computing PCB (Printed Circuit Board) sector took a serious hit, with concept leader Shenghong Technology seeing its stock plunge over 7% in a single day. What sparked this panic? Two viral rumors: one claiming “Nvidia demanded a 10% price cut from PCB manufacturers,” and another alleging “Shenghong Technology’s expansion delays are holding back the Rubin platform.” NUPIAO reached out to various PCB makers and market insiders to verify these claims. The consensus? These rumors are wildly exaggerated and largely misunderstood. Specifically, the idea that “Shenghong caused a delay in the Rubin (Vera Rubin AI chip)” simply doesn’t hold up against industry logic.
A chairman of a listed PCB company told NUPIAO that pricing power has actually shifted upstream to material suppliers. “In the past, the industry relied on locking in capacity with fixed quotes, often for six months or even a year. But now, with raw material costs constantly rising, pricing models have flipped to real-time quotes, and order cycles have shrunk dramatically to just 1–3 months.”
Meanwhile, official word came down hard from Nvidia itself. On the morning of June 23, Nvidia’s Investor Relations department issued a formal clarification: the claim about a forced 10% price squeeze is completely false. There is no blanket policy cutting prices across the board; annual framework orders are still being executed at the original agreed-upon commercial rates.
As for the “Rubin delay” rumor, let’s set the record straight. Nvidia originally planned to start mass production of the Rubin series in the second half of 2026. According to a major AI Server industry survey released by TrendForce in April, the shipment mix for Nvidia’s high-end AI chips in 2026 is expected to change. Due to shifting international dynamics and the time needed for supply chain calibration, the share of Hopper and Rubin series in total high-end GPU shipments is projected to drop, pushing the Blackwell series’ share up significantly from 61% to an estimated 71%.
In that same TrendForce report, they estimate the Rubin series’ shipment share will dip from a planned 29% down to 22%. Why the slowdown? It boils down to four key technical hurdles: first, the certification process for core components like HBM4 took longer than expected; second, upgrading network transmission from CX8 to CX9 required significant adaptation work; third, the sharp rise in per-chip power consumption challenged whole rack power management solutions; and fourth, optimizing the overall efficiency of higher-spec liquid cooling systems proved complex.
So, the rumor that “Shenghong Technology’s expansion dragged down Rubin platform shipments” is clearly baseless and lacks any industrial backing.
Before all this noise, a report from overseas firm SemiAnalysis suggested that two of Nvidia’s core technologies—CPO (Co-packaged Optics) and 800VDC—might face timeline slippages. Regarding CPO mass production, market predictions had previously anticipated annual shipments of 60,000 to 100,000+ units by 2027. However, current产能 falls far short of that target. SemiAnalysis was blunt: the industry’s optimism about a full-scale CPO rollout in 2027 was way too optimistic. Large-scale commercial adoption is likely being pushed back to 2028–2029.
But don’t let the pessimistic institutional views fool you. Gilad Shainer, Nvidia’s Senior Vice President of Networking, publicly addressed these concerns. He confirmed that their CPO switches will proceed exactly as scheduled, kicking off mass production and customer delivery in the second half of 2026. He emphasized that there’s a difference between small-batch commercial use and full-scale adoption. The plan is to introduce products to top-tier supercomputing clients in late 2026, gradually scaling up capacity in 2027. Crucially, CPO technology and traditional optical modules will coexist for the long haul; the technology route is definitely not dying.
Let’s talk about Shenghong Technology. As the global leader in AI computing PCBs, it’s one of Nvidia’s core suppliers. The company traces its roots back to 2003 when Chen Tao founded Shenghua Electronics, officially becoming Shenghong Technology (Huizhou) Co., Ltd. in July 2006. They specialize in R&D, manufacturing, and sales of high-density PCBs. Their core products include high-order HDI boards and multi-layer PCBs, widely used in AI, data centers, smart cars, and 5G infrastructure. In the AI boom, Shenghong has exploded thanks to its high-end PCB tech barriers. While standard consumer electronics PCBs usually have just 4–6 layers, AI server-grade high-end PCBs can exceed 20 layers with micron-level precision—a difficulty level comparable to chip manufacturing.
Today, Shenghong is deeply tied to global tech giants like Nvidia, AMD, Microsoft, Google, and Tesla. It serves as a Tier-1 core supplier for Nvidia’s AI server PCBs. Data from Frost & Sullivan reveals the sheer scale of their dominance: based on revenue from AI and high-performance computing PCBs in the first half of 2025, Shenghong holds a 13.8% market share, ranking #1 globally. Just compare that to the same period in 2024, where their share was only 1.7%, placing them seventh.
Their critical role in Nvidia’s supply chain was made obvious during January’s famous “Trillion Yuan Banquet.” This annual supply chain dinner hosted by Nvidia CEO Jensen Huang in Taipei got its nickname because the combined market cap of attending companies exceeded 1 trillion New Taiwan Dollars. At this elite gathering of Taiwanese tech titans like TSMC, Foxconn, and Quanta, Chen Tao, founder of Shenghong Technology, was the *only* mainland Chinese executive invited—and he attended for two years in a row.
Looking at the financials, the results speak for themselves. For the full year of 2025, Shenghong reported operating revenue of 19.292 billion yuan, a massive 79.77% year-over-year increase. Net profit attributable to shareholders soared 273.52% to 4.312 billion yuan. Their gross margin hit 35.22%, a jump of 12.5 percentage points, perfectly reflecting the high value added by their premium AI PCB products.
Even in Q1 2026, the momentum hasn’t slowed. Revenue reached 5.519 billion yuan (up 27.99% YoY), net profit climbed 39.95% to 1.288 billion yuan, and net cash flow from operating activities surged an incredible 399.38% to 2.117 billion yuan, signaling a dramatic improvement in payment collection quality.
Last May, Shenghong Technology stated clearly to the public: their order book is packed, business progress is smooth, and all order production and deliveries are proceeding normally.
