The fresh hotspot is not another argument about whether Nvidia can keep carrying the tape. It is the next layer down the stack. On June 2, Reuters reported that Hewlett Packard Enterprise surged nearly 29% in premarket trading after pulling forward its long-term financial targets by two years, helped by strong AI infrastructure demand. That matters because traders are reading it as confirmation that the AI buildout is not stalling at the model or chip level. It is moving into the physical systems that let large AI workloads actually run.
HPE’s own June 1 results back up that read. The company said fiscal second-quarter revenue reached $10.7 billion, Cloud & AI revenue rose to $7.7 billion, server revenue climbed 32.7%, and data-center networking revenue jumped 233.3%. Reuters added that management linked the shift to growing enterprise adoption of agentic AI workloads. In plain trading terms, that means the market is starting to price AI as a sustained enterprise capex cycle rather than a narrow chip shortage story.
Once that happens, the watchlist expands quickly. In Japan, Fujikura has become one of the more important read-through names because AI factories need far more optical connectivity than traditional enterprise IT. Filing-based coverage on May 14 showed Fujikura’s operating profit jumped 39.2% on data-center demand, while a March 13 Reuters brief said the company plans to invest as much as 300 billion yen to lift optical-fiber and cable capacity in Japan and the United States to as much as roughly three times current levels. That is not a defensive move. It is management behaving as if AI-related network demand has become structural.
South Korea gives the same theme a different expression. LS Electric’s first-quarter 2026 investor materials explicitly highlighted data-center market expansion, strong performance tied to AIDC, semiconductors and renewables, and a push to target U.S. big tech companies with switchgear, transformers, and distribution equipment. Reuters, citing Yonhap on May 20, also reported that LS Electric won a $64 million order to supply high-voltage switchgear to a U.S. data center. Traders should notice what that says: the bottleneck is no longer just compute. Time-to-power is becoming its own trade.
Europe fits through the energy-management and cooling angle. Schneider Electric said in its April 30 first-quarter revenue report that data centers remained its main growth driver, with strong sales across cooling technologies, prefabricated solutions, and lifecycle services as customers scaled AI-ready infrastructure across geographies. That is an important cross-market signal because Europe may not dominate the GPU narrative, but it is deeply relevant in the electrification, thermal management, and automation layer that hyperscalers need when AI projects move from pilot to industrial scale.
My cautious view is that this is a healthier signal than another pure semiconductor melt-up, but it is also a more selective one. The cleanest trade is not “AI up, buy everything.” It is to watch the companies solving physical constraints: servers that enterprises can deploy now, optical fiber that increases rack-to-rack bandwidth, and power systems that get large clusters energized without waiting years for the grid to catch up. If that thesis stays intact, the winners can keep broadening beyond the usual megacap names. If power delays, permitting friction, or over-ordering show up, these infrastructure names can derate just as violently as the chip leaders did during prior pauses.
Risk notice: This article is for market commentary and education only. It is not investment advice. AI infrastructure trades are highly sentiment-driven and can reverse quickly if capex plans, power availability, supply chains, or enterprise demand weaken.
Sources:
Reuters via Investing.com on HPE’s June 2 surge
HPE fiscal 2026 Q2 results, June 1, 2026
FilingReader summary of Fujikura’s May 14 filings
Reuters brief on Fujikura’s optical-fiber capacity expansion, March 13, 2026
LS Electric 1Q 2026 earnings release
Reuters/Yonhap headline on LS Electric’s U.S. data-center switchgear order, May 20, 2026
Schneider Electric Q1 2026 revenues, April 30, 2026
Schneider Electric AI factory energy automation article, April 7, 2026
原创文章，作者：financial transaction，如若转载，请注明出处：https://www.fanbi.net/archives/118
