The data center "power engine" is experiencing explosive growth! The AI wave is spilling over to the power chain. ON Semiconductor Corporation (ON.US) quadrupled its cash flow in Q2, and its performance outlook exceeded expectations.

ON Semiconductor Corporation (ON.US), a chip manufacturing giant focused on the automotive and industrial sectors, reported its performance and future outlook after the US market close on Monday (morning Beijing time on Tuesday). The company's revenue outlook for the third quarter exceeded Wall Street analysts' consensus expectations, underscoring the explosive growth of its AI data center power engine businessspecifically, the surge in demand for power management chips used in artificial intelligence data centers. This led to a more than 7% increase in the company's stock price during after-hours trading. Amidst the AI infrastructure frenzy, the company's second-quarter performance ending July 3 was also robust, particularly in its power solutions business unit, PSG, which encompasses automotive, industrial, and AI data center operations, showcasing the strongest growth. So far this year, ON Semiconductor Corporation's stock price has surged significantly, driven by strong recovery expectations for demand for analog chips and power-related chips in data centers aligned with the unprecedented wave of AI data center construction. Although the company's stock price has adjusted since July due to global deleveraging related to AI computing power and extreme crowded positions, it has still risen by 50% since 2026; by contrast, the chip giant saw a total stock price decline of 15% in 2025. The chip demand frenzy brought about by AI is spilling over from computational chips themselves (GPU/ASIC/HBM) to the broader categories of data center power and signal chains, with the intensity of this spillover accelerating significantly, further validating the logic that the so-called "end of AI is power" reflects the bull market curve for the data center power chain. Under the AI wave, the so-called endless chip demand driven by AI training/inference is smoothly transmitted from AI chips and memory chips to analog chips and power semiconductors, thereby robustly driving leaders in analog chips like Texas Instruments Incorporated, Infineon, and ON Semiconductor Corporation towards a strong recovery trajectory in performance. The market interprets these strong performance data as the analog chain starting to reap super dividends from AI infrastructure. With a $7 billion acquisition coupled with AI data centers igniting demand for power chips, ON Semiconductor is entering a new growth cycle. As ON Semiconductor Corporation announces robust results and future outlook, it is also striving to seize the growing demand in artificial intelligence devices and the Siasun Robot & Automation sector through its largest acquisition to datethe all-stock acquisition of Synaptics, a subsidiary of EDA chip design software giant Synopsys, Inc., announced in June, valued at $7 billion. ON Semiconductor Corporation's management expects third-quarter revenue to be between $1.65 billion and $1.75 billion. According to data compiled by LSEG, the midpoint of this range is above Wall Street analysts' average expectation of $1.67 billion. CEO Hassane El-Khoury stated in the earnings announcement, The AI data center-related business remains our fastest-growing segment, and we currently expect that revenue for this segment will at least double by 2026, reflecting the strong strength of our intelligent power product portfolio and the expanding adoption of ON Semiconductor throughout the power tree architecture by our customers. For the second quarter ending July 3, the company reported revenue of $1.6 billion, a year-on-year increase of 9.2%, slightly exceeding Wall Street analysts' consensus expectation of around $1.59 billion. The companys adjusted earnings per share (EPS) was $0.74, a year-on-year increase of approximately 40%, surpassing market expectations of $0.71. The management expects the third quarter adjusted EPS to be in the range of $0.81 to $0.93, with the midpoint significantly higher than the Wall Street analysts' consensus expectation of approximately $0.83. ON Semiconductor is also continuing to advance its Fab Right strategy to significantly reduce costs and significantly improve operational efficiency. In July, as part of this strategy, the company sold two manufacturing plants. In other aspects of the second-quarter performance data, ON Semiconductor's GAAP gross margin increased from 37.6% to 38.4%, while the non-GAAP gross margin rose to 39.3%; GAAP operating profit margin increased from 13.2% to 16.1%, and non-GAAP operating profit margin improved from 17.3% to 20.8%. Net profit attributable to shareholders increased from $170.3 million to $226.8 million, marking a year-on-year increase of a substantial 33.2%. GAAP EPS rose from $0.41 to $0.56, up 36.6%. The company's operating cash flow data as of July 3 increased by 149.4% to $459.7 million, and free cash flow surged from $106.1 million to $425.4 million, roughly quadrupling the amount from the same period last year, with the free cash flow rate rising from around 7% to 27%. During the quarter, the company repurchased $333 million in stock. In other words, although revenue grew only about 9%, profits, cash flow data, and shareholder returns expanded at several times the rate, indicating that ON Semiconductor Corporation's Fab Right capacity optimization, cost discipline, and product portfolio upgrades have begun to create real operational leverage. Segment data further demonstrates that growth is increasingly concentrated in the data center power semiconductor business. Revenue in the power solutions segment (PSG) reached $829 million, a year-on-year increase of 18.7% and a quarter-on-quarter increase of 13%, while its proportion of total revenue rose from 47.5% in the same period last year to 51.7%, becoming the source of more than half of the company's income. Revenue from the analog and mixed-signal business (AMG) was approximately $545.7 million, representing a year-on-year decline of 1.8%. The intelligent sensing group (ISG) reported revenue of $228.8 million, up 6.6% year-on-year. In the first half of the year, ON Semiconductor's total revenue was $3.1168 billion, a year-on-year increase of 6.9%. The adjusted EPS rose from $1.08 to $1.38, marking a year-on-year increase of 27.8%. Free cash flow was $642.6 million, an increase of 14.6% year-on-year. The semiannual performance and segment performance data collectively indicate that ON Semiconductor has not seen synchronized prosperity across all its businesses, but rather that AI data centers, high-voltage power supply, and certain electrification demands in the automotive sector are driving PSG to accelerate ahead, while AMG remains in a relatively mild recovery phase. The "chip demand frenzy" triggered by the AI infrastructure wave is spreading from AI chips and memory chips to analog chips and power semiconductors. ON Semiconductor's strong performance indicates that AI chip demand is systematically spilling over from GPU, ASIC, and HBM itself to power semiconductors and analog control and monitoring signal chains. This is not merely the simple exploitation of the AI concept by analog chip companies, but rather a physical necessity brought about by the increase in computational density. ON Semiconductor has defined its AI data center operations as the fastest-growing segment and expects related revenue to more than double by 2026; Texas Instruments Incorporated's data center business revenue grew by approximately 90% year-on-year in the first quarter of 2026, while growth in the second quarter continued to be led jointly by industrial, data center, and automotive sectors; Infineon has also joined the NVIDIA Corporation MGX ecosystem, providing a complete solution from high-voltage conversion to GPU core power supply around the 800V DC power supply, significantly boosting performance figures in the first half of the year. The benefits of AI capital expenditure are thus extending from a few advanced process chips to numerous lower-priced, yet high-volume, long-lifetime devices that must undergo reliability certification in the data center power chain foundational analog devices. In the realms of analog/power chips and power semiconductors/discrete devices, the core logic behind the strong demand expansion fueled by the unprecedented AI wave lies in the fact that AI racks are surging from the tens of kilowatts typical in traditional servers to over 100 kilowatts, and even evolving toward 600 kilowatts and 1 megawatt. This means the power consumption for next-generation GPUs/TPUs/ASICs could reach 2 to 4 kilowatts per unit, while traditional 48V/54V architectures will generate thousands of amperes of current, resulting in copper losses, heat generation, and busbar volume increasing sharply. This is precisely why global AI data centers are being compelled to actively shift towards 400V or 800V DC architectures promoted by ON Semiconductor and NVIDIA Corporation. Before power enters the GPU from the electrical grid, it must undergo AC/DC rectification, PSU, BBU, high-voltage distribution, 800V to 50V intermediate bus conversion, and then be reduced to below 1V required by the GPU core through multi-phase VRM. Each stage requires power switches (Si, SiC, or GaN), gate drivers, digital controllers, power management ICs, and electronic fuses. What ON Semiconductor, Texas Instruments Incorporated, and Infineon provide are precisely these "devices that every watt must go through," thus the higher the rack power and the more complex the conversion stages, the greater the semiconductor value usually associated with each cabinet. ON Semiconductor can provide Elite SiC MOSFET/JFETs, high and low voltage silicon MOSFETs, GaN devices, hot-swappable intelligent fuses, multi-phase controllers, power stages, and PoL voltage regulators within this power tree from "the power grid to the GPU." This means that a single AI rack not only requires more chips but also demands higher-priced, higher-voltage, and more efficient power devices. The NVIDIA Corporation NVIDIA MGX platform can be seen as the core strategic entry point for ON Semiconductor's AI data center business, as ON Semiconductor has supplied power FETs, multi-phase power supplies, SiC JFETs, and GaN solutions to the existing MGX systems and is directly servicing NVIDIA Corporation and future 800V distribution board suppliers within the MGX ecosystem; the standardization of MGX server and rack designs also facilitates easier adoption of certified power devices by multiple OEMs and ODMs.