ON looks misread here. The market is still anchoring on the ugly cyclical numbers — revenue down 15.3% year over year and EPS down 92.1% — while underweighting a much more important shift: onsemi is building a credible AI power franchise that is already showing up in management's commentary and new product launches. The June 9 GaNEXUS launch for AI data centers, paired with Q1 commentary that AI data-center revenue more than doubled year over year and grew more than 30% sequentially, gives this stock a real re-rating path. We think the tape is treating ON like a trough-sensitive auto semiconductor when the next leg of the story is increasingly about power density, efficiency, and AI infrastructure.
The cleanest reason to lean against the recent weakness is that the AI story is no longer just a slide-deck aspiration. Management said in the May 4 quarter that AI data-center revenue more than doubled from a year ago and grew more than 30% sequentially in Q1, then followed that up on June 9 with GaNEXUS, a gallium nitride portfolio aimed directly at AI data centers across 40V to 650V. That sequence matters. It shows both demand traction and product intent, which is exactly how a cyclical chipmaker starts earning a different multiple.
The balance sheet gives ON room to bridge the cycle while that mix shift develops. Financial Health is a perfect 100 inside the TickerSpark Score, and the company repurchased $346 million of stock in Q1. That matters more than the headline P/E of 83.02 suggests, because the earnings base is still depressed by the downturn. When a company says it has moved beyond the cyclical trough, keeps buying back stock, and is launching data-center power products into a market obsessed with efficiency, we pay more attention to the direction of the model than the backward-looking multiple.
The market is also not dealing with a broken chart or a dead name. ON is up 104.5% year to date, crushing the Technology sector's 27.6% gain, and it still sits above both its 50-day and 200-day moving averages. Momentum is 100 in the TickerSpark Score, and news sentiment is strongly positive with a 7-day reading of 0.9171. That combination tells us the 2026 move has been driven by more than hype alone. The next obvious catalyst is September 16 Analyst Day, where management has a chance to put harder edges around the data-center revenue opportunity, silicon carbide, vertical GaN, and margin expansion.
The skepticism is not hard to understand. ON still carries cyclical scars, and the current fundamentals are messy: operating margin is 10.8%, net margin is 9.5%, and year-over-year growth is sharply negative. The AI business also has not been broken out in dollar terms, and GaNEXUS is sampling now rather than contributing at scale today. That leaves room for investors to say this is still mostly an auto and industrial chip company wearing an AI narrative.
That pushback is fair, but it misses where reratings begin. They do not start after every revenue stream is mature and fully disclosed; they start when a company shows credible evidence that the next mix is better than the last one. ON has already done that with sequential AI data-center growth above 30%, a fresh AI-focused power launch, and a management team explicitly saying the trough is behind it. Against peers like STM and GFS, ON's current valuation does not look cheap on trailing earnings, but the market is paying for transition, not perfection.
We're buyers of the disconnect, not the rearview mirror. The setup into September 16 is attractive because the stock has already proven it can outperform when investors start treating ON as a power-enabler of AI infrastructure rather than just another cyclical auto semiconductor. If management uses Analyst Day to quantify the data-center opportunity and show GaN and SiC translating into a more margin-accretive mix, this narrative gets harder to dismiss.
What would change our mind is simple: if the company gets to September without turning the AI commentary into clearer commercial proof, or if the next quarter shows the core business deteriorating faster than AI can offset it, the cyclical label will stick. Until then, we think the market is selling ON for the wrong reason. This is a name we'd own, with respect for volatility, ahead of the catalyst that could finally force a reframe.
