ON Semiconductor Profits Surge as AI Data Centers Drive Demand for Power Silicon
The WSJ framing attributes the revenue uplift to that rack-level proliferation, and even though the available snippet doesn't break out exact figures, the directional read is unmistakable: the analog…

The Wall Street Journal is reporting that ON Semiconductor posted higher profit and revenue, with the AI data center business doing the heavy lifting. If you've been watching the silicon supply chain for the last 18 months, this isn't a shock — it's the load-bearing wall of the entire analog and power semis segment finally announcing itself out loud.
What's actually inside the silicon
ON Semi isn't selling you training accelerators. Their bread and butter is the boring, beautiful stuff — power delivery, isolation, sensing — the unglamorous silicon that quietly makes sure a rack full of GPUs doesn't melt, brown out, or trip a breaker. When AI data centers multiply, every additional megawatt of compute needs more of these support devices around it. The WSJ framing attributes the revenue uplift to that rack-level proliferation, and even though the available snippet doesn't break out exact figures, the directional read is unmistakable: the analog side of the industry is now feeding the same AI capex cycle that has the obvious names printing records.
Why your next device cares
Here's where a consumer-tech reader should pay attention. When the same fabs and product lines that historically fed automotive, industrial, and prosumer gear start tilting toward hyperscaler customers, allocation bends. You — yes, you, hunting for a high-wattage charger or refreshing a router that may or may not still be in stock — feel it in lead times and pricing first, then in the parts bin of whatever "consumer" device lands next. ON Semi making more money from AI data centers isn't just an earnings beat; it's a signal that the analog silicon stack is reorienting around training clusters and inference racks, and that reorientation trickles down.
Meanwhile, the Photonics Spectra week-in-brief from August 7 carries a useful counterpoint: Kubos Semiconductors closed a $2.1M round. Small potatoes against hyperscaler balance sheets, but a reminder that the AI supply chain has a long tail of niche photonics and compound-semiconductor players still raising early money. That long tail is where the next wave of supply diversification will come from, and it's worth keeping an eye on.
What to watch from your desk
If you care about consumer hardware, don't just stare at the obvious accelerator vendors' margins. Track ON Semi's mix shift toward data center revenue, watch for any language about allocation or lead times for legacy automotive and industrial parts, and keep tabs on smaller compound-semiconductor outfits raising fresh capital — they decide whether the consumer side of the aisle eventually catches a break on supply. The AI buildout is great for chipmaker earnings. For your next device's bill of materials? Probably less great, at least until that long tail of smaller suppliers scales up.
And speaking of data analytics reshaping industries nobody saw coming — Luminate's report on non-English music grabbing a massive share of US streams is yet another reminder that the same hyperscale-data backbone now surfaces insights no one could see a decade ago. Different industry, same engine.