To be blunt, Samsung’s been on a streak of bad calls since the final years of the Lee Kun-hee era. Back then, Lee pushed 'next-gen growth engines' like office automation (printers, copiers), digicams, LEDs, and medical equipment, but most were either axed or are barely surviving after a few years. Office automation and medical gear hit a wall against 'patent cartels,' and LEDs weren't profitable because of China. They called them 'future growth engines,' but (except for medical gear) they weren't really future-proof industries. That was the biggest miscalculation. They just dipped their toes in, burned through investment capital, and then pulled out. Closing the HDD business during that exit was an even bigger mistake. I can't help but wonder: what if they had doubled down on semiconductors back then, investing those massive funds into logic chips, power semis, or building up GPU capabilities? That regret lingers.

This streak of misjudgment continued into the Lee Jae-yong era. They almost went under after getting hammered by the memory chicken game, and only recently bounced back thanks to the HBM-driven memory super-cycle. So when I saw the news yesterday, I was skeptical. Like, 'Wait, Samsung is going to develop their own GPU now? At this stage?'

As we all know, Samsung's 'sore spot' is the Foundry. Unlike the blueprint they showed years ago, they’re getting totally outclassed by TSMC. But they can’t quit. If they fold the foundry, Samsung’s entire strategy and future go up in smoke. The sunk costs are astronomical. There are many reasons why Samsung is struggling in foundry, but fundamentally, it's because they carry the risk of 'competing with their own customers.' It’s no exaggeration to say TSMC grew because Apple went all-in on them. The 'competing with customers' risk is that huge. And yet, here they are again, planning to compete with their customers by making GPUs?

Of course, if Samsung's GPU project succeeds, it creates massive opportunities. GPUs are essential in the AI era. However, relationships with existing customers will likely sour. Qualcomm, AMD, and NVIDIA will be pissed, and they'll probably move toward excluding Samsung from their ecosystems as much as possible.

So, this is just my theory, but maybe Samsung’s GPU development is just an 'insurance policy' like BADA OS was? Just like how they developed BADA and Tizen because they were worried Google might shut them out of the ecosystem to focus on Pixels, maybe they're jumping into GPU development just to maintain internal R&D capabilities for a 'what if' scenario. Even if they succeed, I suspect it'll be used limitedly for the internal ecosystem, like in Exynos. If they actually try to sell these in bulk to outsiders, that’s when the bridge with Qualcomm and NVIDIA truly burns.

Just some random thoughts. I could be totally wrong (the odds of that are high).