The tech world is always moving at lightning speed. Lately, the biggest race is all about chips for artificial intelligence, or AI. Companies are scrambling to build the best brains for these smart systems. For decades, Intel was the king of computer chips. But now, they’re facing a tough truth. Even one of their former top leaders admits they made a big mistake.
Pat Gelsinger, who used to run Intel, believed in a different path for AI. He thought the most important part of AI was “inference.” This is when an already-trained AI model is used in real-time, like when your phone recognizes your voice. So, Intel put its efforts there. Meanwhile, rivals like NVIDIA were pouring money into “training” AI models. Training is where you teach the AI from scratch using huge amounts of data. This area quickly became the real powerhouse of the new chip economy. This difference in ideas cost Intel a lot. While others dominated data centers with their training chips, Intel was left with more general products. They just couldn’t deliver the kind of speed AI needed on a large scale. The news of Gelsinger’s admission was reported recently by WCCFTech.
Slow Start, Hard Stops
Intel does have some AI chips out there, like their Gaudi accelerators. But not many companies are using them. Big cloud businesses still prefer chips from other makers. These rivals offer more reliable and efficient options. Some promising projects at Intel also hit a wall. Take “Falcon Shores,” for example. It was supposed to be a big deal for AI. But Intel ended up canceling it. Now, Intel is talking about new “Jaguar Shores” graphics chips, but they won’t arrive until 2025. By then, NVIDIA and AMD are already making billions from their special AI designs.
A New Playbook
Intel knows it needs to change course quickly. They recently brought in Lip-Bu Tan as their new CEO. One big sign of change is moving away from something called “IDM 2.0.” This was Intel’s plan to design and make all its chips in-house. It sounded good on paper. But it led to really high costs, big delays in making chips, and less competitive products. It just didn’t work out.
Tan’s focus will be on chip design and making them energy efficient. He wants Intel to build chips that can truly compete with the best AI performance out there right now. The market is moving incredibly fast. Tan understands there’s no time for long-term gambles anymore. Intel needs to show real results, and they need them now.