For decades, Chipzilla thrived by designing and building its silicon, which worked a treat when it led the pack in chip speeds and transistor size. However, that fairy tale ended years ago when TSMC and Samsung decided to leave it behind.
Those outfits let anyone with a chip design stroll in and get it made in a cutting-edge fab. This meant AMD, Nvidia and other rivals could leap ahead by simply outsourcing their builds to someone competent.
Chipzilla responded by sneakily sending some of its chips to TSMC. That helped avoid complete humiliation, but it torched its profit margins and raised eyebrows about why it was still investing in its fabs.
Investors, unsurprisingly, bailed. Over the past five years, Intel's shares have dropped by about 65 per cent. In that time, AMD more than doubled, TSMC nearly quadrupled and Nvidia grew by a factor of 16.
Former Intel chief executive Pat [kicking] Gelsinger had a vision. Sadly, it looked a lot like setting fire to money. He poured billions into building fabs and chasing foundry dreams, hoping to turn Intel into a credible contract chipmaker. After three years of underwhelming returns and a lack of investor confidence, he was quietly shown the door last December.
Lip-Bu Tan, Intel's new boss since March, hasn't exactly been shouting his strategy from the rooftops. At an event in April, he mumbled vague commitments to being more "customer-centric." Given Intel’s current state, we suspect even its customers aren’t buying that.
The most obvious fix is to break Chipzilla into two. Right now, chip designers don’t want to give business to Intel’s fabs because they’re, well, Intel’s. Separate the factory from the design shop, and maybe Nvidia, AMD and others might consider giving it a go.
The fabs also need a serious dose of customer service, which they lack thanks to a corporate culture that makes Soviet bureaucracy look nimble. TSMC doesn’t just have better tech, it has better manners.
That logic has not escaped Intel’s board or its rivals. TSMC and Broadcom have both been snooping around Chipzilla’s assets this year. But the breakup isn’t simple. Intel’s fabs were built exclusively for Intel designs, and its engineers have never really had to work with outsiders.
Northland Capital Markets semiconductor analyst Gus Richard, who once drew a Chipzilla paycheque, said: "They can create a [chip] design environment that works for Intel, but it’s hard if not impossible for anybody else to use, and they’re failing at getting additional customers."
Even Gelsinger knew something had to give. Two years ago, he pushed Intel toward an internal foundry model, letting chip designers use either internal fabs or external ones. Last September, Intel announced plans to spin off the factories into a separate subsidiary with its board and directors. The results have not been pretty. Intel's factories lost $13.4 billion last year and incurred another $2.32 billion in losses in the first quarter of 2025. Investors aren’t exactly scrambling for a slice of that.
Still, Intel is dangling the idea of bringing in outside capital. Chief Financial Officer David Zinsner told The Wall Street Journal in January that talks had begun with strategic investors and even customers.
For now, Intel is surviving on its substantial $9 billion cash reserve and divesting its non-core businesses. Tan’s latest trick has been mass layoffs and cost-cutting; a clear sign he’s prepping the books for something drastic.
Chipzilla dreams of breaking even by 2027, with some help from cloud outfits like Microsoft and Amazon. But unless Nvidia and friends start queueing up, that’s a pipe dream.