How Intel's factory became a $10.3 billion annual drain on its own products

In the middle of 1985, Andy Grove put a hypothetical question to Gordon Moore: what would a new chief executive do with Intel’s memory business? The company had been losing money on memory chips for three years and had yet to agree on a course of action.

Intel, founded in 1968, went on to set the pace for the computer industry for about forty years. IBM selected an Intel processor for its Personal Computer in 1981. The Windows-Intel pairing, widely known as Wintel, then locked software and hardware makers into a huge ecosystem. Intel’s ability to manufacture chips was its greatest advantage.

The arrangement had a hidden cost. Intel designed and made its products under one roof, so trouble in a factory flowed directly into the products it could sell. Rivals developed a different division of labour: one business designed chips while another manufactured them.

TSMC became the manufacturer serving chip designers across the industry. It began risk production of its 7nm-branded N7 process in 2016 and moved into volume production in 2018. Samsung also began mass production of its 7nm 7LPP devices in 2018. Intel’s 10nm and 7nm-class processes arrived years late.

The timing coincided with a shift in where demand was growing. Phones, graphics processing units and artificial intelligence became the growth engines. Huawei unveiled its Kirin 980, made by TSMC, on 31 August 2018. Apple followed days later at its September event with the TSMC-fabricated A12 Bionic.

On 7 July 2019, AMD officially launched its Ryzen 3000 line of central processing units, using TSMC’s 7nm process and the Zen 2 microarchitecture. AMD’s Rome EPYC 2 processors, built on TSMC’s N7 node, offered up to 64 cores and 128 threads for datacentre work.

The financial arithmetic now captures the difficulty. Intel’s product divisions earned $12.7 billion of operating income in 2025, while its manufacturing arm lost $10.3 billion. Intel’s annual report says leading-edge processes “require manufacturing volumes beyond what we expect from our own products to achieve economic efficiency”.

TSMC controls 72.5% of the foundry market. Intel’s experience shows the pressure built into its old model: a factory once central to its strength must now find enough volume to justify the cost of staying at the leading edge.