In 2019 Intel sold its baseband processor unit, because they figured out that competing with Qualcamm is undoable. Apple back then bought it for 1 billion dollars, and 2200 employees with equipment, leases and IP switched the owner.
The goal of this acquisition was clear: Apple wanted to build a SoC with own baseband modem, and cut costs. The SoC would contain the CPU, GPU and baseband. People estimated that building the first version of such a thing would take 3-5 years.
Fast forward to 2023: Apple just recently renewed its contract with Qualcomm for many years. Why? Because their homebrew baseband processor is a failure. Apple underestimated the complexity and technical challenges. The prototype chips Apple produced were slow, and at least three years behind Qualcomms chips.
Or as the WSJ puts it: "Apple found that employing the brute force of thousands of engineers, a strategy successful for designing the computer brain of its smartphones and laptops, wasn’t enough to quickly produce a superior modem chip."
So instead of using Apple basebands in the newest iPhone, Apple still has to rely on Qualcomm for more years to come. Huawei, who recently introduced their own baseband, is also in the same boat, because their thing consumes more power and heats phones up, which is bad for performance.
Apple is to be expected to continue working on it, and most likely will try to have something better available in 2026, when the contract runs out.
The main obstacles of building such a modem are that there are tons of different regulatory rules around the world, which you have to abide to, as well that it is really, really tough to avoid solutions, which are part of Qualcomm's huge patent portfolio.
A modem is far more complex than any Apple Silicon.
www.theverge.com