Apple’s $30 Billion Broadcom Deal Rewards the One Chip Apple Can’t Copy

Ten months ago, Apple shipped an iPhone with Broadcom’s most visible chip engineered out of it. This week, it committed more than $30 billion to the same supplier — the largest single pledge in the history of its American Manufacturing Program. Both decisions are rational. The space between them is the most precise public map yet of where Apple’s chip ambitions end.

The agreement announced July 8 runs for multiple years and covers advanced radio frequency components — including FBAR filters — and wireless connectivity technologies, all built in the United States. It includes a $1.5 billion expansion of Broadcom’s Fort Collins, Colorado, facility and is expected to yield more than 15 billion American-made chips. Reporting that preceded the announcement says the supply arrangement now stretches through 2031; Broadcom shares jumped as much as 5% in premarket trading when the extension surfaced on July 6, before Apple attached a dollar figure.

Apple already proved it can fire Broadcom

The reason this deal reads strangely is that Apple spent the past five years demonstrating, in shipping products, that it can replace Broadcom’s work. Bloomberg reported in December 2024 that Apple had built an in-house Bluetooth and Wi-Fi chip — then codenamed Proxima — specifically to reduce its reliance on Broadcom. Nine months later it arrived: the N1 networking chip shipped across the entire iPhone 17 lineup in September 2025, handling Wi-Fi 7, Bluetooth 6 and Thread — functions Apple previously bought from Broadcom silicon. The C1 modem did the same to Qualcomm’s slice of the phone. Apple’s insourcing program is not a threat; it is a track record.

Buried in that December 2024 reporting was the caveat that explains this week: even after its own connectivity chip, “Apple will still use a Broadcom-designed radio frequency filter for modems.” That sentence, eighteen months later, is worth $30 billion.

The chip Apple can’t copy is barely a chip

FBAR filters — film bulk acoustic resonators — keep a phone’s dozens of radio bands from bleeding into one another. They are acoustic, analog devices. Their performance comes from piezoelectric materials science and process control on a factory floor, not from transistor design, which is why a company that out-designs nearly everyone at digital logic still buys them from a plant in Colorado. TSMC, which fabricates every chip Apple designs, does not make them.

Fort Collins does. Apple was already supporting more than 1,100 jobs at Broadcom’s FBAR facility there when the two companies signed the previous, undisclosed “multibillion-dollar” version of this deal in May 2023 — back when Apple’s US pledge was $430 billion over five years, not $600 billion over four. The filter problem also compounds rather than shrinking: each new frequency band and carrier combination a phone supports adds filtering content, so the component Apple cannot replace is the one it needs more of every year.

What the Street had written down

Wall Street does not pay for Broadcom’s Apple business; it pays for AI. At Thursday afternoon prices — $405.65, up 4.4% as of 2:47 p.m. Eastern on July 9 — Broadcom is a $1.76 trillion company at 61.6 times trailing earnings and 19.3 times forward estimates, with a consensus analyst target of $523.73. The market’s attention sits almost entirely on custom AI accelerators, the business TECHi examined when Google, Meta and Anthropic commitments made June’s earnings a referendum on the ASIC pipeline.

The wireless segment carried an asterisk instead: Apple is roughly 20% of Broadcom’s total annual revenue, a concentration the market treated as a slow leak because the customer was publicly designing its way out — the tension TECHi mapped in its Apple versus Broadcom comparison this spring. A disclosed $30 billion commitment through 2031 — back-of-envelope, roughly $5 billion a year, though Apple published no schedule — converts that asterisk into visibility. It will not move the AI narrative, but it re-prices the downside case: the segment bears assumed was melting now has a signed floor under it, and the stock still trades about 18% below its 52-week high of $494.18 after the chip-sector pullback of the past two weeks.

The $600 billion umbrella — and what it doesn’t tell you

The deal is the headline act of Apple’s American Manufacturing Program, launched in 2025 with an initial $100 billion inside a $600 billion, four-year US investment pledge. That pledge did not appear spontaneously — it followed tariff threats from the Trump administration aimed at moving iPhone production onto US soil. Announcing $30 billion of chips from Colorado buys Apple something no filter provides: political coverage for the far larger share of its supply chain that remains in Asia.

One caveat belongs in any honest read of the number. “Expected to exceed $30 billion” is Apple’s own multiyear tally, and AMP announcements fold in spending that was already flowing — the 2023 agreement covered the same factory and the same filters. The verifiably new facts are the disclosed figure, the 2031 horizon and the $1.5 billion expansion. How much of the $30 billion is incremental business for Broadcom is not knowable from the outside, and neither company broke it out.

What to watch from here

• Broadcom’s next earnings call. Management has so far let Apple do the talking. Whether the deal shows up as quantified wireless-segment guidance is the first hard test of how much is new money — TECHi’s AI ASIC earnings preview covers the other half of that print.
• The N1’s 2026 roadmap. Bloomberg’s reporting has Apple’s networking chip reaching iPad and Mac this year. Broadcom’s connectivity content in Apple products keeps shrinking even as its RF content is locked in — the net effect on that 20% revenue share cuts both ways.
• Purchase-obligation disclosures. If the commitment is as firm as the press release implies, it should eventually surface in the companies’ filings rather than only in newsroom posts.
• The next AMP headline. Apple has most of a $600 billion pledge left to allocate and a demonstrated preference for supplier deals big enough to make the White House’s case for it.

Apple ran the make-or-buy experiment in public, and both answers came back. The N1 proved Apple can replace Broadcom’s chips. The $30 billion proves it knows exactly which ones it can’t.