ARM Holdings: From Cambridge Barn to $32B SoftBank Acquisition

Use IPO windfalls to fund the next pivot.

Apple’s $1.5 M seed in ARM turned into $792 M of exit proceeds during 1998-2000, cash that kept Apple alive while Jobs readied the iMac and killed the Newton.

Why: Early strategic equity can become life-saving liquidity when core business is burning cash.

"Apple... make $792 million in profit from selling their ARM stock over the next couple of years. And literally, that is what saves the company."

RISC achieves 80% of CISC capability at 20% of the transistor count.

Acorn’s ARM1 used ~30k transistors vs Intel 286’s 134k yet delivered equal or better performance for most workloads by executing common instructions in a single clock cycle.

Why: Fewer transistors mean lower cost, lower power draw, and higher speed for the 80% of instructions that dominate real-world usage.

"you could think of it as, it's the 80-20 rule, right? Like these risk... processors... only could do about 80 percent of the instructions that CISC could do, but they did those instructions... much, much faster"

License the architecture, then monetize three ways: upfront fee, engineering services, and per-unit royalty.

ARM charges an initial license for the ISA, fees for co-designing custom chips, and a small royalty on every device shipped—turning pennies into billions across 130 B units.

Why: Aligns ARM’s incentives with customers’ volume success while keeping capex near zero because ARM never fabs chips.

"we want you to ship a bunch of units. So how about we take a small royalty on every device you ship that has our technology in it"

Enter new waves early via partnership, not competition.

Instead of fighting the Wintel duopoly in PCs, ARM pivoted to PDAs with Apple (Newton) and then to Nokia candy-bar phones—markets where power efficiency mattered more than software compatibility.

Why: Low-power niche markets grow into mass markets (smartphones) before incumbents can re-architect.

"it turned out that because there were so many fewer transistors on the chip, it needed much less electricity... it was actually just sucking power from the other components"

Become the default ISA for an entire operating system.

iOS and Android were both built on ARM; switching would require recompiling the OS, toolchains, and every app—an almost insurmountable coordination cost.

Why: Network effects of developer tools, compilers, and app stores create a moat deeper than patents.

"Android and iOS run on risk architecture chipsets. Are they really going to re-architect them for something else? There is no other thing to re-architect to."

Buy the tollbooth to every connected device before the IoT wave.

SoftBank paid $32 B—a 43 % premium—for ARM in 2016, betting that 15 B units/yr would explode into 100 B+ IoT endpoints, each paying ARM a royalty.

Why: Owning the foundational IP captures upside from autonomous cars, AI edge chips, and ambient computing without picking individual winners.

"what would you do if money were no constraint?... I want to buy you guys. And I'm going to offer you $32 billion"

Design chips, don’t manufacture them.

ARM outsources fabrication to TSMC, Samsung, and others, keeping fixed costs low while scaling to every process node.

Why: Capital-light model lets R&D dollars chase the next process shrink instead of building fabs.

"this chip company manufactures a total of zero chips"

Embed engineers inside customer teams to co-design SoCs.

ARM’s 4k employees in 2016 grew to 6k+ under SoftBank, embedding with Apple, Qualcomm, and car OEMs to tailor cores for specific thermal and performance envelopes.

Why: Custom silicon beats generic CPUs in mobile and automotive, ensuring ARM remains indispensable.

"we give them this core processor... we could embed within their teams and we can help them develop essentially custom silicon for their use cases"