ExpertsFMQualcomm
Acquired · November 15, 2022
5 Selected Lessons:
reasoning_pattern · 2 heuristic · 1 framework · 1 priority · 1
Don't over-extend past your core competency
Source
When this applies
when a successful company considers diversifying into adjacent markets that feel strategically appealing but are outside its demonstrated core competency
The takeaway
Rank 'extend the proven core competency deeper' above 'diversify into adjacent-sounding but structurally different markets,' because the latter has a track record of failure even for excellent, well-capitalized teams.
Why it works
Capabilities and ecosystem relationships built over decades (patents, carrier relationships, RF engineering) don't transfer automatically to unrelated product categories, so bets outside the core repeatedly underperform even when well-funded.
One level up
A company's demonstrated pattern of wins and losses across its own history is a more reliable predictor of future success in a new category than the strategic narrative management tells investors.
Where it stops holding
This stops working when the 'adjacent' market genuinely does leverage the same underlying competency (e.g., Snapdragon chips did transfer into automotive/IoT connectivity) rather than being superficially similar.
Detail
hierarchy:
- 1. Defend and extend proven core competency (phone modems/chips) as the primary value driver
- 2. Pursue adjacent bets only where there's demonstrated competency overlap (e.g., connectivity for automotive)
- 3. Be skeptical of narrative-driven TAM expansion (e.g., 'intelligent connected edge' as $700B opportunity) that isn't yet proven
rationale: Qualcomm historically has failed every time they've tried to do servers or watches or smart home or displays — every time they've strayed too far from their core competency, it hasn't been good.
In their words
Qualcomm historically has failed every time they've tried to do servers or watches or smart home or displays. Like every time they've strayed too far from their core competency, it hasn't been good.
Patent-licensing flywheel funding R&D
Source
Acquired — Qualcomm (November 15, 2022) · at 1:43:46 ▸ · at 2:18:47 ▸
When this applies
when a business has one predictable, high-margin revenue stream (IP/licensing/subscriptions) and one more volatile, lower-margin operating stream (product/hardware) — consider whether keeping them together funds a virtuous cycle
The takeaway
A high-margin, durable revenue base justifies and funds continued R&D risk-taking that generates the next generation of IP/product — even if activist investors argue the two businesses should be split for 'focus,' the combination can be the actual moat.
Why it works
Predictable high-margin recurring revenue de-risks R&D investment, which generates new products and defensible IP, which feeds back into the licensing base, compounding advantage over time.
One level up
Durable cash-generative assets are strategic fuel for frontier R&D, not merely a separable 'cash cow' to be spun off for short-term value unlock.
Where it stops holding
This stops working when the cash-generative business becomes a crutch that funds uncompetitive or unfocused ventures rather than genuine frontier R&D (and activist investors' split argument gains merit).
Detail
procedure:
- Maintain a high-margin IP licensing stream.
- Use that predictable cash to fund R&D for new chip products.
- New products and R&D generate new patentable IP.
- New IP feeds back into the licensing stream.
benchmarks: {"licensing margin":"69% earnings before tax margin","chip margin":"34%","total 2022 revenue":"$44 billion, $37B chips / $7B licensing"}
rationale: Because it's a big predictable revenue stream, it lets them take bets on new R&D, and new R&D fuels both new products and new IP for the licensing flywheel.
In their words
And when they do more RD, that fuels the flywheel where they both get new products and they get more IP that they can continue putting into the licensing flywheel.
Betting on Moore's Law before it's obvious
Source
Acquired — Qualcomm (November 15, 2022) · at 1:03:00 ▸ · at 1:16:57 ▸
When this applies
when deciding whether to build a product that is technically infeasible today but could become feasible by the time you ship, based on a predictable technology trend
The takeaway
Design for the cost/performance curve at your delivery date, not at today's levels, when a trend is predictable and few competitors are tracking it.
Why it works
If you can correctly forecast where a well-understood exponential trend (like Moore's Law) will be by your ship date, you can commit to an architecture others dismiss as impossible today.
One level up
Advantage often comes from believing in a slow, well-known trend more precisely and earlier than everyone else, not from secret new information.
Where it stops holding
This stops working when the underlying trend stalls, reverses, or when the timeline to exploit it is too long for your capital runway.
Detail
approach: Identify a theoretically superior but currently infeasible design; map its resource requirements against a known exponential curve; commit if the curve closes the gap within your realistic timeline.
signals:
- technology works in theory but needs more compute than currently exists
- a predictable trend (Moore's Law) is underappreciated by competitors
- multi-year runway exists to reach feasibility
rationale: Few people in 1986 understood Moore's Law well enough to bet a company on 'one or two more turns of the crank' making real-time CDMA decoding possible on a handset.
In their words
But the Qualcomm guys, they know about Moore's Law, which most people didn't know about at that time. And they're like, yeah, I'm pretty sure you give it one or two more turns of the crank on Moore's Law here. And I think we could maybe do this.
Diagnosing 'cornered resource' and other powers in a business
Source
Acquired — Qualcomm (November 15, 2022) · at 2:09:13 ▸
When this applies
when trying to diagnose the structural source(s) of a company's competitive advantage rather than just describing its success
The takeaway
When evaluating a dominant company, separately identify each type of structural power at play (cornered resource, network economy, scale economy, process power) rather than treating 'they're just really good' as sufficient explanation.
Why it works
Different advantages (cornered resources like patents, network effects between complementary products, scale economies in R&D-heavy products, and process power from an irreplicable team) compound differently and have different durability, so naming them precisely clarifies how defensible an advantage really is.
One level up
Durable competitive advantage is usually a composite of multiple distinct structural powers, and naming them precisely is necessary to forecast whether the advantage will persist.
Where it stops holding
This stops working when the categorization becomes an after-the-fact label rather than a predictive tool — i.e., when it's applied retroactively to explain success without testing whether the 'powers' identified would have predicted it in advance.
Detail
approach: List out candidate structural powers (cornered resource, network economy, scale economy, counter-positioning, switching costs, branding, process power) and test each against the company's specific history.
signals:
- patents as cornered resource
- infrastructure-handset interdependency as network economy
- R&D cost spread over huge unit volume as scale economy
- a uniquely talented, irreplaceable team producing a wave of spin-off companies as evidence of process power
rationale: The relative lack of a large successful startup diaspora from Qualcomm compared to Fairchild/Intel's Silicon Valley diaspora is used as a 'proof by example' that Qualcomm's advantage was more about a unique, non-transferable team process than a transferable culture or methodology.
In their words
I think Qualcomm during the golden years that we told the history of had real process power. I think it was equivalent to the Pixar Braintrust. Like that set of people working together under those set of circumstances were wholly unique in the industry and the world.
Evangelize competitors to make customers feel safe, while keeping a trade-secret edge
Source
Acquired — Qualcomm (November 15, 2022) · at 2:15:58 ▸
When this applies
when you are the sole supplier of a new technology and customers are afraid of single-vendor lock-in, deliberately helping competitors reach 'good enough' status while keeping your own edge via undisclosed know-how
The takeaway
When sole-supplier risk is blocking adoption, actively cultivate competitors to just-good-enough status rather than trying to remain the only option — then compete on execution quality you keep proprietary.
Why it works
Customers avoid adopting a technology with only one viable supplier due to dependency risk; enabling credible-but-inferior competitors removes that objection while your retained trade secrets preserve your performance lead and pricing power.
One level up
Sometimes the fastest way to grow your own market is to deliberately strengthen weak competitors just enough to remove the adoption-blocking fear of monopoly dependency.
Where it stops holding
This stops working when you can't control the gap between 'good enough' and 'best,' or when competitors you help catch up entirely and erode your differentiation.
Detail
rule: Patent broadly enough that no one can build the technology without licensing you, but keep key implementation know-how as trade secrets sold via consulting/service engagements.
threshold: Keep competitors 'just good enough' to remove single-vendor dependency fear, not good enough to match your actual performance.
rationale: No phone company was going to take a dependency on CDMA technology when just one vendor existed, so Qualcomm needed to evangelize and create their own competitors while keeping enough secret to remain the performance leader.
In their words
No phone company was going to take a dependency on the CDMA technology when just one vendor existed. And so they needed to evangelize and create their own competitors so that their customers could feel safe with this new technology. But of course, as long as they kept some thing secret of how to eke out the absolute best performance from the innovations, they actually could still be the leader.