ExpertsFMHamilton Helmer & Chenyi Shi on How to Build an AWS-Like Second Business
Acquired · April 4, 2023
5 Selected Lessons:
framework · 4 reasoning_pattern · 1
Value creation vs. value capture distinction for transforming decisions
Source
Acquired — Hamilton Helmer & Chenyi Shi on How to Build an AWS-Like Second Business (April 4, 2023) · at 32:23 ▸
When this applies
when deciding which of several plausible growth directions actually captures durable profit rather than merely identifying a market opportunity
The takeaway
Treat opportunity-generation (where to look) and defensibility-assessment (whether you can keep what you capture) as two separate, sequential analyses; popular frameworks only do the first.
Why it works
Popular growth narratives identify where economic value might exist (demand/opportunity) but say nothing about whether the company can defend a share of that value once competitors arrive; without a power analysis, 'good idea' and 'good business' are conflated.
One level up
Any heuristic that reliably predicts success must address capture, not just creation, of value.
Where it stops holding
This stops working when the capture mechanism (power) genuinely doesn't apply to the business type (e.g., quickly commoditized markets where capture is irrelevant and share is transient regardless).
Detail
procedure:
- Use popular narratives (customer obsession, TAM expansion, broad industry redefinition) as idea generators for where value might exist.
- Then separately assess, via power analysis, whether the company can actually capture and keep that value.
- Reject any narrative-only approach as insufficiently predictive, since it conflates value creation with value capture.
rationale: Companies following the same popular narrative (e.g., broad industry definition) succeed (Disney) or fail (Uber mobility) depending on whether they could actually capture value, not on following the narrative itself.
In their words
a lot of these popular narratives kind of tell you about economic value, but they don't tell you much about business value
Business definition test: does your power extend to the new segment?
Source
Acquired — Hamilton Helmer & Chenyi Shi on How to Build an AWS-Like Second Business (April 4, 2023) · at 37:14 ▸
When this applies
when deciding whether to extend into any adjacent market, geography, or customer segment
The takeaway
Before expanding, identify precisely what structural element creates your current defensibility, then test whether that specific element — not just your brand or technology — transfers to the new context.
Why it works
If the drivers of your current power (e.g., shared fixed-cost content, geographic density) transfer to the new segment, the expansion inherits the existing defensibility at low incremental risk; if the power driver is segment-bound (e.g., local network density), the new segment is effectively a brand-new business requiring you to rebuild the expensive part from scratch.
One level up
Treat every expansion decision as first a diagnostic of 'is my moat the same here,' not a question of market size or strategic fit alone.
Where it stops holding
This stops working when the apparent power source is largely non-structural (e.g., brand recognition without network/scale effects), in which case the test may wrongly suggest safety where none exists.
Detail
situation: Netflix expanded into Korea assuming international streaming was 'the same business'; Uber expanded into China assuming ride-sharing was inherently an international, transferable business.
decision: Netflix leveraged shared content fixed-cost economics across countries; Uber had to rebuild rider/driver density from scratch against an entrenched local competitor (Didi).
outcome: Netflix's expansion worked because the cost driver (content) transferred; Uber's failed because its real power (localized two-sided network density) did not transfer and the expensive part (acquiring both sides of the marketplace) had to be redone entirely.
lesson: Test literally whether the mechanism of your power, not your technology or brand, carries into the new segment before expanding.
In their words
The key thing there is, if you have an established business, is the drivers of power in that extensible to that additional segment you're considering, whether it's geographic or customer or whatever.
Two-by-two matrix for where to expand: diversification, reinvention, co-action
Source
Acquired — Hamilton Helmer & Chenyi Shi on How to Build an AWS-Like Second Business (April 4, 2023) · at 50:36 ▸
When this applies
when any organization is choosing an adjacent space to expand into, beyond its current core
The takeaway
Classify any expansion candidate along the skills-shared / needs-shared axes before committing, and heavily prefer the co-action quadrant (shared skills, different need) since it has the best empirical track record.
Why it works
Expansion directions can be classified by whether they share customer needs and/or company skills with the current business; sharing neither (pure diversification) means building a wholly unknown business with large agency/bureaucracy costs and no edge; sharing needs but not skills (reinvention) is rare opportunity and usually opposed internally by incumbents with P&L power; sharing skills but serving a different need (co-action) preserves capability advantage while opening new markets, and empirically produced ~90% of new value among the largest US companies studied.
One level up
Expansion risk is best managed by anchoring new ventures in what the organization already structurally knows how to do, even when the market served is new.
Where it stops holding
This stops working when a company possesses a truly proprietary/distinctive capability that itself constitutes defensible power (rare, per Corning's glass science), in which case even 'different needs, different skills' moves can work — or when reinvention is the only viable path despite internal resistance.
Detail
procedure:
- Map a candidate expansion on two axes: does it serve the same customer need, and does it use the same company skills?
- Same need + same skills = your current business (no move needed).
- Different need + different skills = pure diversification (rarely works; consider spinning off entrepreneurs instead).
- Same need + different skills = reinvention (rare opportunity, requires founder sponsorship against internal resistance, e.g., Netflix into streaming).
- Different need + shared skills = co-action (where ~90% of SP100's new business value came from, e.g., AWS).
benchmarks: {"co_action_share_of_new_value":"~90% of new business value in SP100 per Helmer's recollection"}
rationale: Unrelated diversification carries large agency costs and no informational edge; co-action preserves capability edge while accessing new demand.
In their words
I don't have the numbers in front of me, but I think it's 90% of the value in the SP 100 of the new stuff came from coaction. And that's saying something that's pretty straightforward.
Acquisition logic: why is the asset worth more to you than to the seller?
Source
Acquired — Hamilton Helmer & Chenyi Shi on How to Build an AWS-Like Second Business (April 4, 2023) · at 1:08:25 ▸ · at 1:12:18 ▸
When this applies
when any buyer is evaluating an acquisition that isn't purely financial arbitrage
The takeaway
Justify any acquisition premium by identifying a structural reason (power-related) the asset specifically is worth more under you than under the seller or any other buyer — not by assuming cost synergies will cover the premium.
Why it works
Sellers generally know their asset better than buyers (information asymmetry), and empirical M&A studies show acquirers tend to roughly break even on average while sellers capture most gains; cost-synergy arguments (reduced overlap) usually wash against the diseconomies of being part of a larger bureaucracy, so a genuine case for paying a premium must rest on something structural — typically power — not headcount savings. Separately, cost savings are capped while new revenue capture is unbounded, and the cheapest new revenue is revenue whose underlying cost of production/exploitation is also favorable (e.g., an underexploited brand).
One level up
In any transaction with asymmetric information, the burden of proof for value creation lies in identifying a truly structural edge, not in generic efficiency arguments that likely wash out.
Where it stops holding
This stops working when genuine cost synergies are unusually large and verifiable (e.g., eliminating fully duplicated physical infrastructure), in which case cost-based logic can still apply, though the speakers suggest this is atypical.
Detail
procedure:
- Ask why this asset is worth more to you specifically than to the seller, given the seller knows more about it than you do.
- Distrust cost-synergy justifications (headcount overlap reduction) since diseconomies of scale in large orgs tend to roughly offset the savings.
- Look instead for a power-related reason (e.g., the acquirer can plug the asset into an existing high-switching-cost customer base, or can fully exploit an under-leveraged brand/franchise).
- Prefer acquisitions aimed at new revenue over cost savings, since cost savings are capped but new revenue capture is unbounded — and favor cases where the cost of generating that revenue is also favorable (not just the revenue itself).
rationale: Antitrust authorities specifically scrutinize horizontal acquisitions that amplify scale-economy advantage, underscoring that power-based rationale (not cost-cutting) is the real driver of acquisition value.
In their words
Well, it's very related. If you're acquiring a company, the primary question you have to ask yourself is why is this worth more to me than to the seller?
Why founders and VCs both struggle to see power early
Source
Acquired — Hamilton Helmer & Chenyi Shi on How to Build an AWS-Like Second Business (April 4, 2023) · at 23:39 ▸ · at 26:52 ▸
When this applies
when any decision-maker tries to extrapolate from a single recent success, or judges early-stage ventures mainly on top-line momentum
The takeaway
Distinguish between signals of market demand (growth) and signals of durable structural advantage (persistent margin); don't let the former substitute for the latter in decision-making.
Why it works
Early success is some unknown mix of skill and luck; founders lack a basis to decompose it, so they wrongly assume repeatability. Similarly, VCs' business model and the absence of a profitability track record push evaluation toward top-line growth, which signals demand but not defensibility.
One level up
Be suspicious of generalizing from a single successful outcome without being able to attribute how much of it was skill versus luck.
Where it stops holding
This stops working when a company's skill-based process really is reliably repeatable and documented, reducing the luck component significantly.
Detail
approach: Recognize that a first success contains an unknown skill/luck mix; don't assume the process will replicate. Separately, recognize that growth-focused investor evaluation (used because profitability isn't yet observable) is a proxy for demand, not for durable power.
signals:
- founder claims 'I will just repeat the exact same process again'
- investors citing top-line growth as the main metric
- VC complaints that companies 'waited too long to IPO' signaling a late realization the real driver was bottom-line not top-line
rationale: Mistaking growth or a single win for durable advantage leads to overconfident repetition of unproven playbooks.
In their words
It could be skill, it could be luck, of course, it is some combination of it, but you have no idea the percentage that's skill and the percentage that luck