← Back to list
Colossus (Invest Like the Best / Business Breakdowns)Podcast2 Jun 2020Source: investlikethebest.libsyn.comHost: Patrick O'Shaughnessy

Ben Thompson – Platforms, Ecosystems, and Aggregators - [Invest Like the Best, EP.176]

In plain words

This piece argues that tech may not be as special as it thinks—smartphones and cloud computing have hit their natural limits, and giants like Apple, Google, Amazon, Microsoft, and Facebook could become as entrenched as the Big Three automakers after the 1930s. Ben Thompson explains his 'Aggregation Theory': aggregators like Google and Facebook profit from user habits and scale, while platforms like Windows and iOS App Store rely on technical lock-in. Key holdings: Netflix (actively burning cash on original content, seen as smart long-term), Spotify (music squeezed by record labels, but podcast strategy promising), and Shopify (empowers small merchants; high failure rate costs nothing, success pays off).

AI SummaryAI-generated · may contain errors · verify against the original

Ben Thompson, in his appearance on the Invest Like the Best podcast, delved into how the internet has reshaped the business landscape, centering his discussion on his well-known "Aggregation Theory" and the "Smiling Curve." He argues that successful aggregators (such as Amazon and Netflix) are built

~14 min full read · 9 sections
Deep Analysis

Ben Thompson – Platforms, Ecosystems, and Aggregators

At a Glance

Ben Thompson is the author of the tech strategy newsletter Stratechery and co-host of the Exponent and Dithering podcasts. This episode centers on his Aggregation Theory and Smiling Curve, offering a deep analysis of how the internet has reshaped the logic of business competition.

The most consequential takeaway from the entire episode: Ben Thompson argues that the tech industry may not be as "special" as it believes itself to be — smartphones and cloud computing have reached the natural endpoint of the general-purpose computing paradigm, and the currently dominant companies (Apple, Google, Amazon, Microsoft) are likely to become long-term, unshakeable cornerstones, much like the Big Three automakers after the 1930s, rather than being disrupted by the next wave. This judgment directly challenges Silicon Valley's narrative of "perpetual self-disruption."


Theme 1: The End of Tech "Exceptionalism" — Smartphones and Cloud Computing Have Reached Their Limits

Thompson argues that the tech industry is not a series of discrete "eras" (mainframe → PC → internet → mobile), but a continuous evolution that has reached the natural endpoint of the general-purpose computing paradigm.

  • Interaction layer evolution: From "destination batch processing" (mainframe rooms) → "desktop personal computers" (still requiring sitting at a desk) → "always-on smartphones." Thompson notes that the smartphone is the "truly personal computer" — large enough to get work done, small enough to fit in a pocket, and always connected. "Watches are too small, AR glasses are too specialized — they are all dedicated devices that cannot replace the smartphone's versatility."
  • Computing/storage layer evolution: From "a room-sized mainframe" to "cloud computing spanning the globe." Thompson believes the next step (e.g., edge computing) is merely a supplement for specialized scenarios, not a disruption of the existing paradigm.
  • Historical analogy: Thompson compares the current phase to the automotive industry in the 1930s — "Between 1900 and 1930, hundreds of car companies emerged in the U.S. every decade; by 1930, the top three were set, and for the next 67 years, almost no new players appeared until Tesla." He implies the same may hold true for the current Big Five tech giants (Apple, Google, Amazon, Microsoft, Facebook).

> "Maybe tech isn't so special after all."


Theme 2: Aggregation Theory and the Smiling Curve

Thompson distinguishes between "Aggregators" and "Platforms"—a core concept for understanding the business models of internet giants.

  • The Smiling Curve: Originally proposed by Acer founder Stan Shih, it describes how value in the PC industry chain concentrates at the two ends—one end is brand/marketing/user relationships (high value), the other is patents/core technology/manufacturing processes (high value), while the middle (production and assembly) is low value. Thompson extends this to the internet: Aggregators occupy the "user relationship" end (high value), differentiated suppliers occupy the "core technology" end (high value), and the middle layer (e.g., traditional publishers) gets squeezed.
  • Three characteristics of an Aggregator:

1. Direct relationship with users (users come to you directly, not through intermediaries)

2. Zero marginal cost to serve users (infinite scalability)

3. User acquisition costs decline with scale (the product improves as more users join)

  • Essential difference from Platforms: Platforms (e.g., Windows, iOS App Store) connect multiple parties via APIs and drivers, and their network effects are "externalized"—they require highly differentiated suppliers to attract users. Aggregators (e.g., Google, Facebook) have "internalized" network effects—they absorb user data through their own products to form feedback loops, and suppliers are highly commoditized. Thompson emphasizes that this distinction is critical for regulation: "The power of an aggregator comes from user choice, not technological lock-in."

> "An aggregator, I can always go around and form a one-on-one connection. A platform type moat is all about making connections that can't be made without the platform." (Meaning: An aggregator's moat is user habit and scale—you can bypass it; a platform's moat lies in the connections it enables, which cannot be bypassed.)


Theme 3: Supplier Power Determines Aggregator Profitability – A Comparison of Netflix and Spotify

Thompson points out that an aggregator's profitability depends on the degree of supplier differentiation and bargaining power.

  • Spotify's music business is a "weak aggregator": The three major record labels possess irreplaceable "historical catalogs," and new artists still require label support. "The record labels have immense bargaining power over Spotify because they can easily pivot to Apple Music." As a result, Spotify earns razor-thin margins on music.
  • Spotify's podcast strategy is an attempt at being a "strong aggregator": The podcast industry is highly fragmented, and suppliers (independent creators) lack unified bargaining power. Spotify attracts users through exclusive content (e.g., Joe Rogan) and leverages its user data (email, geographic location, payment information) to achieve ad targeting precision far beyond traditional dynamic insertion. "If Spotify's CPM is significantly higher than the industry average, podcasters will come to it proactively in the future, rather than Spotify paying them."
  • Netflix's "vertical integration": Netflix has shifted from "licensing third-party content" to "producing original content," transforming suppliers from external partners into internal production units. "Netflix front-loads all its cash flow in exchange for retaining all future revenue for itself—Orange is the New Black will remain on Netflix forever, continuously attracting new users." This is precisely why Netflix's free cash flow was deeply negative from 2017 to 2019—it was actively transforming its business model, not mismanaging operations.
  • The right way to compete with Netflix: "Taking on Netflix head-on is foolish—it can spread the cost of a show across 180 million users, while you cannot." The correct approach, as demonstrated by Disney, is to use streaming as a "funnel top" to drive other businesses like theme parks and cruise lines, rather than competing on subscription revenue in Netflix's own arena.

Theme 4: Shopify vs. Amazon – Differentiated Supplier vs. Aggregator

Thompson argues that Shopify and Amazon represent two fundamentally distinct competitive logics, and that Shopify's model is closer to a "platform" than an "aggregator."

  • Amazon is a classic aggregator: It controls user demand, and suppliers are highly commoditized. "When you sell on Amazon, you use Amazon's labels, Amazon's website, Amazon's boxes—you even pay Amazon to manage your inventory." Suppliers tolerate this because that is where the users are.
  • Shopify is an "anti-aggregator": It empowers suppliers to build their own direct customer relationships. "Shopify's success lies not in how many merchants succeed, but in how many try—950,000 fail, 50,000 succeed, and that is a win for Shopify." Because Shopify's marginal cost of service is zero, failed merchants incur no cost, while successful ones generate revenue.
  • Shopify's long-term challenge: "The gravity of the internet always flows toward whoever owns user attention." Shopify's merchants still need to acquire users through Facebook/Google ads, meaning Facebook will eat up a significant portion of profits. Thompson believes Shopify's true moat lies in becoming the "e-commerce infrastructure layer"—like Windows connecting OEMs and application developers, linking small merchants with underlying services such as logistics and payments. "No single 3PL company can individually serve a small cookie shop, but Shopify can coordinate in the middle."

Theme 5: From Scarcity to Abundance — Value Chain Restructuring in the Internet Era

Thompson argues that the core feature of the internet is the replacement of "scarcity" with "abundance," fundamentally altering which segments can generate profits.

  • Constraints of the Old World: The physical world is full of constraints—limited broadcast channels, limited shelf space, limited time (a 22-minute episode can only accommodate 8 minutes of ads), and limited purchasing power.
  • Logic of the New World: "On the internet, all constraints disappear. Netflix can stream endlessly; Amazon has infinite shelf space; Facebook and Google have infinite ad inventory." Therefore, "a decline in costs is a good thing, not a bad thing"—when Google went public, the drop in CPC was misinterpreted as a negative signal, but it actually meant Google was adding inventory faster, implying greater long-term potential.
  • Logic for Winners: Either become the "starting point" from which users set out (aggregators), or become a highly differentiated "destination" (independent creators/brands). "The middlemen—traditional publishers, traditional retailers—are being squeezed."
  • Insights for the Media Industry: "Selling 'content' on the internet is a loser's game—content is just bits with zero marginal cost. What you need to sell is 'service'—continuously making users feel informed and inspired." Thompson uses himself as an example: he sends four emails per week, and these emails are the "byproduct" of the service, not the product itself.

Mentioned Positions

Position Guest Stance Key Data
Netflix Bullish long-term, believes its proactive business model shift is the right strategy Free cash flow deeply negative from 2017-2019, but due to upfront costs; 180 million user base dilutes content costs
Spotify Bullish on podcast strategy, notes music business is constrained by suppliers Podcast ad CPM expected to far exceed industry levels; exclusive Joe Rogan deal
Shopify Bullish on its platform model, but notes long-term dependence on Facebook ads Merchant failure rate is extremely high but costs Shopify nothing; marginal service cost is zero
Amazon Viewed as a typical aggregator, suppliers are commoditized Infinite shelf space; merchants bear all costs
Apple Platform power is strong, App Store control is a "potential abuse of platform power" iOS/Android are the "endpoint" of the smartphone era
Google Typical aggregator, driven by user choice SEO industry is essentially suppliers paying to optimize themselves; declining CPC is a positive signal
Facebook Typical aggregator, internalizes network effects Stories increase ad inventory, declining CPM is a positive
Disney Bullish on its streaming strategy, believes the model is superior to Netflix's direct competition Cross-subsidizes streaming via theme parks/cruise lines and other businesses
Epic Games Bullish on its platform/infrastructure logic Game engine licensing threshold raised from $50,000 to $1 million; Fortnite is free-to-play, monetized through cosmetic items
Uber Believes its business model has a fundamental misalignment Spends money simultaneously on acquiring users and recruiting drivers, inconsistent with aggregator logic
Stripe Viewed as an infrastructure platform, bullish on its cost structure moat Scale brings cost advantages, enabling investment in developer tools and documentation
Walmart Believes its traditional e-commerce strategy has failed, but grocery business has advantages Grocery supply chain differs from Amazon's; Walmart has existing infrastructure

Judgments Worth Remembering

1. Thompson believes the tech industry may no longer be "special"—smartphones and cloud computing have reached the end of general-purpose computing, and current giants could remain as entrenched as the Big Three automakers after 1930. Support: Both the interaction layer (phones → watches/AR represent specialization, not upgrades) and the computing layer (cloud → edge represent specialization, not upgrades) have hit the natural limits of generality.

2. "Aggregators" and "platforms" are two entirely different business models, and regulators must treat them distinctly. Support: Aggregators derive power from user choice (Google/Facebook), while platforms' moats come from technological lock-in (Windows/iOS App Store). Applying antitrust tools to aggregators often amounts to "pushing a rope"—Europe's penalty against Google Shopping failed to change user behavior.

3. Netflix's "negative cash flow" is not a problem, but a signal of an active business model transformation. Support: Shifting from "buying third-party content" to "producing proprietary content" means front-loading all costs in exchange for retaining all future revenue. In a zero-interest-rate environment, this is a savvy capital move.

4. Directly confronting Netflix head-on is foolish—the correct competitive approach is like Disney's, using streaming as a funnel top to drive other businesses. Support: Netflix can spread the cost of a show across 180 million users; competitors cannot. Disney Plus's low-price strategy is rational because its returns come from cross-selling theme parks, cruises, etc.

5. Shopify's high merchant churn rate is a positive signal, not a negative indicator. Support: Marginal service cost is zero, and failed merchants incur no cost; a large number of trials increases the probability of hitting a "jackpot"—this is the logic of an "abundance" world.

6. "Selling 'content' on the internet is a loser's game—you need to sell 'service'." Support: Content is just bits with zero marginal cost; service is about continuously making users feel informed and inspired. Thompson's emails are a "byproduct" of the service, not the product itself.

7. Spotify's podcast strategy is essentially about ad monetization, not subscription growth—it aims to replicate Facebook's advertising model. Support: Spotify allocates content costs to its ad division; it has users' emails, locations, and payment information, enabling precise advertising. If CPM is far above industry average, podcasters will proactively join, creating a positive feedback loop.

8. "Abundance" replacing "scarcity" is the core economic feature of the internet—cost declines are a good thing, not a bad thing. Support: At Google's IPO, CPC decline was misunderstood; it was actually inventory growing faster than consumption. Facebook Stories increased ad inventory, and CPM decline is a long-term positive. The old-world "scarcity mindset" is entirely obsolete.