How Mark Zuckerberg Built Meta: Facebook’s Rise, Its Biggest Crises, and the Fight to Control the Future of Computing

By 2024, Meta had built something few institutions in history could claim: a network reaching approximately half the world's population.

Facebook, Instagram, WhatsApp, Messenger and Threads collectively served nearly four billion people every month. More than three billion used Meta's applications daily. The company had survived the collapse of early social networks, Google's attempts to compete, the smartphone revolution, privacy scandals, Apple's advertising restrictions and TikTok's rapid expansion.

Its success was not inevitable. Facebook repeatedly came close to losing its advantages. On several occasions, Mark Zuckerberg chose to sacrifice profitable existing businesses to protect what he believed would become more important.

In a 2024 episode of Acquired, hosts Ben Gilbert and David Rosenthal examined the company's history, including interviews and accounts from people involved in its development. Their investigation revealed a recurring pattern: Zuckerberg built dense networks, protected his authority, invested in several possible futures and redirected the company when technology changed.

The financial results were extraordinary. Meta generated $134.9 billion in revenue in 2023. By 2025, annual revenue had reached $201 billion. In September 2026, its applications averaged 3.64 billion daily users, according to Meta's October 21, 2026 earnings release.

But Meta's history raises questions beyond how Facebook made money. How much of its dominance came from better products, and how much from buying competitors? Why did privacy failures continue despite repeated warnings? Can a company remain successful while spending tens of billions on technology that may never become profitable?

The answers begin with an unusual arrangement: one person retained the authority to make decisions that most public companies would struggle to approve.

1. Zuckerberg’s Control of Facebook Changed Its Entire History

Sean Parker helped build a company its founder could not easily lose

Mark Zuckerberg's control over Facebook was established long before the company became enormously valuable.

In 2004, Facebook's original business financing came from Zuckerberg and Eduardo Saverin, who each contributed approximately $1,000. Their initial arrangement gave Zuckerberg two-thirds of the business and Saverin one-third.

That ownership structure changed as Facebook grew and moved from Harvard to Palo Alto. Saverin stayed on the East Coast for an internship and later returned to Harvard while Zuckerberg, Dustin Moskovitz and other members of the team continued building the company in California.

Saverin's ownership eventually fell to approximately 2% during Facebook's corporate reorganization, creating a major dispute over his financing contributions, operational involvement and treatment. The Acquired hosts did not establish a definitive judgment about whether the dilution was fair.

The more consequential long-term influence came from Sean Parker.

Parker had co-founded Napster and Plaxo. At Plaxo, he experienced the loss of control of a company he had helped create. When he became involved with Facebook in 2004, he warned Zuckerberg about venture investors obtaining board control and replacing founders with professional executives.

Parker encouraged Zuckerberg to build a governance structure that protected his authority while still accepting outside financing.

That protection shaped Facebook's future. Zuckerberg could reject takeover offers, choose expensive investments and withstand shareholder pressure without facing the same immediate risk of dismissal as a conventional chief executive.

Parker also introduced Zuckerberg to important Silicon Valley figures, including Reid Hoffman and Mark Pincus. Hoffman recommended Peter Thiel as an early investor.

Thiel participated in a financing round of approximately $500,000 at a $5 million pre-money valuation. Facebook later raised $12.7 million from Accel in 2005 at a $98 million post-money valuation, retaining an unusually founder-friendly governance structure.

Kevin Efrusy and Jim Breyer were important to Accel's investment. Breyer also invested personally. Facebook used venture debt from Western Technology Investment to support infrastructure development and obtained additional funds through arrangements that included early founder liquidity.

These were not ordinary financing decisions for a young internet company. Venture investors were accepting less control than was customary because they considered Facebook's growth sufficiently attractive.

The Yahoo acquisition was closer to happening than the popular story suggests

Zuckerberg's determination to remain independent was tested in 2006.

Facebook's growth had slowed. Its original college market was becoming saturated, and expansion into high schools and workplaces was less successful than expected.

Viacom offered approximately $750 million to acquire Facebook. Yahoo subsequently proposed a transaction valued around $1 billion.

The familiar story suggests Zuckerberg immediately rejected Yahoo's offer because he always knew Facebook would become enormous. The account presented by Acquired is more complicated.

Zuckerberg initially allowed the deal to advance toward documentation.

Then Yahoo reported disappointing results, and its share price declined approximately 20%. Yahoo chief executive Terry Semel proposed leaving the stock component unchanged, reducing the effective value of the proposed transaction to approximately $800 million.

Zuckerberg walked away.

Some Facebook executives had favored selling and expected substantial personal liquidity. The collapse of the negotiations contributed to management departures.

Moskovitz remained committed to the company. Zuckerberg, meanwhile, concluded that Facebook needed to generate more dependable revenue so that financial uncertainty would not again place it in a position where selling appeared necessary.

He also believed two developments already underway—News Feed and unrestricted registration—could restart growth.

This was not simply a billionaire refusing another billion dollars. At the time, Facebook had not yet demonstrated that it could become a universal social network.

Founder control made risky transitions possible

The pattern repeated throughout Facebook's history.

Zuckerberg declined subsequent acquisition opportunities, including a complicated Microsoft proposal reportedly valued around $24 billion. He continued investing during the company's troubled 2012 IPO, persisted through the advertising crisis of 2022 and financed long-term artificial-intelligence and hardware programs despite substantial losses.

The Acquired hosts emphasize a necessary qualification: retaining founder control does not automatically produce good decisions.

Zuckerberg combined control with product judgment, technical understanding, competitiveness and an unusual willingness to revise strategy. Other founders with similar power might not have produced similar results.

His control remains a central part of Meta's governance.

An important development occurred after the original interview. On October 21, 2026, Meta announced plans to create publicly listed, non-voting Class C shares, effective in January 2027.

The proposed structure would provide additional flexibility for capital raising, employee compensation and acquisitions while maintaining the existing distribution of voting power. Meta explicitly connected the change to preserving Zuckerberg's founder-led leadership.

More than twenty years after Sean Parker warned about losing control, Meta was still organizing its capital structure around the same concern.

2. Facebook Started With Dense Communities, Not a Global Audience

Zuckerberg was building social software before Harvard

Zuckerberg was born on May 14, 1984, in Dobbs Ferry, New York. His father, Ed, was a dentist interested in computers and mathematics; his mother, Karen, was a psychologist.

Three interests shaped his childhood: programming, classical history and strategic video games.

He became particularly attached to Sid Meier's Civilization, a game in which players accumulate resources, develop technology and pursue several possible routes to victory.

The hosts draw a comparison between that game's strategic flexibility and Zuckerberg's later approach to business. Rather than commit everything to one possible outcome, Facebook repeatedly invested in multiple technologies and competitive positions.

Zuckerberg learned C++ around age ten, received private programming lessons and developed projects including ZuckNet, which connected computers in his family's home and his father's adjoining dental practice.

AOL Instant Messenger provided another important influence.

AIM, launched in 1997, became an enormous communications network with approximately 36 million users by 2001. It was free, accessible and easily modified. Young programmers experimented with automated messages, custom interfaces and software that extended its capabilities.

Zuckerberg was one of those programmers.

After transferring from Ardsley High School to Phillips Exeter Academy, he met Adam D'Angelo, another teenager fascinated by computing.

In 2002, the two developed Synapse, a music-recommendation program that predicted what someone would enjoy listening to next. Winamp and reportedly Microsoft expressed interest in the software. The hosts mention a possible acquisition worth around $1 million but acknowledge that the amount is uncertain.

The experience mattered because Zuckerberg learned that software written by students could possess commercial value.

D'Angelo went to Caltech while Zuckerberg attended Harvard. They remained in contact, discussing programming projects over AIM.

At Caltech, D'Angelo developed BuddyZoo, which analyzed users' AOL contact lists and the relationships connecting them. It attracted several hundred thousand users.

Unlike Synapse, which helped an individual choose music, BuddyZoo became more useful as more people joined.

D'Angelo also discovered Friendster in 2003. It was a dedicated social network backed by venture investors, with millions of members and features that combined friendship discovery with dating-related interactions.

These products gave Zuckerberg and D'Angelo practical experience with recommendation systems, online identities and network effects before Facebook existed.

CourseMatch and Facemash supplied important ingredients

At Harvard, Zuckerberg lived in Kirkland House with Chris Hughes, Dustin Moskovitz and Billy Olson.

He built several small applications, including CourseMatch, which allowed students to see what classes their friends were considering.

It demonstrated that ordinary information about other people's choices could become surprisingly engaging.

Facemash was more controversial.

Built in 2003, the application displayed photographs of Harvard students and invited users to compare their attractiveness. Zuckerberg obtained photographs without permission from institutional resources, creating a privacy dispute and university disciplinary proceedings.

The website attracted sufficient traffic to strain campus systems. Zuckerberg was placed on disciplinary probation rather than expelled.

The episode also illustrated how inexpensive it had become to develop consumer software. Using technologies such as Linux, Apache, MySQL and PHP, an individual could create a website without purchasing expensive proprietary infrastructure.

Facemash made Zuckerberg known as a capable programmer. Cameron and Tyler Winklevoss subsequently approached him about helping develop HarvardConnection, another proposed online student directory.

The resulting dispute eventually led to a reported $65 million settlement, including approximately $45 million in Facebook shares.

Zuckerberg also met his future wife, Priscilla Chan, during the period surrounding the disciplinary proceedings.

The important product lesson was that university relationships, photographs, profile information and online identities could be combined into something students genuinely wanted to use.

Facebook's early restrictions made the network stronger

TheFacebook.com launched on February 4, 2004.

Students needed Harvard email addresses to register. They could create profiles, upload photographs, identify classmates, inspect social connections, list courses and find contact information.

There was no News Feed, no elaborate photo-sharing service and no ecosystem of games.

The product was essentially a searchable student directory created and maintained by students themselves.

That restricted environment became its advantage.

Because members belonged to the same university, new users already knew a substantial proportion of the people appearing on the website. Facebook did not need millions of strangers to make the service useful.

According to contemporary reporting in The Harvard Crimson, approximately 650 people joined during its earliest day of operation. Within two weeks, membership exceeded 4,300.

That contemporary account is important because it independently confirms the explosive adoption described in the interview.

Facebook's popularity came partly from the unusually intense engagement of a small community. Students repeatedly browsed profiles, examined classmates' interests and followed connections between people they knew.

The founders called this behavior the trance.

Facebook rapidly introduced its Wall, allowing public messages and conversations on members' profiles. The company also added functionality that encouraged repeated visits.

Early engagement was extraordinary: approximately 70% of registered users were active on a typical day.

Facebook expanded one campus at a time

Instead of immediately opening the service to everyone, Zuckerberg selected new universities carefully.

Columbia was chosen partly because an existing social network operated there. Entering a competitive campus would test whether Facebook could win against an established service.

Stanford and Yale followed.

Each university initially operated as a separate network. Students at Columbia could not freely browse Harvard profiles.

This helped preserve the sense of belonging that made the original product attractive. It also simplified engineering because smaller networks were easier to manage.

Waiting lists generated demand before Facebook opened at new institutions, increasing the likelihood that new members would immediately encounter people they knew.

Within approximately three to four months, Facebook was operating at more than 100 colleges.

Harvard's reputation also contributed. A product that succeeded at Harvard attracted attention from students at other prestigious institutions.

There were limits, however.

When Facebook attempted to establish workplace networks, employees were often less interested in sharing their personal lives with colleagues. High-school expansion was complicated by the lack of standardized institutional email addresses.

These difficulties mattered because Facebook could not achieve Zuckerberg's ambitions by serving university students forever.

The company needed a product that remained useful after graduation.

That led directly to News Feed.

3. News Feed Turned Facebook Into a Daily Media Business

From visiting profiles to receiving updates automatically

Facebook's early product roadmap included a website redesign, photographs, events, local businesses, entertainment and a personalized newspaper.

The newspaper concept became News Feed.

Engineers observed that users repeatedly visited their friends' profiles to discover whether anything had changed.

An early experiment highlighting recent profile changes increased engagement. That suggested Facebook could save users considerable effort by bringing updates together.

But constructing a personalized feed was technically difficult.

Facebook needed systems that could collect activity from millions of profiles, determine which updates mattered to each person and distribute the results efficiently.

Early News Feed versions updated information in batches roughly four times daily because the company lacked sufficient computing capacity for continuous personalization.

The feature launched on September 5, 2006.

It changed Facebook from a collection of individual destinations into a continuous stream of information.

Instead of deciding whose profile to visit, users increasingly received updates selected by Facebook.

That shift became fundamental to both user engagement and advertising.

The first major privacy revolt

News Feed immediately provoked strong opposition.

Facebook reportedly received approximately 30,000 angry messages on launch day, and roughly 10% of users joined an anti-News-Feed group.

Their concern was understandable.

Information that previously required effort to locate was now automatically displayed to large numbers of people.

Even where formal access permissions remained unchanged, the practical visibility of information increased enormously.

News Feed therefore revealed an important distinction: information can technically be accessible without users expecting it to be actively distributed.

There was also a commercial risk. Facebook previously earned display revenue when users opened many individual pages. A consolidated feed could reduce page views and advertising impressions.

Zuckerberg nevertheless refused to withdraw the product.

Engagement data showed that people were using it extensively, including to discover the protests against News Feed itself.

Facebook introduced additional privacy controls, and the opposition weakened within approximately two weeks.

The decision established a lasting management principle: users' observed behavior could sometimes reveal more about product value than their immediate reactions.

But the principle was dangerous when applied too broadly. Later controversies demonstrated that a feature could generate engagement while still violating reasonable privacy expectations.

Photographs, tagging and events reinforced the feed

News Feed was part of a broader product system.

Facebook Photos allowed users to share extensive collections of photographs. Photo tagging connected images to identifiable people and their social relationships.

Events produced invitations and attendance information. Parties and gatherings produced photographs, which generated tags, comments and more notifications.

Each feature made the others more useful.

Rather than maintain separate businesses equivalent to Flickr, Eventbrite and local social directories, Facebook connected these behaviors inside one network.

In November 2005, before News Feed's introduction, Facebook reportedly generated approximately 230 million page views daily. By October of that year, it had reached about 8.3 million users and approximately $1 million in monthly advertising revenue.

Its engagement was unusually strong even before the major feed transformation.

Opening registration restarted growth

News Feed launched while Facebook was preparing to open registration beyond institutional networks.

Because of the backlash, Zuckerberg postponed the expansion briefly.

Late in September 2006, anyone could join.

The combination of open registration and News Feed worked.

New registrations accelerated from approximately 5,000–10,000 per day during the summer to roughly 70,000 per day by late fall.

Facebook had escaped the boundaries of its original university market.

The consequences were not entirely positive. As parents, employers and older adults joined, younger users became increasingly aware that their social audiences were changing.

This helped create opportunities for services offering different sharing conventions, including temporary photographs and more private conversations.

Facebook had solved one growth problem while creating the conditions for future competition.

4. Facebook Built an Operating System Inside a Website

The developer platform attracted thousands almost immediately

On May 24, 2007, Facebook launched a platform allowing outside developers to build applications using its APIs, social relationships and user information.

Facebook had more than 20 million users by then.

The company hoped to recruit 5,000 developers during the platform's first year. It reportedly reached that target within two days.

Developers could distribute games, quizzes and other applications through Facebook's existing social network.

Zynga became an important beneficiary. Games including FarmVille used social connections, invitations and activity updates to attract users.

This fundamentally expanded Facebook's purpose.

People could now use the service not only to follow real-world relationships but also to play games and participate in entirely digital activities.

The platform created commercial opportunities through advertising and virtual payments, including Facebook Credits.

However, the same incentives also generated enormous amounts of spam.

Developers wanted to encourage users to contact friends because doing so produced free distribution. Facebook experienced the resulting deterioration in notifications and News Feed quality.

Eventually, the company restricted automatic application promotion.

Developers who had depended on free viral distribution increasingly purchased advertising instead.

That change improved Facebook's control over the experience while expanding its paid advertising business.

The ambition was bigger than social networking

Facebook initially wanted to become an indispensable environment for software developers, comparable in some respects to Microsoft Windows.

The obstacle was that Facebook did not own the underlying operating system.

Its applications operated inside web browsers, which themselves depended on hardware and software controlled by other companies.

Later, Apple and Google's smartphone platforms would expose this dependency more severely by controlling application distribution, payments and access to operating-system features.

Facebook also experimented with another potential business, Wirehog, during its early Palo Alto period. It was a college-oriented file-sharing service, and the founders seriously considered whether Facebook should serve partly as its distribution network.

Reid Hoffman encouraged Zuckerberg to focus on Facebook instead, particularly because the legal risks of music and movie sharing resembled those surrounding Napster.

The Like button became a commercial signal

Facebook subsequently acquired FriendFeed, a social-sharing service developed by engineers including Bret Taylor and Paul Buchheit.

FriendFeed had introduced the Like button.

Facebook adopted the feature, choosing a simple response that required less effort than writing a comment.

Engineers initially worried that Likes would replace more meaningful interaction. Instead, the feature encouraged participation and supplied valuable information about users' preferences.

A Like became both social feedback and an input for recommendation algorithms.

Businesses and public figures accumulated followers through their Facebook pages, hoping to distribute posts to those audiences.

Over time, Facebook reduced the unpaid distribution available to commercial pages. Businesses increasingly needed to purchase advertising to reach people who had previously followed them voluntarily.

External Like buttons also spread Facebook's infrastructure across the web, creating additional opportunities to understand user behavior outside its applications.

That expansion of data collection would eventually become a major source of regulatory conflict.

5. Facebook Turned Growth Into an Engineering Discipline

The growth team measured relationships, not just registrations

By 2008, Facebook had experienced another slowdown in user growth.

The company's response was unusual for the period. Rather than simply increase advertising expenditure, it created a team responsible for improving growth through changes to the product itself.

The original group included Chamath Palihapitiya, Alex Schultz, Naomi Gleit and Javier Oliván.

Their work combined functions normally divided among engineering, marketing and product management.

They studied registration processes, incomplete sign-ups, email invitations, notifications, user activity and retention. They identified obstacles preventing people from becoming regular Facebook users and developed experiments to remove them.

A central discovery concerned the number of meaningful connections someone established shortly after joining.

People who quickly found enough real friends were much more likely to remain active.

The team developed practical activation goals such as finding seven friends within ten days or ten friends within fourteen days.

These numbers were not permanent rules about human relationships. They were product-management tools identifying users likely to become engaged.

People You May Know became an important part of this strategy.

By analyzing contact lists, mutual friends and other social signals, Facebook helped new members locate people they already knew.

The relevance of those suggestions mattered enormously. A list containing a user's closest friends could make the service immediately useful. A large collection of strangers would achieve much less.

Translating Facebook became part of expanding it

Language presented another growth obstacle.

Facebook allowed users to help translate its interface, review suggestions and adapt terminology to regional preferences.

This approach accelerated international expansion while involving early members in introducing the product to their communities.

It also allowed variations within languages, such as regional differences in Spanish.

Facebook grew from approximately 100 million users in August 2008 to around 150 million at year-end. By the end of 2009, it had approximately 350 million users.

The daily-to-monthly active-user ratio, a measure of how frequently the monthly audience returned, improved from around 50% toward 70% over subsequent years.

The growth team's influence remained visible in Meta's later leadership. Oliván became COO, Schultz became CMO, and Gleit continued into senior management.

Meta also invested in internet-access initiatives, telecommunications partnerships and infrastructure intended to bring more people online.

As the number of reachable internet users approached saturation, expanding internet access itself became another possible growth strategy.

Continuous experimentation became institutional knowledge

Facebook supported these methods with internal experimentation systems.

Engineers could release changes to selected users, compare results, monitor technical performance and reverse unsuccessful features.

The company developed tools allowing numerous simultaneous experiments while attempting to distinguish the effects of each one.

This mattered because Facebook could learn from actual behavior instead of treating every product decision as an irreversible prediction.

Over time, the ability to test, measure and improve became an organizational advantage.

New employees could use established infrastructure rather than rebuild experimentation tools themselves.

The company increasingly treated product-led growth as part of its operating system.

6. Sheryl Sandberg Helped Turn Attention Into an Advertising Business

Microsoft supplied early advertising revenue

Facebook was already attracting extraordinary attention before it developed a sophisticated advertising business.

By late 2005, it reportedly generated more daily page views than Google, despite having far fewer users.

But Google had an established advertising operation, while Facebook remained dependent on comparatively ordinary display advertisements.

Microsoft helped close that gap.

In August 2006, Microsoft agreed to sell Facebook's US display advertising inventory and provide guaranteed advertising rates.

The arrangement gave Facebook revenue without requiring an extensive internal sales organization.

Microsoft also needed Facebook. Google dominated search advertising, and Microsoft was seeking ways to expand its online advertising position.

In October 2007, Microsoft expanded its advertising relationship worldwide and invested $240 million in Facebook at a valuation of approximately $15 billion.

Google's interest strengthened Facebook's negotiating position.

According to the interview, the final Microsoft negotiations continued through the night at Facebook's Palo Alto offices. A company hackathon began while negotiations were underway, organized by Javier Oliván as part of Facebook's internationalization work.

The agreement was completed the following morning.

Microsoft also received selected product integrations, including Bing Maps within Facebook Events.

The investment eventually proved highly profitable. The hosts estimate Microsoft realized approximately $8 billion as its Facebook shares were sold after the IPO.

Sandberg professionalized Facebook's commercial operations

Sheryl Sandberg joined Facebook in 2008 after helping develop Google's advertising organization.

Her arrival established a division of responsibilities that shaped the company for years.

Zuckerberg concentrated on products, engineering and long-term technology. Sandberg managed advertising, operations, commercial relationships and important parts of organizational development.

Facebook began treating advertising as a central business rather than an unavoidable way to pay operating expenses.

It explored several alternatives along the way.

Beacon attempted to make commercial transactions visible through social activity. Facebook Credits tried to capture value from virtual payments. The developer platform created opportunities to charge for transactions and distribution.

Ultimately, targeted advertising became the dominant model.

Facebook possessed information about relationships, interests, activities and engagement that could help advertisers identify relevant audiences.

As the advertiser marketplace expanded, its algorithms could choose from more advertisements. Better selection could improve commercial results while reducing irrelevant advertising.

The business expanded rapidly.

Facebook generated approximately $153 million in revenue in 2007, $280 million in 2008, nearly $800 million in 2009, $2 billion in 2010 and $3.7 billion in 2011.

User growth reinforced advertising demand, while advertising income financed the infrastructure needed for continued expansion.

Advertising benefited from network effects of its own

Facebook's advertising advantage was not simply that it possessed information about billions of people.

Advertisers also benefited from the size of the marketplace.

More advertisers increased competition for relevant audiences. More users created additional opportunities to select appropriate advertisements. More engagement produced signals that improved recommendations.

This created a reinforcing relationship between advertising scale, targeting quality and revenue.

Facebook also became an important way to discover mobile applications.

Game developers and other companies purchased advertisements encouraging people to install software. Users could click an advertisement and proceed directly to an app-store listing.

That business placed Facebook in an increasingly uncomfortable relationship with Apple: Facebook earned substantial revenue influencing which applications people discovered inside Apple's ecosystem.

The disagreement became much more important after the smartphone revolution.

7. The Smartphone Nearly Destroyed Facebook’s Business Model

The 2012 IPO exposed a serious weakness

The iPhone launched in 2007. Apple introduced third-party native applications in 2008.

At first, the consequences for Facebook were not obvious.

Facebook continued growing through desktop computers and web browsers. Its advertising systems were designed around those environments.

As smartphones became more important, however, Facebook confronted several connected problems.

Its desktop advertisements did not fit naturally on small screens. Its web-based developer platform could not exercise the same control inside mobile operating systems. App stores introduced approval processes that restricted Facebook's ability to release changes immediately.

Apple and Google also controlled payments and access to important device capabilities.

Facebook initially relied extensively on HTML5-based mobile development, attempting to retain some of the flexibility of the web. The resulting applications often performed poorly.

Meanwhile, more users were spending more time on smartphones.

The danger was visible before Facebook went public.

In its April 2012 registration statement, Facebook disclosed that 488 million monthly users accessed its mobile products in March 2012.

The company acknowledged that it did not yet generate meaningful revenue directly from mobile usage and that successful monetization remained unproven.

This was a remarkable position for a company preparing one of the largest technology IPOs in history.

Facebook priced its shares at $38 on May 17, 2012, raising approximately $16 billion.

Its public debut was troubled by NASDAQ trading problems, investor skepticism and concerns about its mobile future.

General Motors had publicly withdrawn Facebook advertising spending shortly before the IPO, adding to questions about the effectiveness of its commercial model.

The stock initially traded above its offering price but soon declined.

On September 4, 2012, shares reached approximately $17.68, more than 53% below the IPO price.

Employee share-sale restrictions were also approaching expiration, threatening additional market pressure.

Facebook had reached enormous scale, but investors were questioning whether its business could survive the next computing platform.

Zuckerberg and Sandberg chose to cannibalize desktop revenue

The response required accepting a difficult trade-off.

Facebook could protect existing desktop revenue by prioritizing the products and advertising placements that already worked.

Alternatively, it could redirect its strongest engineers toward mobile advertising, risking weaker short-term financial results while rebuilding the business.

Zuckerberg and Sandberg chose the second approach.

Andrew Bosworth, who had worked on Facebook's profile and Timeline systems, was moved into mobile advertising.

Bosworth worked with Will Cathcart, an engineer associated with News Feed who later became WhatsApp's leader.

Their first solution was relatively modest: introducing sponsored brand recommendations into areas previously used for friend suggestions.

It generated some mobile revenue but did not resolve the underlying problem.

The larger project was called Project Whale Shark.

Its purpose was to place native advertisements directly inside News Feed, the central experience of Facebook's mobile application.

This was different from squeezing desktop banners onto smaller screens.

A News Feed advertisement would appear where users were already directing their attention. It would compete with ordinary posts rather than occupy a separate, easily ignored section.

The approach carried substantial risks.

If advertisements were irrelevant or overwhelming, users might abandon the product. If Facebook restricted them too aggressively, it would fail to generate sufficient revenue.

Initially, the advertisements were not particularly good.

Bosworth had planned a six-month honeymoon and sabbatical after completing the engineering work. Zuckerberg wanted him to remain involved because building the product was only the first step.

Facebook still needed to convince advertisers to purchase the new placements and then improve the system based on their experience.

Bosworth shortened his planned absence and worked with Sandberg on advertiser relationships.

The new advertising format became more valuable than the old one

Facebook discovered that prominent mobile advertisements could command significantly more attention than desktop sidebar placements.

The company expanded its advertiser marketplace and invested heavily in ranking systems capable of selecting more relevant advertisements.

Zuckerberg initially worried about the quality and volume of ads. Bosworth argued that a larger, better-functioning marketplace was necessary to improve relevance.

The distinction mattered.

Restricting advertising volume could reduce annoyance, but it would not necessarily produce better advertisements. Increasing the number of eligible advertisers and improving selection technology could make each impression more valuable.

By the fourth quarter of 2013, mobile advertising accounted for approximately 53% of Facebook's total advertising revenue, compared with approximately 23% a year earlier.

Facebook reported $2.34 billion in advertising revenue that quarter, representing 76% annual growth. These figures are confirmed in its official fourth-quarter 2013 earnings release.

The turnaround had taken approximately eighteen months.

Facebook subsequently extended the same advertising infrastructure to Instagram, allowing advertisers to buy placements across both applications through a shared system.

The technology that initially threatened Facebook's existence became the foundation of a larger business.

The episode also clarified why founder control mattered. Zuckerberg could authorize a companywide transition that deliberately weakened existing revenue to develop a replacement.

8. Instagram and WhatsApp: Growth Strategy or Buying the Competition?

The Instagram acquisition was designed to buy time

Facebook announced its agreement to acquire Instagram on April 9, 2012.

The approximately $1 billion purchase was extraordinary at the time.

Instagram had around 27 million users and little meaningful revenue. Facebook already had roughly 900 million users, but Instagram was growing rapidly within mobile photography.

Zuckerberg negotiated the acquisition over a weekend, acting with limited internal consultation.

The company kept Instagram operating as a distinct product instead of immediately absorbing it into Facebook.

The episode explains this as part of a wider acquisition strategy: preserve promising networks, remove immediate competitive pressure and incorporate their successful features into Facebook's larger ecosystem.

Independent evidence makes the competitive motivation especially clear.

In February 2012, Zuckerberg exchanged emails with colleagues about possible acquisitions. He considered both integrating attractive product capabilities and neutralizing emerging competitors.

One of his explanations was:

“What we're really buying is time.”

The statement later appeared in the US House Judiciary Committee's published hearing record.

Zuckerberg reasoned that buying emerging networks could give Facebook time to integrate their features before competitors achieved comparable scale.

The distinction was significant.

An acquisition could produce commercial value by adding users and capabilities. It could also reduce competitive pressure by placing an independent rival under Facebook's ownership.

Both explanations could be true simultaneously.

WhatsApp protected Facebook against a different kind of network

The next major acquisition concerned private messaging.

WhatsApp was not simply another version of Facebook.

It allowed people to communicate directly, often across national boundaries and different mobile operating systems, without requiring participation in a public social feed.

Facebook already offered Messenger, but WhatsApp's growth revealed a broader behavioral transition.

People were becoming more comfortable sharing certain information through smaller, private conversations instead of publicly visible posts.

The hosts described the distinction as a movement from the town square to the living room.

In February 2014, Facebook agreed to acquire WhatsApp for approximately $19 billion.

The purchase strengthened its presence in private communications and reduced the possibility that a rapidly growing messaging network would develop into a broader competitor.

WhatsApp largely retained its independent identity and functionality, rather than being folded into Facebook's public-feed experience.

That separation also allowed Meta to serve different communication needs within one corporate portfolio.

Snapchat showed why new sharing formats mattered

Facebook's acquisition strategy was not always successful.

It attempted to acquire Snapchat, but the negotiations did not produce a deal.

Snapchat introduced sharing conventions that challenged established social networks.

Temporary photographs and Stories reduced the pressure to publish permanent, carefully selected posts.

Instagram had increasingly become a place where users worried about how their profiles appeared to others. A temporary Story offered a lower-commitment alternative.

Facebook responded by incorporating similar features into Instagram and other applications.

Instagram Stories improved aspects of the interface, using a horizontal row of profile images and recommendation systems to decide which Stories appeared prominently.

Facebook had previously responded to Twitter's status updates and later developed Facebook Live to compete with services such as Meerkat and Periscope.

These examples illustrate an important feature of Meta's competitive history.

It did not consistently invent the original interaction format. It often identified a successful format elsewhere, adapted it and distributed it through an enormous existing audience.

Onavo provided intelligence about rival applications

Facebook also acquired Onavo, an Israeli mobile-data company, in 2013.

Onavo provided data-compression and VPN-related services. Its technology also gave Facebook information about how people used competing applications.

That intelligence helped identify the growth of products such as WhatsApp and Snapchat.

Competitive analysis had become harder on mobile because app stores and operating systems restricted access to information about competing products.

Onavo supplied an alternative source of insight.

It also raised serious privacy questions. People using a service to manage data consumption or protect their connections might not expect their activity to inform its owner's acquisition and competitive strategies.

The antitrust questions continued long after the interview

Facebook's acquisition strategy eventually attracted major regulatory scrutiny.

In 2020, the Federal Trade Commission sued Facebook, alleging that its acquisitions of Instagram and WhatsApp, together with restrictions affecting developers, formed part of an unlawful strategy to maintain a social-networking monopoly.

The case required distinguishing legitimate business expansion from acquisitions intended to suppress competition.

Meta contested the allegations.

A federal district court ruled in Meta's favor on November 18, 2025. The FTC then appealed that ruling in January 2026.

As of October 2026, the appeal remained a material unresolved legal issue.

The evidence supports two observations without requiring a premature legal verdict.

Instagram and WhatsApp became enormously successful products within Meta. Zuckerberg's own internal communications also show that neutralizing potential competition was part of the acquisition analysis.

Whether the transactions constituted unlawful monopolization is a separate question for the courts.

9. Cambridge Analytica Exposed Facebook’s Privacy Failures

News Feed and Beacon revealed an early misunderstanding

Facebook's privacy controversies did not begin with Cambridge Analytica.

News Feed had already demonstrated that users could object when personal information became substantially more visible, even without changes to formal access permissions.

Beacon created a more serious problem.

Launched in 2007, Beacon allowed participating external websites to share certain purchases and activities through Facebook.

A transaction made on another website could become visible to friends who were not expected to know about it.

Facebook initially responded by introducing stronger controls, moving toward explicit permission and eventually discontinuing the feature.

The incident showed why the successful News Feed response could not become a universal management rule.

Increased engagement did not necessarily mean that users accepted a feature's treatment of personal information.

Developers could obtain information about people who had never installed their apps

The developer platform introduced another privacy risk.

Facebook originally allowed applications to request extensive personal information. Depending on the permissions and period, developers could obtain information about users' friends.

That meant one person's decision to install an application could expose information associated with other people who had never personally authorized it.

Facebook also changed visibility settings and public-profile features in ways that sometimes conflicted with users' expectations.

The company entered into a settlement with the FTC in 2011, formalized through a 2012 consent order, addressing misleading privacy representations and requiring additional safeguards.

The underlying problem remained: Facebook had allowed third parties to obtain information that could be difficult or impossible to retrieve after distribution.

Cambridge Analytica demonstrated the consequences.

What Cambridge Analytica actually obtained

Cambridge Analytica was a political consulting company promoting voter profiling and advertising services.

Its work involved data collected through a Facebook personality-quiz application developed by researcher Aleksandr Kogan.

The original interview estimates that approximately 320,000 people completed the relevant quiz. There is an important difference between that retrospective figure and the numbers appearing in contemporaneous records.

Facebook stated in March 2018 that approximately 270,000 people downloaded the application. The FTC subsequently described approximately 250,000–270,000 US app users whose information was collected, together with information associated with tens of millions of their Facebook friends.

These figures reflect different descriptions and populations. They should not be treated as interchangeable counts of willing participants.

The broader exposure was much larger because the application could collect information associated with friends who had not installed it.

Facebook estimated that information associated with as many as 87 million users might have been improperly shared.

The FTC's investigation of Kogan and former Cambridge Analytica chief executive Alexander Nix alleged deceptive collection practices involving personal information used for voter profiling and targeted advertising.

Facebook had restricted access to friends' data for newer applications, but legacy access arrangements allowed some existing developers to continue collecting information during a transition period.

The company had requested deletion of the relevant data in 2015. However, enforcing deletion after information had left its systems proved difficult.

Cambridge Analytica therefore exposed weaknesses in both Facebook's original permissions model and its ability to control downstream uses of information.

Cambridge Analytica’s influence on elections remains a separate question

The scandal became closely associated with the 2016 US presidential election.

But the episode separates several different issues that are frequently combined into one narrative.

Russian-linked operations used fake accounts and pages to spread political content. Some publishers produced fabricated or sensational political stories to generate advertising revenue. Cambridge Analytica promoted psychological voter profiling. Political campaigns also used Facebook's conventional advertising tools.

These activities involved different participants and mechanisms.

The Cambridge Analytica controversy generated especially dramatic claims about the ability of psychological profiles to predict and influence voting behavior.

The hosts argue that the available evidence did not establish that Cambridge Analytica's techniques decisively influenced the election.

They point to doubts about the models' predictive accuracy, including the difference between performance on training data and performance when predicting unfamiliar people.

That distinction matters. A system that explains information already used to develop it may be much less useful when applied to voters whose behavior is unknown.

The hosts also distinguish Cambridge Analytica's techniques from ordinary digital campaign operations.

Donald Trump's campaign made extensive use of Facebook advertising. Barack Obama's earlier presidential campaigns had already demonstrated the platform's usefulness for organizing supporters and targeting political messages.

The hosts consider the Trump campaign's conventional advertising operations potentially more important than Cambridge Analytica's claimed psychological methods. This remains their interpretation, not proof that any particular advertising technique determined the election.

The central privacy violation does not depend on establishing electoral impact.

Personal information was obtained and used under arrangements that exposed millions of people to risks they had not knowingly accepted.

Facebook paid $5 billion, but the consequences extended further

In July 2019, Facebook agreed to pay $5 billion to settle FTC charges concerning privacy violations.

The settlement addressed misconduct broader than Cambridge Analytica alone.

According to the FTC's official announcement, the company was charged with violating its earlier privacy order and misleading users about their control over personal information.

The agreement imposed additional requirements involving privacy governance, oversight and accountability.

The reputational consequences were substantial.

Facebook's advertising business generally did not require selling identifiable user databases directly to advertisers. Advertisers purchased access to selected audiences while Facebook retained control over the underlying targeting infrastructure.

But that distinction did not eliminate concerns about what information Facebook collected, how it obtained consent or what developers could access.

Public expectations were also changing.

In Facebook's earlier years, extensive social sharing had often been presented as desirable. By 2018, many users were more interested in private messaging, controlled audiences and limits on tracking.

Facebook's historical practices increasingly conflicted with those expectations.

Content moderation became a global institutional responsibility

Privacy was only one element of Meta's growing social responsibilities.

Its applications operated across languages, cultures and legal systems with different standards concerning expression, defamation, harassment, political activity and harmful material.

Meta developed large moderation teams, contractor operations, automated detection systems, policies and escalation procedures.

It faced difficult questions about hate speech, fabricated information, violent content and live broadcasts.

These systems imposed major costs. They also became capabilities that smaller competitors would struggle to reproduce at comparable scale.

The hosts regard this as an unusual competitive consequence of Facebook's controversies.

Years of regulatory scrutiny and public criticism forced Meta to develop specialized expertise that subsequently became part of its operational advantage.

The existence of that expertise does not establish that Meta's moderation decisions are consistently correct. It means that operating a comparable global platform requires far more than building an attractive application.

10. TikTok Changed the Basis of Social-Media Competition

Facebook's existing network was no longer sufficient

For most of Facebook's history, its strongest protection was the social graph: the network of relationships connecting its users.

A rival could develop better software, but convincing millions of people to recreate their relationships elsewhere was difficult.

Friendster and Myspace had demonstrated that the strength of a network could change. Facebook nevertheless spent years benefiting from the enormous switching costs associated with established connections.

TikTok challenged that advantage differently.

It did not require users to find friends before enjoying the product.

Instead, TikTok presented personalized entertainment selected by recommendation algorithms.

A new user could open the application and immediately encounter videos likely to be interesting, regardless of whether anyone they knew was participating.

TikTok's parent, ByteDance, had developed substantial recommendation capabilities through earlier products, including Toutiao, before expanding internationally through TikTok and the acquisition of Musical.ly.

The underlying innovation was not simply short video.

It was the ability to distribute engaging content without depending primarily on existing social relationships.

That weakened Facebook's traditional defense.

Users could spend more time on TikTok without migrating their friends or abandoning Facebook accounts.

Reels required Facebook and Instagram to change what they showed users

Meta responded through Reels, its short-video format within Instagram and Facebook.

The company increasingly recommended content created by people users did not follow.

This represented a major change in the purpose of its feeds.

Originally, Facebook helped users discover information about people they knew. TikTok demonstrated the commercial potential of predicting what entertainment someone wanted to consume, regardless of its source.

Meta began combining both approaches.

A user might discover an entertaining video through Reels and then send it privately through WhatsApp, Messenger or Instagram messaging.

Public media discovery and personal communication could become distinct but mutually reinforcing activities.

The transition was not financially painless.

Reels initially generated less advertising revenue than Facebook's established News Feed formats. Encouraging users to consume more Reels therefore shifted attention toward a less mature commercial product.

Nevertheless, Zuckerberg treated the transition as necessary.

The hosts argue that losing advertising effectiveness was dangerous, but losing the audience itself would be worse.

Apple's tracking restrictions made the transition harder

The TikTok competition coincided with another major threat.

Apple introduced App Tracking Transparency in 2021, requiring applications to seek permission before tracking users across other companies' apps and websites.

Many users declined.

Meta lost access to signals that had helped it target advertising, attribute purchases and measure whether campaigns produced results.

The effects were especially significant for app-install advertising and e-commerce.

In its February 2022 earnings discussion, Meta estimated that Apple's iOS changes would create an advertising revenue headwind of approximately $10 billion during 2022.

The figure was a company estimate, not a precise independent measurement of lost revenue.

Meta's former CFO Dave Wehner explained that the calculation involved uncertainty, and the company could not identify a single exact amount.

Apple's policy also had competitive implications.

Apple's hardware-centered business differed from Meta's advertising-centered model. Stronger privacy restrictions could benefit users while simultaneously limiting the information available to competing advertising platforms.

Meta argued that some restrictions affected app-based advertising differently from search advertising.

The hosts cite an example of a business selling physical goods whose Facebook advertising costs approximately doubled during the transition.

Such businesses could not necessarily replace their lost efficiency by purchasing advertisements elsewhere in Apple's ecosystem.

Meta lost hundreds of billions before rebuilding its advertising system

Meta had already experienced a historic financial decline in July 2018, when warnings about slowing growth and privacy-related costs contributed to a stock collapse of approximately 19% in one day.

That wiped roughly $119 billion from its market value.

In February 2022, Meta experienced another record-breaking decline.

Its market value fell approximately $232 billion in a single trading session.

The company had reported the first quarterly decline in Facebook's daily users, while Apple tracking restrictions and TikTok competition intensified concerns about its future.

From February to its October 2022 low, Meta's market value declined approximately 72%.

The company responded by improving its recommendation systems and rebuilding advertising capabilities around information still available under the new restrictions.

Products including Advantage+ expanded automated campaign management, targeting and optimization.

Meta's enormous quantity of internal engagement data, engineering resources and computing infrastructure made it better positioned than many smaller competitors to develop alternatives to cross-application tracking.

Reels also improved as a product and advertising format.

The recovery illustrated a recurring feature of Meta's history: when the surrounding technology changes, the company is willing to redesign major parts of its business rather than preserve existing arrangements indefinitely.

However, the distinction between the mobile crisis and the TikTok crisis is important.

In 2012, Facebook already possessed the audience but lacked the ability to monetize its new behavior.

With TikTok, the company faced the possibility that audiences themselves would move their attention elsewhere.

That is a more fundamental competitive threat.

11. Meta Invested in Artificial Intelligence Years Before ChatGPT

Yann LeCun established Facebook's long-term AI research program

Facebook's AI strategy began well before large language models became widely popular.

Around 2012–2013, Zuckerberg and engineering leader Mike Schroepfer began examining the technologies that might define future computing.

The smartphone revolution had exposed Facebook's dependence on Apple and Google. AI offered both immediate improvements to Facebook's existing applications and an opportunity to develop strategically important technology internally.

In 2013, Facebook recruited Yann LeCun, a leading artificial-intelligence researcher, to establish Facebook AI Research, commonly known as FAIR.

LeCun wanted to remain connected to New York University, continue academic teaching and publish research rather than operate exclusively within a closed corporate laboratory.

Facebook accepted those conditions.

The arrangement helped the company attract researchers who valued scientific publication and academic collaboration.

Facebook concentrated its research around AI technologies relevant to its products and anticipated future needs.

Early practical applications included recognizing people in photographs, improving tagging, ranking social content and selecting advertisements.

Unlike companies investing in AI without an established commercial use, Facebook could immediately benefit from incremental improvements in recommendations.

Better predictions meant more relevant content, stronger engagement and potentially more valuable advertising.

These improvements generated financial returns that helped justify continuing investment.

The investment became essential when TikTok arrived

By the time TikTok challenged Facebook's social graph, Meta had accumulated almost a decade of serious machine-learning experience.

That investment became strategically important.

Reels required systems capable of ranking entertainment from creators users had never encountered.

The task differed from ranking updates produced by established friends.

Meta needed to understand the qualities of individual videos, predict likely interest and continually improve recommendations based on engagement.

AI research provided part of the foundation for that transition.

The hosts suggest Meta might have struggled to respond effectively to TikTok without its earlier investment.

This illustrates a broader characteristic of Zuckerberg's strategy.

Some investments are justified by immediate commercial value while simultaneously creating capabilities whose importance becomes apparent only after markets change.

Llama was designed to reduce dependence on proprietary AI suppliers

Meta's AI ambitions expanded with the rise of large language models.

The company developed Llama, a family of AI models made broadly available to researchers and developers, alongside Meta AI, its consumer-facing assistant.

The two serve different purposes.

Llama provides underlying model capabilities that other software can incorporate. Meta AI uses AI technology to deliver assistance directly to users through Meta's applications and other experiences.

Zuckerberg's motivation involved a familiar concern: avoiding dependence on a technology platform controlled by another company.

If Meta relied entirely on proprietary systems from OpenAI, Google or Anthropic, those suppliers could influence its technology costs, product capabilities and access to important future innovations.

Developing Llama gave Meta greater independence.

Broad distribution also encouraged outside developers to adopt its technology, expanding the ecosystem around Meta's models.

Openly available models do not necessarily mean unrestricted open source

An important distinction emerges from independent research.

Meta has described Llama as part of its open approach to AI. However, the terms attached to some Llama releases have not met conventional open-source standards.

For example, the Llama 3.1 Community License contained additional licensing requirements for organizations exceeding 700 million monthly active users, alongside other conditions.

In February 2025, the Open Source Initiative stated that Llama 3.x did not meet its Open Source Definition.

This does not negate the practical value of distributing model weights. It establishes that access and licensing freedom are different things.

Meta can encourage widespread model adoption while preserving control over certain commercial uses.

That arrangement fits its broader strategy of making complementary technologies widely accessible without surrendering every commercial advantage.

The strategy resembles Meta's open infrastructure investments

Meta previously developed the Open Compute Project, encouraging more standardized hardware and data-center designs.

Instead of depending entirely on proprietary infrastructure suppliers, the company could work with manufacturers supporting shared technical specifications.

The commercial logic was straightforward.

If Meta could reduce infrastructure costs, it could improve the profitability of its applications without needing to sell computing services itself.

A similar argument applies to AI.

More widely available model technology can reduce reliance on expensive proprietary suppliers while allowing Meta to concentrate on consumer products, advertising and distribution.

The strategy requires substantial financial resources.

Most startups cannot afford to develop competitive AI systems merely to reduce their dependence on vendors.

Meta can finance the effort through an existing advertising business that already benefits from machine learning.

AI remained central in 2026

By October 2026, Meta had moved well beyond the research ambitions discussed in the 2024 interview.

In its third-quarter 2026 financial announcement, Zuckerberg described the launch of Muse and Muse Wonder as significant milestones in the company's development of frontier AI models.

Meta also reported that AI investment was producing incremental advertising revenue through better recommendations, stronger engagement and more efficient monetization.

The company forecast $135–145 billion in capital expenditures during 2026, including finance-lease principal payments, reflecting the growing expense of infrastructure investment.

That provides an important distinction between the original interview and later evidence.

In 2024, Meta's AI spending was often discussed in terms of future technological independence.

By 2026, Meta was explicitly connecting larger AI investments to present advertising performance while pursuing more ambitious AI products.

The commercial return from individual future models remains uncertain, but AI's value to Meta's existing business is already established.

12. Reality Labs: A Multibillion-Dollar Bet on Escaping Apple and Google

Oculus gave Facebook an entry into immersive computing

Facebook began exploring augmented and virtual reality around the same period it established FAIR.

The ambition was not simply to manufacture entertainment devices.

Zuckerberg wanted Meta to possess greater control over whatever computing platform followed smartphones.

The company had already experienced the limitations of being dependent on Apple's and Google's operating systems.

In February 2014, Facebook announced the $19 billion WhatsApp acquisition.

Only about a month later, it announced another major transaction: acquiring Oculus for approximately $2 billion.

Oculus provided virtual-reality expertise, hardware development capabilities and an early foundation for a potential new computing environment.

Meta subsequently expanded these efforts through Reality Labs, Quest headsets, smart-glasses projects and augmented-reality research.

The hosts compare Reality Labs with FAIR.

Both began as concentrated investments in technologies Zuckerberg believed could become strategically important, even though their eventual commercial applications remained uncertain.

Orion suggested a different future from bulky VR headsets

The Acquired hosts had tested Meta's Orion augmented-reality prototype before recording their episode.

They were particularly impressed by how close it brought immersive computing to an ordinary glasses-like form.

Earlier virtual-reality devices generally required bulky headsets that isolated users from their surroundings.

Orion demonstrated a different possibility: digital information and interactive content presented within a person's view of the real world.

The system used glasses together with external computing components.

Meta publicly introduced Orion in September 2024, explaining that the device was intended for internal development and selected external testers rather than general consumer sale.

The company's official Orion announcement identified improvements in display quality, smaller hardware and manufacturing affordability as continuing priorities.

The hosts considered glasses a plausible next computing platform after mainframes, personal computers and smartphones.

But an impressive prototype does not establish that Meta will control the eventual market.

Even if augmented-reality glasses become widely adopted, another manufacturer could develop the preferred hardware or operating environment.

The cost of Reality Labs was enormous

By the time of the 2024 interview, Reality Labs had accumulated approximately $60 billion in operating losses over the period the hosts examined.

The financial question was whether the eventual hardware and software business could become profitable enough to recover that investment.

The hosts developed an illustrative scenario using the iPhone's extraordinary commercial success as a benchmark.

Even assuming a hypothetical Meta product could begin following iPhone-like growth economics in 2024, their model did not reach cumulative break-even until approximately 2035.

That was not a forecast.

It was an optimistic comparison designed to demonstrate how difficult it would be to justify the historical spending through direct hardware and services profits alone.

The actual financial position became more demanding after the interview.

Meta reported Reality Labs operating losses of approximately:

These figures are drawn from Meta's 2025 annual report filed with the SEC.

The company subsequently forecast Reality Labs losses of approximately $18–19 billion for 2026 and indicated that losses could decline further in 2027 as investment shifted toward AI.

The figures show that Reality Labs remained a substantial financial commitment long after the original Oculus acquisition.

The investment is also a form of insurance

Direct profits from hardware are only part of Meta's reasoning.

The company also wants to avoid becoming permanently dependent on operating systems that can restrict its products or advertising capabilities.

Apple's tracking changes demonstrated how much value a platform owner could influence through policy decisions.

The hosts therefore describe Reality Labs partly as a hedge.

If Meta eventually controls an important computing platform, it could determine more of the rules governing applications, distribution, commerce, payments and technical interfaces.

Zuckerberg reportedly told the hosts that Meta believed its core media products could be twice as profitable on a platform the company controlled.

That estimate was Meta's strategic assessment, not an independently established result.

The platform argument also explains why Zuckerberg remained interested in third-party developers.

Facebook previously attempted to build a software ecosystem inside a website. A successful hardware platform could potentially provide greater control over the underlying environment.

Reality Labs therefore combines three possibilities: a profitable device business, a broader software and developer economy, and reduced exposure to outside platform restrictions.

Each is commercially significant. None guarantees recovery of the investment.

13. The Scale and Economics of Meta’s Advertising Empire

Meta reached almost everyone it reasonably could

By the end of 2023, Meta reported approximately 3.98 billion monthly active people across its family of applications.

Facebook alone had approximately 3.07 billion monthly users and 2.11 billion daily users.

Instagram and WhatsApp each had audiences numbering roughly two billion monthly users.

Those figures informed the hosts' comparison between Meta's reach and historical empires.

They estimated that the Roman Empire governed at most approximately 40% of the world's population at its peak, while the British Empire governed around 23%.

Meta's applications reached almost half the world's population.

The comparison illustrates scale, but it has an obvious limitation: using an application is not equivalent to being governed by an empire.

It does establish that Meta operates communications infrastructure involving an extraordinary proportion of humanity.

The company's own figures confirm the historical scale. Its fourth-quarter 2023 earnings report recorded 3.98 billion monthly active people and 3.19 billion daily active people.

Meta subsequently concentrated public reporting on daily active people across the family of applications rather than publishing every earlier Facebook-specific audience metric.

These totals are estimates. Because users can maintain multiple accounts across different applications, Meta employs statistical methods to estimate unique people rather than simply adding account counts.

Existing internet access increasingly limited user growth

The hosts estimated that around 5.4 billion people were connected to the internet worldwide in the period discussed.

After accounting for China's substantial internet population, where Meta's primary applications are restricted, they estimated that approximately 450 million internet users outside China were not monthly users of any Meta application.

The precise number is uncertain because it combines several estimated populations and usage measures.

Nevertheless, the underlying conclusion was significant.

Meta had approached saturation among internet-connected populations in the markets it served.

Its future revenue growth could not depend indefinitely on adding new users at historical rates.

The company would need to earn more from existing audiences, improve engagement, introduce additional services or reach people who were not yet connected.

Facebook Marketplace demonstrated that the original Facebook product could also attract practical use beyond social posting.

WhatsApp continued developing in markets such as the United States, where traditional messaging and iMessage had historically remained prominent.

Revenue per user became increasingly important

Advertising efficiency helped Meta grow even as audience growth became more difficult.

The hosts estimated annual revenue per user of approximately $227 in the United States and Canada, compared with around $44 globally.

The difference reflected variation in advertiser demand and the commercial value of reaching consumers across markets.

A billion users are not equally valuable to advertisers in every country.

Meta's ability to generate revenue therefore depends on both the number of people using its applications and the sophistication of its advertising systems.

By 2023, the company generated approximately $134.9 billion in annual revenue and $46.8 billion in operating income.

Its operating margin was roughly 35%.

Capital expenditures reached approximately $28 billion, much of which supported computing infrastructure.

Meta operated data centers at a scale comparable with major cloud providers, but primarily used that infrastructure to support its own applications rather than selling general-purpose cloud services.

This allowed infrastructure engineers to optimize systems for Meta's particular workloads.

Meta became more profitable while spending heavily on the future

The financial position continued developing after the original interview.

According to Meta's 2025 annual report, total revenue reached approximately $200.97 billion.

Operating income was approximately $83.28 billion.

Advertising accounted for approximately $196.18 billion in revenue within its Family of Apps segment.

Reality Labs, by comparison, generated only approximately $2.21 billion in revenue while reporting a $19.19 billion operating loss.

This contrast explains much of Meta's financial model.

Its established advertising applications generate substantial profits. Management uses part of those profits to finance research, infrastructure and products whose future economics remain uncertain.

By the third quarter of 2026, Meta reported $64.60 billion in quarterly revenue, up 26% year over year.

Daily active people across its applications averaged 3.64 billion in September.

Meta also reported approximately $82.45 billion in cash, cash equivalents and marketable securities at September 30, 2026.

Its capital expenditures for that quarter, including finance-lease principal payments, reached approximately $34.91 billion.

The company's financial capacity therefore remained substantial, but so did the expense of maintaining technological leadership.

14. Meta’s Engineering Culture and Leadership Explain Its Adaptability

Zuckerberg built an organization capable of changing direction

Many technology companies successfully develop one important product but struggle when the underlying market changes.

Meta repeatedly managed major transitions.

It moved from university directories to News Feed, from desktop advertisements to mobile advertising, from public posts toward messaging, and from social-graph-based discovery toward AI-powered entertainment.

The hosts attribute much of this adaptability to Zuckerberg's leadership.

Former colleagues described him as competitive, technically capable, willing to learn quickly and open to changing his position when presented with persuasive evidence.

He was also unusually interested in both product design and the computing architecture required to deliver an experience at scale.

That combination mattered because decisions about software design, infrastructure, distribution and business economics often affected one another.

Zuckerberg had also admired Microsoft's iterative product philosophy.

Instead of requiring perfection before release, Facebook frequently launched early versions, measured performance and improved them through repeated development cycles.

Facebook developed specialized infrastructure to support speed

Facebook encouraged engineers to release software rapidly.

Its internal systems allowed teams to manage large codebases, conduct controlled experiments and identify failures before they affected the entire audience.

Engineers sometimes shipped changes during their first day with the company.

The company developed specialized technologies to address limitations in its original software architecture.

HipHop for PHP improved the performance of software written in PHP without forcing developers to abandon the language.

Hack introduced stronger typing and development safeguards while retaining familiarity with PHP.

TAO helped Facebook handle large-scale social-graph information efficiently.

Facebook also created or contributed important technologies including Cassandra, React and React Native.

These investments helped preserve development speed while improving performance, reliability and efficiency.

Meta's infrastructure was designed primarily for its own applications, allowing product and engineering teams to coordinate closely.

That focus offered advantages over infrastructure intended to accommodate every possible external customer's requirements.

Long-serving executives became an organizational advantage

Zuckerberg also maintained a substantial group of long-serving senior executives.

Chris Cox became chief product officer. Javier Oliván became COO. Andrew Bosworth became a central leader in hardware and technology. Alex Schultz led marketing, while Naomi Gleit continued into senior management.

The team had accumulated experience across Facebook's major product and business transitions.

Several important former employees established other technology companies.

Adam D'Angelo founded Quora. Dustin Moskovitz founded Asana. Sean Parker later influenced Spotify's development. Bret Taylor became an important technology executive after serving as Facebook's CTO.

The hosts suggest that Meta's organizational continuity made it easier to redeploy resources, communicate decisions and implement major changes.

This does not mean Facebook's culture was universally successful.

Its rapid-development habits contributed to privacy controversies, product failures and decisions that underestimated public objections.

But the same organizational systems also allowed the company to recover from serious business disruptions.

15. Facebook Failed Repeatedly While Building Its Successful Products

A long history of abandoned experiments

Meta's success can create the misleading impression that Zuckerberg consistently selected winning products.

The company's history contains numerous failures.

It attempted to enter local deals, payments, location sharing, video entertainment, independent messaging and social hardware.

Facebook Deals competed with businesses such as Groupon, while Facebook Places responded to location-based services such as Foursquare.

Facebook Gifts attempted to develop social commerce. Facebook Credits supported virtual payments.

Neither became a dominant permanent business.

The company also experimented with standalone applications including Poke, Slingshot, Paper, Rooms, Riff, Notify, Hello, Lifestage, Moments and Lasso.

Some were developed to counter new competitors. Others tested emerging communication behaviors or specialized use cases.

Products such as Facebook Watch and IGTV attempted to expand Meta's role in video entertainment.

Facebook Live responded to emerging live-streaming services, although it did not become the independent commercial platform originally envisioned.

Acquisitions including Beluga and TBH also contributed to experimentation without becoming enduring independent businesses.

Hardware, cryptocurrency and internet-access experiments

Facebook pursued products outside conventional social software.

Portal and Portal TV attempted to introduce social communication through dedicated home hardware.

A solar-powered drone project explored ways to provide internet access to underserved populations.

The company also proposed Libra, a cryptocurrency and payments network involving outside corporate partners.

These initiatives did not achieve their original visions.

Facebook even considered expanding messaging into an email alternative, offering communication through email-style addresses before concentrating on other messaging products.

The unsuccessful projects mattered because they illustrate Meta's willingness to enter adjacent industries.

Some ambitions were overly expensive or poorly timed. Others produced knowledge, talent or technology that could be used elsewhere.

The company generally redirected resources when products failed to demonstrate sufficient value.

The hosts distinguish these failures from major bets such as AI research and augmented reality, where Zuckerberg maintained investment over much longer periods.

Meta's biggest advantage was not perfect prediction

The company's history shows a pattern of experimentation rather than consistent foresight.

It copied some successful ideas, acquired others, created important original innovations and abandoned many concepts that did not work.

News Feed and photo tagging were especially significant original product developments.

Instagram, WhatsApp and Snapchat's Stories format demonstrated different approaches to acquiring or adapting successful products.

Its early developer platform was ambitious but ultimately weakened by the smartphone operating systems.

Meta's repeated competitive responses involved different combinations of engineering, acquisition, distribution and financing.

The hosts identify adaptability as a central characteristic.

Their description is that Meta moves like water: it changes its products and resource allocation as the competitive environment changes.

That ability has not eliminated mistakes. It has helped the company survive several of them.

16. Meta’s Regulatory and Financial Risks Continued Growing After 2024

Teen mental health became a major financial issue

The original interview identified teenage mental health as a potentially serious future challenge.

Meta faced growing concern about whether social-media products could contribute to harmful experiences among young people.

These concerns extended beyond the immediate effects of advertising or content moderation.

They raised questions about product design, recommendation systems, age verification, parental controls and the responsibility of technology companies toward younger users.

By 2026, the issue had developed into major litigation and financial exposure.

In August 2026, Meta announced an agreement with 52 US attorneys general concerning litigation related to teen mental health and social media.

The arrangement included potential payments of up to approximately $18.59 billion.

According to Meta's announcement of the agreement, payments were structured over ten years, with a portion conditional on other platforms adopting specified safety measures and providing matching funds.

The conditions concerned YouTube and TikTok adopting measures including daily usage limits, night-mode protections and age-assurance systems.

Meta's third-quarter 2026 financial statements included a $9.89 billion charge related to the agreement.

The company reported quarterly operating income of approximately $12.99 billion. Without that charge, it stated operating income would have been approximately $22.88 billion.

This was a material development beyond the original interview.

Teen safety had moved from a possible long-term regulatory concern to a financial obligation measured in billions of dollars.

The agreement also reflected Meta's attempt to encourage wider industry adoption of specific protections.

Whether those measures ultimately resolve the underlying health and safety concerns remains a separate question.

Privacy, antitrust and platform control remain connected

Meta's major legal issues concern different subjects, but they intersect with its business model.

Privacy regulation can limit data collection and advertising targeting.

Antitrust enforcement can challenge acquisitions, restrictions imposed on developers and market concentration.

Youth-safety regulation can require product changes that affect engagement and operating costs.

Meanwhile, Apple and Google retain influence over application distribution and technical capabilities.

The company's responses also intersect.

Investment in AI can improve advertising without relying as heavily on restricted tracking methods.

Investment in Reality Labs may eventually reduce dependence on outside operating systems.

Acquisitions can add capabilities but also create legal scrutiny.

Growth systems can increase engagement while raising questions about user well-being.

Meta's economic model therefore involves more than generating advertising revenue. It requires managing relationships among users, advertisers, developers, regulators and technology-platform owners.

Saturation increases the importance of earning more from existing users

Meta has fewer opportunities to repeat the extraordinary audience growth of its early years.

By 2024, its applications already reached billions of people.

That places greater importance on increasing engagement, improving advertising performance and introducing services capable of producing additional revenue from established audiences.

AI recommendations and advertising tools address that commercial need directly.

Messaging, business communication and commerce provide other possible opportunities.

Reality Labs and future hardware platforms represent more distant possibilities.

The scale of Meta's existing business makes these investments possible. It also makes replacing the core advertising model especially difficult.

A new product must either strengthen an already enormous company or eventually become substantial enough to justify major changes to its operations.

17. What Meta’s History Establishes About Its Future

The most revealing feature of Meta's history is not that Facebook grew from a Harvard project into a global network.

It is how often the company changed what that network was used for.

Facebook began as a directory of real student relationships. News Feed transformed it into a continuous information service. Photos and Events reinforced engagement. The developer platform made it an entertainment destination.

Advertising converted that attention into a profitable business.

The smartphone revolution almost destroyed that business, forcing Zuckerberg and Sandberg to rebuild advertising around mobile News Feed.

Instagram and WhatsApp expanded Meta into photography and private messaging while bringing serious questions about competition.

Cambridge Analytica exposed the consequences of permissive data practices and weak control over third parties.

TikTok weakened the importance of traditional social-graph advantages, forcing Meta to invest more heavily in algorithmic entertainment.

Apple's tracking restrictions demonstrated that access to users did not guarantee control over advertising technology.

AI investment helped Meta respond, while Reality Labs represented a much more expensive attempt to influence the next computing platform itself.

All these transitions were financed by increasingly profitable established applications.

They also depended on a governance structure that allowed Zuckerberg to continue making long-term decisions despite repeated market crises.

There is a tension within that history.

The same control that allowed Facebook to survive technological disruption also allowed controversial practices, unsuccessful products and enormous speculative investments to continue without ordinary shareholder intervention.

Meta's 2025 revenue reached approximately $201 billion, while Reality Labs lost more than $19 billion. In 2026, the company was forecasting extraordinary infrastructure expenditure while absorbing a multibillion-dollar legal charge related to youth safety.

The financial scale makes both kinds of decisions possible.

Meta's lasting strengths are identifiable: deeply established networks, enormous advertiser demand, proprietary recommendations, computing infrastructure, experienced executives and an organization capable of repeatedly changing direction.

Its vulnerabilities are equally concrete: dependence on advertising, regulation, social trust, competition for attention, outside operating systems and uncertain returns from expensive future technologies.

The evidence does not establish that Zuckerberg will successfully control the next generation of computing. Nor does it suggest that Meta's earlier success was simply the inevitable consequence of owning a popular social network.

It shows that the company's position was built through a long sequence of decisions involving product design, acquisitions, financial risk, organizational structure and technological adaptation.

Facebook became Meta because Zuckerberg repeatedly chose not to preserve Facebook in its original form.

Whether that strategy continues to succeed will depend on the company's ability to make its next transitions without repeating the costs of its earlier mistakes.