The Algorithm Knows You Better Than You Know Yourself
On January 18, 2025 — a Saturday night — TikTok went dark in the United States. For roughly fourteen hours, 170 million Americans lost access to what had become, by most measures, the most consequential consumer technology product of the previous half-decade. The app stores pulled it. The website returned nothing. Users who had the app already installed saw a black screen with a message directing them to a website that didn't load. It was the first time a major social platform had been forcibly shut down in a functioning democracy, and it lasted just long enough to prove an extraordinary point: the Supreme Court had upheld a law requiring ByteDance to sell TikTok or face a ban, and ByteDance — the Beijing-headquartered parent company valued at roughly $300 billion — had called America's bluff, choosing to shut down rather than sell. Then, within hours of
Donald Trump's inauguration the following day, the app flickered back to life, resurrected by executive order into a 75-day reprieve that would eventually stretch into months of negotiations, culminating in a January 2026 deal that split TikTok's U.S. operations into a new entity controlled by a majority-American board. The whole episode — the shutdown, the resurrection, the hasty restructuring — crystallized something that had been accumulating for years: TikTok is not merely a social media application. It is a geopolitical instrument, a cultural factory, an advertising engine, and an algorithmic achievement of staggering sophistication, all packaged in an interface so frictionless that the average American user watches it for 80 minutes a day. More time than Facebook and Instagram combined.
The paradox at the center of TikTok is simple to state and impossible to fully resolve. The most American-seeming app in the world — with its messy democratic creativity, its exhibitionism, its vast variety of hustlers — is not American. It was built in Beijing by a company whose relationship with the Chinese state is, at minimum, structurally ambiguous. It reshaped American culture — the music, the movies, the shopping habits, the political campaigns, the conspiracy theories — while remaining, in the eyes of the U.S. government, a national security threat. Half the country uses it. Congress wants to kill it. Both impulses are entirely rational.
By the Numbers
TikTok at Scale
~1.6BMonthly active users worldwide (2024)
170M+U.S. monthly active users
$120B+Estimated ByteDance revenue (2024)
~$23BEstimated TikTok ad revenue (2024)
80 minAvg. daily U.S. user watch time
$300BByteDance estimated valuation (2024)
150+Markets where TikTok operates
110K+ByteDance employees globally
The Information Factory
To understand TikTok, you have to understand the man who did not build it — or rather, the man who built the machine that built it. Zhang Yiming is not a social media visionary in the Zuckerberg mold. He is not particularly charismatic. He does not give TED talks or pen manifestos about connecting humanity. Born in 1983 in Longyan, a mid-tier city in Fujian province, he studied software engineering at Nankai University and, by his own account, spent his early career devouring information — reading voraciously, experimenting with side projects, building a reputation as a relentlessly analytical operator who treated human attention as an engineering problem. His central insight, developed across a series of startups in the early 2010s, was that content distribution should be determined not by social graphs — who you know, who you follow — but by algorithms that learn what you want faster than you can articulate it yourself.
In 2012, Zhang founded ByteDance in a four-bedroom apartment in Beijing. The company's first product was not a video app. It was Toutiao, a news aggregation platform that used machine learning to personalize content feeds for each individual user. Toutiao did not produce journalism. It did not employ editors. It simply watched what you read, how long you read it, what you skipped, and then served you more of what you couldn't stop consuming. It was an attention arbitrage machine, and it worked. By 2016, Toutiao had over 120 million daily active users in China. Zhang Yiming had proven his thesis: you do not need a social graph to command attention. You need a recommendation engine.
This is the DNA that would become TikTok. Not the lip-syncing. Not the dance challenges. Not the creator economy. The recommendation engine. Everything else is a content layer sitting atop a distribution machine of unprecedented precision.
We are not in the social networking business. We are in the information distribution business.
ByteDance, from its inception, was structured as what Zhang called an "app factory" — a company that could spin up, test, and iterate on consumer applications faster than any competitor, feeding them all into a shared infrastructure of machine learning models, ad tech, and content moderation systems. By 2016, ByteDance was running dozens of apps across news, entertainment, education, and humor. The company's competitive advantage was not any single product. It was the organizational architecture: centralized AI capabilities, decentralized product teams, and a culture that prized speed over polish. Zhang reportedly structured his day around eliminating "organizational entropy" — his term for the bureaucratic friction that slows decision-making. He was, in his own way, building a factory that manufactured attention.
The Musical.ly [Arbitrage](/mental-models/arbitrage)
The product that became TikTok began its life as someone else's idea. Musical.ly launched in Shanghai in 2014, founded by Alex Zhu and Luyu Yang, who had originally tried to build an educational short-video platform — think Khan Academy for mobile — and pivoted when no one cared. The pivot was elegant: lip-syncing videos set to popular songs, an interface designed for teenagers, a creation workflow so simple that a thirteen-year-old could produce something watchable in under a minute. Musical.ly crossed the Pacific in a way few Chinese apps had managed, cracking the U.S. teen market and accumulating roughly 60 million users worldwide by 2016, with the bulk of its audience in America.
Alex Zhu understood something about creative tools that Zhang Yiming, for all his algorithmic brilliance, had not yet fully internalized: that the unit of viral content in a mobile-first world is not text or photographs but short-form video, and that the way to win is not to find the best creators but to make everyone capable of creating. Musical.ly's genius was in its templates — the preloaded songs, the lip-sync formats, the duet features that let users riff on each other's content. It lowered the creative floor to zero while leaving the ceiling unbounded.
ByteDance had launched its own short-video app in China, Douyin, in September 2016. Douyin was built on ByteDance's recommendation engine and attracted 100 million users in China and Thailand within a year. But ByteDance wanted the world, and Musical.ly already had the foothold that mattered most: the American teenager. In November 2017, ByteDance acquired Musical.ly for approximately $800 million to $1 billion — a price that seemed generous at the time and would prove, in retrospect, one of the great bargains in technology history.
The merger happened in August 2018. Musical.ly's users were migrated into a new app — TikTok — that combined Musical.ly's creator tools with Douyin's recommendation algorithm. The existing Musical.ly audience provided the cold-start problem's solution: enough content and enough users to begin training the algorithm on Western tastes. And the algorithm, once fed, learned fast.
📱
From App Factory to Global Platform
Key milestones in TikTok's formation
2012Zhang Yiming founds ByteDance in a Beijing apartment; launches Toutiao news aggregation app.
2014Alex Zhu and Luyu Yang launch Musical.ly in Shanghai; it gains traction with U.S. teenagers.
2016ByteDance launches Douyin in China; reaches 100M users within a year.
2017ByteDance acquires Musical.ly for ~$800M–$1B.
2018Musical.ly merged into TikTok; global expansion begins in earnest. TikTok surpasses Facebook, Instagram, Snapchat, and YouTube in downloads.
2020COVID-19 lockdowns accelerate growth; TikTok becomes most downloaded app worldwide.
2021TikTok surpasses 1 billion monthly active users globally.
The For You Page and the Death of the Social Graph
Every social platform before TikTok was built on a single structural assumption: that the best predictor of what content you want to see is who you already know. Facebook's News Feed ranked posts from your friends. Twitter's timeline showed tweets from people you followed. Instagram's Explore page was an extension of your existing social graph. The model worked — spectacularly, in Facebook's case — but it carried an inherent limitation. Your social graph is a lagging indicator of your interests. The people you friended in college, the accounts you followed during a brief obsession with sourdough, the relatives you're too polite to unfollow — they all pollute the signal. And for new users who hadn't yet built a graph, the experience was hollow.
TikTok obliterated this assumption. The For You Page — the default screen every user sees upon opening the app — is not powered by who you follow. It is powered by what you do. Every micro-behavior is a signal: not just what you like and share, but how long you watch before swiping, whether you rewatch, whether you pause, whether you tap through to the creator's profile, whether you read the comments. The system maps each video along hundreds of content attributes — the music, the visual style, the pacing, the subject matter, the text overlays — and correlates them against every user's behavioral fingerprint. The result is a recommendation engine that can, within minutes of a new user's first session, begin serving content with uncanny relevance.
One in six American teenagers reports using TikTok 'almost constantly.'
The strategic implication was profound. Because TikTok's distribution was algorithmic rather than social, a creator with zero followers could go viral with their first video. This inverted the power dynamics of every other platform, where audience-building required months or years of patient content production and cross-promotion. On TikTok, quality — or more precisely, engagement — was the only currency that mattered. A former pizza-shop worker from Idaho could post a 16-second clip of herself playing Fortnite and wake up to 3 million views. A gummy bear reenacting an Adele concert could rack up tens of millions of impressions. The algorithm did not care about provenance. It cared about retention.
This architectural choice — content-first, graph-second — also made TikTok extraordinarily sticky. Users didn't need to curate a feed. They didn't need to follow anyone. They just opened the app and started watching. The algorithm did the rest. U.S. adults were spending, by Q2 2024, a movie's worth of time on TikTok every day — 1.3 times more minutes than on YouTube, 2.5 times more than Facebook, 3.1 times more than Instagram, according to Data.ai. The For You Page had become, in the language of behavioral design, a perfectly tuned variable-ratio reinforcement schedule: a slot machine where every pull delivers something engineered to be just interesting enough that you pull again.
The Lockdown Accelerant
TikTok's growth trajectory, already steep by late 2019, entered a different atmospheric layer when the world locked down. COVID-19 was, for the attention economy, the equivalent of a controlled experiment: billions of people confined to their homes, desperate for entertainment, with nothing to do but stare at screens. TikTok was perfectly positioned. The app's format — short, endlessly scrollable, optimized for passive consumption — was ideal for the anxious, listless quality of pandemic time. But it was the creation side that truly exploded. People who had never made a video in their lives suddenly had time, boredom, and a platform that made it trivially easy to participate.
The numbers were staggering. TikTok was the most downloaded app in the world in 2020, 2021, and 2022. Monthly active users surged by 45% between July 2020 and July 2022. By 2023, TikTok had crossed 1 billion monthly active users globally — a milestone that had taken Facebook eight years and Instagram six to reach. TikTok did it in roughly three years from its global relaunch.
The pandemic also accelerated TikTok's evolution from an entertainment platform into a cultural infrastructure. #BookTok resurrected backlist novels and sent them onto bestseller lists. #FoodTok created overnight sensations of baked feta pasta and air-fried everything. #LearnOnTikTok became, for millions, a more compelling educational medium than anything traditional institutions had devised. Political campaigns discovered that TikTok could reach young voters whom television and Facebook had lost. Hollywood — after the requisite stages of dismissal, denial, and fear — found that TikTok was indispensable for marketing films. Sony's Anyone But You, a mid-budget romantic comedy, was rescued from obscurity by a viral TikTok campaign. The BBC News theme tune went viral as Brits made light of daily coronavirus briefings.
The platform was no longer a place where teenagers did dance challenges. It was where culture was made, remixed, and distributed. As Emily Dreyfuss of Harvard's Shorenstein Center put it: TikTok became "a source of culture, a reinforcement of culture, and a cauldron where culture is created." Even if you had never opened the app, you lived in a culture that existed downstream of what happened there.
The Shou Zi Chew Problem
If Zhang Yiming is TikTok's architect — the systems thinker who designed the machine — then Shou Zi Chew is the face that was chosen to stand in front of the world and explain why the machine should be trusted. Chew was an unusual choice: Singaporean by birth, educated at University College London and Harvard Business School, a former intern at Facebook, a Goldman Sachs alum, and the former CFO of Xiaomi. He was, by design, a bridge figure — Western-credentialed enough to reassure American regulators, Asian enough to maintain credibility with ByteDance's Beijing headquarters, and young enough (he was 38 when named CEO in 2021) to project dynamism. He also faced a task that may have been structurally impossible: convincing the United States government that a Chinese-owned platform with 170 million American users posed no national security risk.
Zhang Yiming had stepped back from ByteDance's day-to-day operations in May 2021, citing a desire to focus on "long-term strategy" and learning — a move that was widely interpreted as a response to mounting regulatory pressure both in China and abroad. Liang Rubo, a co-founder and close confidant of Zhang's, became ByteDance's CEO. Chew was installed as TikTok's CEO — the public-facing operator of the international business — and immediately inherited a political firestorm.
In March 2023, Chew was hauled before the House Energy and Commerce Committee for over five hours of questioning. Lawmakers asked about data security, content moderation, Chinese government influence, and — in a moment that crystallized the hearing's combative theater — his own nationality. "I'm Singaporean," Chew responded, politely and repeatedly, to questions implying fealty to Beijing. The hearing was not, in any traditional sense, a productive exchange of information. It was a performance, staged for an audience of voters who were nervous about China, and Chew was the available representative of anxieties that transcended any single app.
But the hearing also reflected a genuine structural problem. ByteDance is headquartered in Beijing. It operates under Chinese law, which includes provisions requiring companies to cooperate with state intelligence agencies. TikTok has repeatedly stated it has not and would not share U.S. user data with the Chinese government. Chew testified as much. But the structural exposure remains: even if TikTok has never complied with a Chinese government data request, the legal framework exists to compel it. News investigations revealed that China-based ByteDance employees had, in fact, accessed non-public data about U.S. TikTok users — a revelation that undermined the company's assurances and accelerated legislative action.
Let me state this unequivocally: ByteDance is not an agent of China or any other country.
Project Texas and the Architecture of [Trust](/mental-models/trust)
TikTok's response to the national security threat was not merely rhetorical. It was infrastructural. Beginning in 2022, the company invested an estimated $1.5 billion in what it called Project Texas — a massive data migration effort designed to route all U.S. user data through Oracle's cloud infrastructure, physically separating it from ByteDance's servers in Singapore and China. The project involved building dedicated data centers, hiring a separate trust and safety team, and establishing governance structures that would, in theory, prevent Beijing-based employees from accessing American data.
The name was deliberate — Oracle's headquarters are in Austin — and the partnership with
Larry Ellison's company was strategic. Ellison, a vocal Trump ally, provided both the technical infrastructure and the political cover that TikTok needed. Oracle would host the data, audit the algorithms, and serve as the American intermediary that could reassure regulators.
Project Texas was, by any measure, an extraordinary concession. No other foreign-owned technology company had ever voluntarily submitted to this level of data sovereignty requirements. It was simultaneously a genuine security investment and a negotiating tactic — an attempt to demonstrate good faith that would, TikTok hoped, forestall the forced sale that Congress was increasingly determined to impose.
It was not enough. In April 2024, the House bundled a TikTok divest-or-ban bill into a foreign aid package, ensuring bipartisan passage. Biden signed it. The law gave ByteDance approximately nine months to sell TikTok's U.S. operations or face a ban. TikTok challenged the law in court, arguing that it violated the First Amendment rights of its 170 million American users. The case moved with unusual speed through the judicial system, reaching the Supreme Court in late 2024. On January 17, 2025, the Court upheld the law unanimously.
The following evening, TikTok went dark.
Fourteen Hours of Darkness
The brief shutdown was both a negotiating tactic and a demonstration of power. By going dark rather than quietly complying with a sell order, ByteDance showed the American public — and the incoming Trump administration — exactly what a TikTok-less America would feel like. The answer was: bad. Within hours, the political dynamics had shifted. Trump, who had once tried to ban TikTok during his first term in 2020 (via executive order that was blocked by courts), now positioned himself as the app's savior. He signed an executive order on Inauguration Day delaying enforcement by 75 days, buying time for a negotiated sale.
The 75-day reprieve stretched into months. Multiple potential buyers circled — including consortia involving former Treasury Secretary Steven Mnuchin, tech investors, and various private equity firms. The fundamental challenge was the algorithm: ByteDance's recommendation engine was the heart of TikTok's value, and China's export control laws — updated specifically to cover "personalized content recommendation algorithms" — appeared to prohibit its transfer to a foreign buyer. A TikTok without the algorithm was, functionally, an Instagram clone with worse social features.
The eventual deal, finalized in January 2026, threaded an improbable needle. A new entity — TikTok USDS Joint Venture LLC — was established under a majority-American board of seven directors, with Shou Zi Chew retaining a seat. ByteDance kept a 19.9% stake, below the threshold that would classify it as a controlling interest. The algorithm was not sold but licensed to Oracle, which would retrain and update it using only U.S. user data within Oracle's secure cloud environment. The deal meant that the core intellectual property remained ByteDance's, but its American implementation would be operated, maintained, and audited by U.S. entities.
It was, depending on your perspective, either a triumph of pragmatic diplomacy or a cosmetic restructuring that left the fundamental national security concerns unresolved. Democrats, led by Senator Ron Wyden, warned that the deal's investor group — heavy with Trump allies and foreign capital — created new risks. Senator Ed Markey called for congressional investigation, citing a "lack of transparency." The debate was not over. It had merely changed shape.
The [Attention](/mental-models/attention) Arms Race
While the geopolitical drama consumed headlines, TikTok's competitive impact on the social media industry was rearranging the landscape at a structural level. Every major platform responded to TikTok's rise by cloning its core mechanic — the algorithmic short-video feed — with varying degrees of success and desperation.
Instagram launched Reels in August 2020, initially as a feature buried within the app, and then, under increasing pressure from Meta's leadership, as an experience that effectively became Instagram. By 2024, Reels accounted for more than half of time spent on Instagram. The shift was profitable — Meta's ad revenue reached $164.5 billion in FY2024 — but it fundamentally altered the platform's identity. Instagram, which had been built on curated aesthetics and the social graph, was now a TikTok clone with better social features and more advertiser trust.
YouTube launched Shorts, which by 2023 was generating over 70 billion daily views. Snapchat introduced Spotlight. Even Netflix, a streaming service with no obvious connection to short-form video, acknowledged TikTok as a competitor for attention. The imitation was flattering and also, in a structural sense, validating. TikTok had not merely launched a product. It had established a new primitive — the algorithmically distributed short video — that was rapidly becoming the default unit of digital content consumption.
But TikTok's competitors had advantages that TikTok lacked. Meta had 3.3 billion monthly active users across its family of apps, an ad platform with decades of optimization, and no geopolitical baggage. YouTube had the world's largest library of long-form video and deep integration with Google's search and advertising infrastructure. TikTok had the algorithm and the cultural cachet, but it was fighting the ad revenue war from a standing start, against incumbents who could cross-subsidize their short-video products with existing cash flows.
The competitive question was whether TikTok's algorithmic advantage was durable or whether, once every platform had adopted the same content-first distribution model, the advantage would erode. The evidence was mixed. Instagram Reels was growing rapidly and had begun, in some demographics, to close the engagement gap. But TikTok retained a quality of cultural primacy — the sense that it was where trends originated, where content was native — that Reels and Shorts, for all their scale, had not replicated.
The Commerce Bet
TikTok's advertising business, while substantial, was not the company's long-term strategic ambition. That ambition was commerce. In China, Douyin had already demonstrated the playbook: an entertainment platform that seamlessly integrates shopping, using live-stream selling, affiliate links, and in-app checkout to convert attention directly into transactions. Douyin's e-commerce gross merchandise volume (GMV) reportedly exceeded $200 billion in 2023. The question was whether this model could be exported to the West.
TikTok Shop launched in the U.S. in September 2023, initially targeting small merchants and creator-led live-stream commerce. The early results were promising in volume if not in margin. TikTok leveraged heavy subsidies — discounted seller fees, promotional credits, algorithmic boosting for shop-linked content — to bootstrap the marketplace. The strategy was a direct copy of the Douyin playbook, which had itself borrowed from Alibaba's Taobao and Pinduoduo: subsidize the buyer, subsidize the seller, and use the platform's attention as the distribution layer that makes the economics work at scale.
E.l.f. Beauty became one of the most visible examples of a brand building its strategy on TikTok's rails. The affordable cosmetics company had debuted on TikTok in 2019 with an original song and the #eyeslipsface campaign, which became one of the platform's first branded viral moments. By 2023, E.l.f.'s Power Grip Primer alone had accumulated 70 million views on TikTok. Net sales grew 49% year-over-year to $146.5 million in Q3 FY2023, marking 16 consecutive quarters of growth. E.l.f. went from TikTok to the Super Bowl — a progression that illustrated both the platform's power as a brand-building engine and its limitations: at some point, brands that succeed on TikTok still need television to reach mass audiences.
We learned how to be a music producer, how to be a movie producer, how to be an entertainment brand. Initially, it was short-form content: how do you do all this in 15 seconds? Now, it's long-form content: how do you create these incredibly engaging pieces of branded content that don't feel like commercials?
The commerce pivot was, in strategic terms, ByteDance's attempt to follow the Alibaba and Amazon playbook: own not just the attention but the transaction. If TikTok could capture even a fraction of the U.S. e-commerce market — worth roughly $1.1 trillion in 2024 — the advertising business would become a stepping stone rather than the destination.
The Sonic [Trojan Horse](/mental-models/trojan-horse)
There is a small, almost absurdly granular detail in TikTok's competitive strategy that reveals the company's operational DNA better than any strategic memo. When TikTok videos began going viral on other platforms — Instagram, Reddit, Twitter — they carried with them a watermark and, more importantly, a sound. TikTok had invested heavily in sonic branding: a distinctive audio logo, a series of micro-jingles and sound cues that accompanied every piece of exported content. The sounds were carefully engineered to be catchy but unobtrusive, embedding TikTok's brand identity into content that users were sharing on competitors' platforms.
It was, as NPR's Planet Money described it, a Trojan Horse strategy. Every TikTok video that went viral on Instagram was, functionally, an advertisement for TikTok. The sonic branding was the mechanism by which the advertisement was smuggled past the audience's defenses — you didn't think of it as marketing, you thought of it as the sound of the internet. Instagram eventually built tools to strip TikTok watermarks from reposted videos, an implicit acknowledgment that the strategy was working.
The sonic branding campaign reflected a broader truth about ByteDance's operational culture: the company thought about distribution with a granularity and intentionality that Western platforms often lacked. Where Facebook relied on network effects and Google relied on search intent, ByteDance treated every surface of the internet as a potential acquisition channel. The watermark was not vanity. It was growth engineering.
The Brain and the Machine
The criticism of TikTok, when it wasn't geopolitical, was neurological. The app's design — the bottomless scroll, the variable reward schedule, the auto-playing videos calibrated to exploit novelty-seeking behavior — triggered genuine alarm among psychologists, pediatricians, and parents. Jonathan Haidt, the social psychologist whose 2024 book The Anxious Generation catalyzed a national conversation about social media's effect on children, called TikTok a "national security threat" — but the threat he was describing was not Chinese espionage. It was the systematic rewiring of adolescent attention.
The numbers supported the concern. Sixteen percent of American teenagers reported using TikTok "almost constantly," according to Pew Research Center. Two-thirds of American teens used the app at all. A report from Qustodio found that TikTok was simultaneously the most-used social media app for children and the app parents were most likely to block. The average session length — 80 minutes per day for U.S. users — exceeded anything other platforms had achieved, and the behavioral science behind that engagement was, by design, extraordinarily sophisticated.
TikTok's defenders argued, with some justification, that the same concerns had been leveled at every new medium — television, video games, the internet itself — and that the platform also functioned as a genuine educational tool, a creative outlet, and a community-building mechanism for marginalized groups. Both things were true. TikTok was a place where #LearnOnTikTok could flourish alongside content designed to exploit the dopamine circuits of developing brains. The platform's power was indiscriminate, and indiscriminate power is, by definition, both a tool and a weapon.
The 2026 U.S. deal included new terms of service that prohibited children under 13 from using TikTok outside of a dedicated "Under 13 Experience" — a concession to the growing political consensus that algorithmic recommendation engines should not be pointed at children without guardrails. Whether the guardrails would be enforced, or enforceable, remained an open question.
The Asymmetry of Influence
There is a structural asymmetry at the heart of TikTok's global position that is easy to state and difficult to solve. TikTok operates in over 150 markets. Douyin, its Chinese twin, operates exclusively in China. American social platforms — Facebook, Instagram, YouTube, Twitter — are banned in China. Chinese users cannot access TikTok. American users cannot access Douyin. The information flows in one direction: Chinese companies can study, influence, and monetize American attention, while American companies are locked out of the Chinese market entirely.
This asymmetry is not TikTok's fault, exactly. It is a product of China's Great Firewall and the Chinese government's long-standing policy of protecting domestic technology companies from foreign competition. But the asymmetry creates a strategic imbalance that goes beyond any individual company. ByteDance operates in a regulatory environment where the Chinese government has the legal authority to access company data and influence content moderation decisions. TikTok operates in a regulatory environment where the U.S. government cannot do the same with Chinese platforms — because there are no Chinese platforms in China for the U.S. government to regulate.
The divest-or-ban law, the Supreme Court ruling, and the eventual USDS Joint Venture restructuring were all attempts to address this asymmetry. Whether they succeeded depends on whether you believe that licensing the algorithm to Oracle, establishing a majority-American board, and limiting ByteDance's stake to 19.9% meaningfully changes the power dynamics — or whether you believe, as some lawmakers argue, that the ties remain too deep, the transparency too limited, and the structural exposure too fundamental to be resolved by corporate restructuring alone.
The $300 Billion Question
ByteDance, TikTok's parent, is one of the most valuable private companies in the history of capitalism. Its estimated valuation of approximately $300 billion places it alongside public companies like Salesforce and Coca-Cola. Its estimated revenue of over $120 billion in 2024 — generated across TikTok, Douyin, Toutiao, and a portfolio of other apps — makes it larger by revenue than Meta. Zhang Yiming, who retains a significant stake despite stepping back from operations, is consistently ranked among China's wealthiest individuals, with a net worth estimated at over $40 billion.
The company has never gone public. An IPO was reportedly planned and then shelved in 2021, amid regulatory crackdowns by the Chinese government on technology companies. The decision to remain private has given ByteDance a freedom of strategic action that public companies do not enjoy — no quarterly earnings calls, no activist investors, no obligation to explain the economics of TikTok Shop's subsidies or the cost of Project Texas. But it has also made the company's financials opaque. Estimates of TikTok's standalone economics vary widely: some analysts peg TikTok's 2024 ad revenue at roughly $23 billion, others higher, with profitability in the U.S. market still uncertain given the enormous spending on infrastructure, content moderation, creator subsidies, and the USDS restructuring.
The creator economy that TikTok catalyzed is itself a significant economic force. TikTok claims to have contributed $24.2 billion to the U.S. economy. The platform's Creator Fund, launched in 2020 with $200 million (eventually expanded to $2 billion over three years), was an explicit attempt to retain top talent who might otherwise migrate to YouTube or Instagram. But the fund's per-view payouts were widely criticized as paltry — fractions of a cent per view — and many creators found that TikTok's real value was not in direct monetization but in brand deals, audience-building, and cross-platform leverage.
The Culture That Exists Downstream
Roughly 170 million Americans use TikTok. That is half the population of the United States. The music America listens to, the movies it sees, the conspiracies it believes, the products it buys, the celebrities it creates — all of it has been shaped, inflected, and in some cases determined by what happens on a platform whose algorithmic logic was designed in Beijing and whose content is generated, overwhelmingly, in the countries where it operates.
Lil Nas X's "Old Town Road" became the longest-running number-one single in Billboard history in part because of TikTok virality. Fleetwood Mac's "Dreams" returned to the charts in 2020 after a TikTok video of a man skateboarding while drinking cranberry juice accumulated hundreds of millions of views. The "Renegade" dance — created by a Black teenager from Atlanta named Jalaiah Harmon — became one of the most replicated choreographies in internet history, raising urgent questions about credit, appropriation, and who captures the value of viral content.
TikTok did not invent virality. But it industrialized it. The platform's algorithm is a distribution machine that can take any piece of content — a dance, a recipe, a conspiracy theory, a 15-second comedy sketch — and, if the engagement signals are right, push it to tens of millions of users within hours. This power is value-neutral. It works the same for #BookTok recommendations and election misinformation, for small-business marketing and mental health pseudoscience. The algorithm does not have values. It has metrics.
The result is a cultural system of extraordinary vitality and extraordinary danger, operating at a scale and speed that no previous medium has achieved. TikTok is the most consequential media distribution platform of the early 21st century. It is also, depending on which committee hearing you attend, a threat to national security, a threat to children's mental health, a threat to competitive markets, and a threat to the epistemic foundations of democratic society. All of these things can be true simultaneously. That is the nature of a technology this powerful.
On any given night in America, a teenager opens TikTok, swipes past a video of a cat, pauses on a cooking tutorial, watches a conspiracy theory about a celebrity, laughs at a sketch, saves a song she's never heard before, and closes the app 80 minutes later having consumed more algorithmically optimized content than her parents consumed in a week of television. The algorithm has learned something new about her. She has learned nothing about it.
