We are in a new era of geopolitical rivalry. It is one that has engulfed world politics since the Russian invasion of Ukraine in 2022, and it is referred to increasingly as a second Cold War. Once again, the world is divided into an Eastern bloc and a Western bloc. They spy intensively on each other, quarrel over trade policy, and face off in the UN Security Council. They are expanding their nuclear arsenals, and they are planning dramatic feats of space exploration, lunar and even inter-planetary colonisation, even as they attempt to contain each other’s economic and technological advancement. They build competing alliances, with Nato acquiring a counterpart in the Shanghai Cooperation Organisation, and they compete for the loyalty of larger states, such as India and Brazil, that are yet to unambiguously take a side. Elsewhere, the indirect antagonists engage in proxy wars (Ukraine being an example) and coups in Africa, while scrambling to secure resources of critical industrial minerals — lithium, cobalt, uranium. In some ways, it is as if the original Cold War never went away. 

Yet there is at least one significant difference between Cold Wars 1.0 and 2.0. In Cold War 1.0, ideological dispute between Left and Right mapped directly onto geopolitical conflict. The First Cold War was a trial of state strength, measured in terms of military might and the extent of a country’s diplomatic influence. But it was also a competition between social systems, political visions and rival ideologies: the capitalist markets of the West versus the socialist command economies of the East. It was a trial of whether competitive, multi-party liberalism could extract more political legitimacy from its citizens when compared to communist single-party states. The trial, then, was about more than innovation in weapons systems; it also involved questions of individual rights and state legitimacy.

By contrast, the ideological fissures of today’s cold war do not clearly map onto these international rivalries. Instead, they yield a much more complex picture of political tension and geopolitical rivalry. It is true that China is still ruled by a Communist Party that won power back in 1948, but China is not the champion of Leftism in the way the USSR was. There is now a stock market in Beijing — China’s third stock market, after Shanghai and Shenzhen — and, by some reckonings, China has more billionaires than the US itself. Many of these billionaires made their fortunes after China began to enmesh its economy with the world’s. The USSR was part of Comecon, the autarkic trading regime of the Eastern bloc that kept its members sequestered from the wider world economy, but the Chinese and US economies are deeply intertwined. This much is evident from the complexity of the tariff negotiations currently underway between the two countries’ trade negotiators. 

At the same time, the two powers’ navies menace each other in the Pacific. The tensions between the countries are real, but so is their inter-dependence. Each side represents one of the two wings of global capitalism, with industrial capital concentrated in China and finance capital in the US. While Donald Trump has been vocal in accusing China of stealing American jobs and exporting poisonous fentanyl to US consumers, he and his administration have been relatively muted in their political criticism of the Chinese regime. This makes sense when one considers that China, like the US, is capitalist, and that China, by the same token, offers little ideological challenge to US global leadership. Ironically, as China’s industrial might has grown, its ideological influence has diminished. Red China was never ideologically more powerful than in the heyday of Maoism, when China was much weaker. Back then, Red China sought to inspire a global guerrilla war among the Third World’s legions of poor peasants, also claiming the allegiance of student radicals on Western campuses. (Impressed by China, the Black Panthers notoriously sold copies of Mao’s Little Red Book in order to raise money for guns.)

Today, China’s infrastructural investments are helping those countries of the Global South convert their poor rural folk into urban city-dwellers. But former Western Maoists such as Bernard-Henri Lévy have become neo-conservatives, obsessed with bloody crusades to spread democracy to the Middle East, while China quietly consolidated its rise. 

In Cold War 1.0, Maoist ideological excess compensated for China’s relative weakness when compared to the military might of Soviet Russia. Today the roles are reversed: it is Russia that now occupies the relatively weaker position in the new Sino-Russian alliance. Predictably enough, it is Russian leader Vladimir Putin, more than Chinese leader Xi Jinping who has put most effort in trying to cultivate his own state as an alternative ideological pole in the international system. Putin rails against Western hegemony and has tried to style himself as a champion of global conservatism and family values against the spread of transnational Western liberalism. Yet Putin’s ideological reach is even more limited than that of Chairman Mao. His posture is fundamentally reactive and defensive, and his cultural defence of the Orthodox Church and “Russian world” will, by its very nature, have limited reach beyond Russia itself. While Trump is slowly tightening the noose of anti-Russian sanctions in an effort to induce a ceasefire in Ukraine, it is unsurprising that the White House has expended so little effort in criticising Putin’s vision of the crabbed, distempered Russian bailiwick. 

So where, if not between China and the US, or the US and Russia, can we find the key ideological battle line of Cold War 2.0? The line does not run so much between East and West as within the West itself: between the US and Europe. That the US is fighting an ideological cold war with Europe is most obvious in the diplomacy of Trump’s vice-president, J.D. Vance. Vance unleashed the opening volley in this ideological attack in his extraordinary speech delivered at the Munich Security Conference in February, when he calmly and methodically took apart the dismal record of European states on individual liberty and democratic rights. He criticised EU interference in the Romanian presidential elections of the previous year and even named the British anti-abortion activist Adam Smith-Connor, arrested in 2024 for silently praying outside an abortion clinic. In doing so, Vance adopted the same manner and tone that American leaders adopted when speaking out about the squashing of popular opposition in the Eastern bloc and the maltreatment of Soviet dissidents such as the physicist Andrei Sakharov or writer Aleksandr Solzhenitsyn. Following his speech, Vance met with Alternativ für Deutschland (AfD) leader Alice Weidel — a figure of obloquy for the European liberal orthodoxy. On his current holiday in the Cotswolds, Vance has not only once-again criticised the British record on free speech, he is also meeting Robert Jenrick — the dissident rival to the current Tory leader Kemi Badenoch.

When was the last time that a Chinese or Russian dissenter achieved the renown of Smith-Connor, by virtue of being cited in a speech by a major US political figure? The barbs do not come from Vance alone. Marco Rubio, Trump’s Secretary of State, has criticised the German state’s suppression of the national-populist AfD party as “tyranny in disguise”. EU leaders rallied together against Vance’s assault, with former German Chancellor Olaf Scholz defending continental restrictions on free speech. The German foreign office even took to X to criticise Rubio for interfering in Germany’s internal affairs. This is to say nothing of the years of bitter invective that have been hurled by European politicians and commentators against Trump ever since 2016. They have denounced him as a pro-Russian stooge, a demagogue and a fascist — only to downplay such remarks years later. Even Trump, who has mostly left the international ideological battles to Vance, embarrassed UK prime minister Keir Starmer on his recent private visit to Scotland. The mere mention of free speech by Trump was enough to induce gurning, tortured denials from Starmer.

There are some important differences between the former cold war and its new pan-Atlantic counterpart. When Western leaders denounced communism in the first cold war, they did so by arguing that the Soviets’ centralised economy, and appropriation of private property, was necessarily a totalitarian enterprise. Vance, by contrast, takes the Europeans to task for falling foul of their own ideals rather than realising some larger sinister purpose. More recently, he has also accused Europe of engaging in civilisational suicide through its unwillingness to halt mass migration. The historically unprecedented inflows represent not only a divergence from the legacy of the Western civilisation shared by the US and Europe, but a flouting of the elites’ democratic compact with their people. Repeatedly, voters have indicated their exasperation with the scale of mass migration; repeatedly, they have been ignored.

Vance’s ideological confidence, as he suggested in his speech in Munich, rests on the democratic mandate won by Trump. Barracked by Vance and his colleagues, European leaders react defensively, for they are more used to bureaucratic haggling in the committee rooms of Brussels than dealing with European voters. As with any cold war, both sides recoil from full-frontal assaults: they beat hasty retreats, seek to patch up relations and make assurances of enduring alliances. But there is no mistaking the underlying political drift.

This Euro-American ideological cold war reflects the underlying political and even economic divergence between the US and Europe. Not only does the EU cling to its outdated political model — that of a supranational trade bloc designed for the technocratic regulation of globalisation — it also remains ideologically and strategically committed to the economic policies of that bygone era, too. At the top of the EU’s agenda remain multilateral free trade deals, deindustrialisation and expensive renewables. By contrast, the Trump administration is wielding a strong democratic mandate in its effort to re-industrialise North America behind tariff barriers that penalise Europe. It is expanding its use of fossil fuels and reducing its reliance on renewable energy. 

Ideological crusades abroad, including Vance’s, always reflect domestic needs. In this case, MAGA’s desire is to assert national interests over the liberal globalism championed by the Democrats. That Vance can afford to launch an ideological cold war on Europe reflects a hard material reality: Europe simply matters much less to the US in Cold War 2.0 than it did in the first cold war. Vance can afford to be so pointed in his criticisms of the liberal, globalist EU because he knows that Europe needs America more than America needs Europe. The EU is dependent on the US for both its security and its energy, but offers little to the US in its struggle with China. 

Whereas the first cold war fused ideological and geopolitical rivalry into one great bipolar stand-off, Cold War 2.0 seems like it will be characterised by financial multipolarity (as the primacy of the dollar is gradually eroded), geopolitical bipolarity (between China and the US), and ideological tripolarity, between the populist-democratic US, the liberal-technocratic EU and a conservative Russia. As an island nation formally independent of the EU, Britain — like other middle powers — might find that this new world order is rich in opportunities. It also benefits from lower tariffs than those imposed on the EU. But seizing such opportunities also requires a willingness to duck and weave in pursuit of our national interest, and, so far, our stubbornly globalist uniparty has been unwilling to show much political dexterity. For Britain, as for many other states, seizing the national interest will require internal political renewal. 

---

• Who/What/When/Where/Why: On July 23, 2025 President Donald Trump unveiled "America's Artificial Intelligence Action Plan" via three executive orders in the U.S. to fast‑track data‑center permitting, promote exports of U.S. AI models, and prioritize working with purportedly "anti‑woke" neutral AI models to counter authoritarian influence.
• New IP warning: Cites China’s 2019 New IP proposal that would have tied identities to online actions and enabled surveillance and censorship, which was shelved but served as a wake‑up call.
• U.S. withdrawal: Argues that a decade‑long U.S. retreat from international standards-setting created a vacuum Beijing has filled.
• China's influence evidence: Notes that by 2021 China held over 30% of ITU technical‑committee leadership while the U.S. held under 8%, and that Huawei sent more engineers to an IETF meeting than Google and Apple combined.
• Authoritarian agenda: Describes China’s promotion of "cyber sovereignty" and efforts to legitimize facial‑recognition and smart‑city surveillance standards through U.N. channels.
• Belt and Road export: States that China exports surveillance technologies via the Belt and Road Initiative, often tied to loans that expand adoption of Beijing’s standards.
• Risk of splinternet: Warns that standards enabling government control could produce a divided internet with reduced freedom, privacy, and interoperability.
• Policy prescriptions: Recommends funding American engineers’ participation in standards bodies, forming allied coalitions, promoting forums like IEEE, and defending security, human‑rights, and device interoperability principles; piece authored by Ms. Tews (AEI) and Mr. Hogg (Foundation for American Innovation).

---

Your browser does not support HTML5 video.

WSJ Opinion: Trump’s Liberation Day for American AIPlay video: WSJ Opinion: Trump’s Liberation Day for American AI

Keep hovering to play

Unveiling 'America's Artificial Intelligence Action Plan' on July 23, 2025, Donald Trump signed three executive orders that will fast-track permitting for data centers, promote the export of U.S.-made AI models, and commit to only working with 'anti-woke' AI models that follow neutral ideologies.

Imagine an internet where your identity is automatically attached to everything you do—every website you visit, every click you make. That was the vision behind New IP, a proposal Chinese engineers introduced at a United Nations telecom forum in 2019. New IP would have replaced the current open internet with a government-controlled system designed for surveillance and censorship.

China shelved the proposal in the face of fierce opposition from Western governments, but it served as a wake-up call. China wants to dominate the technical standards that shape our digital future, filling a void that the U.S. has created as it has gradually withdrawn from this arena over the past decade. For too long, Washington has remained on the sidelines while Beijing has set the rules for new technologies.

President Trump aims to change that. His administration’s recent AI Action Plan and executive orders direct the State and Commerce departments to foster tech standards that reflect democratic values and resist authoritarian influence.

These specifications determine how our devices communicate, how networks operate, and how much freedom we have online. From 5G telecommunications to artificial intelligence protocols, tech standards have shaped the digital landscape for decades. Without them, innovation runs into incompatible systems.

Historically, international standards organizations operated with significant influence from Western countries. The process for developing standards was generally consensus-driven and based on merit. But as America pulled back from setting tech standards, China stepped in. Chinese officials now flood key technical meetings, submit many proposals, and vote as a bloc to promote Beijing’s agenda. Today, Chinese experts chair many of the committees setting the rules for next-generation technologies.

America has itself to blame for China’s growing power in the tech sphere. For years, U.S. policymakers and companies grew complacent, assuming that our technological dominance would naturally ensure that standards would follow our lead. Western participation in standards organizations declined after 2008 as companies cut back travel budgets during the recession. Beijing capitalized on this, updating laws in 2018 to encourage participation in international standards organizations and processes.

By 2021, China held more than 30% of the leadership positions in the United Nations’ International Telecommunication Union technical committees, while the U.S. held less than 8%. [The same year, the Chinese tech powerhouse Huawei sent more engineers than Google and Apple combined to a meeting of the Internet Engineering Task Force, which develops core internet protocols.](https://democracyandsociety.net/ds-volume-18-truth-and-information/1619-2/#:~:text=As of April 2021, Chinese,societal and human rights impacts.)

This is all part of Beijing’s strategy to embed its authoritarian values into the global technology sector. China’s “cyber sovereignty” doctrine, the principle that governments should control and monitor their digital infrastructure, is increasingly influencing global standards. Chinese firms have promoted facial recognition and “smart city” surveillance standards through U.N. channels, legitimizing technologies that enable real-time population tracking.

Meanwhile, China’s Belt and Road Initiative exports these technologies to developing countries, often tied to predatory loans. Once enough nations adopt Chinese tech, Beijing’s standards automatically gain worldwide influence.

If the technical underpinnings of AI, telecommunications and the internet enable government control, international freedom and privacy will decline. We face a potential “splinternet”—a divided digital world filled with government checkpoints that block the free exchange of information and innovation.

America doesn’t, and shouldn’t, control its tech companies the way China does. But the White House can encourage and empower American tech experts to be more engaged in shaping global technology standards. When American engineers and delegates make strong proposals built on democratic values, they can effectively challenge dangerous Chinese alternatives.

The federal government and the U.S. tech sector can work together to take steps in this direction: Fund American engineers’ involvement in standards organizations. Form coalitions with allies to share information and strategies. Promote industry-led forums such as the Institute of Electrical and Electronics Engineers, in which China’s influence is reduced by transparency. Finally, stand firm on principles of security, human rights and interoperability between devices made by different manufacturers.

Technical standards matter in the struggle between open societies and authoritarian regimes. By re-engaging in standards development, America would send a clear message: We’re back, not to dictate to the world but to ensure that free societies set the pace. It’s up to the U.S. to define tomorrow’s digital rules before China writes them for us.

Ms. Tews is a nonresident senior fellow at the American Enterprise Institute. Mr. Hogg is director of tech policy at the Foundation for American Innovation.

PHOTO: GETTY IMAGES/ISTOCKPHOTO

• Who/What/When/Where/Why: Nvidia CEO Jensen Huang drew a rock‑star reception in Beijing in July after the Trump administration allowed Nvidia to resume selling certain H20/MMNVDAMM AI chips to Chinese customers as part of U.S.–China trade and export‑control negotiations.
• U.S. export deal and concessions: The U.S. relaxed controls for some chip sales and secured an unprecedented 15% cut of China‑bound chip sales via export licenses, with chip sales also tied to China’s agreement to restart rare‑earth mineral exports to the U.S.
• H20 role and limits: The H20 is positioned as a leading inference chip for running AI models but is not powerful enough for training large AI models, and the most advanced training chips remain barred from export.
• Chinese demand and production: Major Chinese tech companies ordered at least 700,000 H20 units since controls were relaxed, prompting Nvidia to arrange additional manufacturing capacity beyond existing inventory.
• Security concerns and regulatory review: China’s cyberspace regulator summoned Nvidia over alleged “backdoor” risks in H20 (tracking, kill‑switch concerns) and has demanded a cybersecurity review before approving further shipments.
• Beijing’s push for self‑reliance: Chinese authorities have encouraged domestic chip adoption to reduce reliance on U.S. technology, and some domestic chip makers say U.S. controls accelerated that shift toward homegrown alternatives.
• Manufacturing and technology limits: U.S. export controls on advanced chip‑making equipment constrain China’s ability to produce cutting‑edge training chips at scale, leaving domestic manufacturers with less advanced tools and lower yields.
• Nvidia’s stance: Nvidia denies H20 backdoors, says the chip is commercial not military, notes China has domestic options for government use, and expresses a desire to resume sales to approved commercial customers after reviews.

---

Your browser does not support HTML5 video.

Nvidia CEO Draws Rock-Star Reception in China After Chip Sales OK’dPlay video: Nvidia CEO Draws Rock-Star Reception in China After Chip Sales OK’d

Keep hovering to play

Jensen Huang heaped praise on Chinese AI models and signed autographs a day after the Trump administration allowed Nvidia to resume selling one of its advanced chips to Chinese customers. Photo: Andy Wong/Associated Press

The Nvidia booth at an event in Beijing last month. PHOTO: SHELDON COOPER/SOPA IMAGES/ZUMA PRESS

Nvidia CEO Jensen Huang at an AI summit. PHOTO: CHIP SOMODEVILLA/GETTY IMAGES

The Global Chip Battle

U.S. export controls on Nvidia’s MMNVDAMM artificial-intelligence chips have been at the center of U.S.-China trade negotiations for months, yet it appears China is discouraging its companies from buying one of them: the H20.

Why?

Sales of the chips to Chinese companies helped get China to agree to reboot its exports to the U.S. of rare-earth minerals used in cars, electronics and other products, according to U.S. officials. In more recent days, the Trump administration extracted an unprecedented 15% cut of chip sales to China by Nvidia and AMD in exchange for granting the companies export licenses, which could result in billions of dollars for the U.S. government.

Beijing wants access to advanced Nvidia chips because its companies need them to help train state-of-the-art AI. The H20 isn’t advanced enough to train large AI models, but it is one of the best chips on the market for powering inference, the ability of AI programs to tap their training to respond to user prompts. Still, Nvidia can’t sell its most powerful chips used for AI training to China because of U.S. export controls.

Chinese authorities are also worried about becoming too reliant on U.S. technology. They have repeatedly pushed domestic chip users to support Chinese chip makers whenever possible so that—eventually—the country can be self-reliant.

The Nvidia booth at an event in Beijing last month. PHOTO: SHELDON COOPER/SOPA IMAGES/ZUMA PRESS

China’s Huawei and a few other companies already have chips that are useful for AI inference, similar to what Nvidia’s H20 chips can do. But there are other obstacles.

China’s biggest issue with making chips for AI training is that its access to advanced chip-making equipment and other technology has been blocked by U.S. export controls that seek to maintain America’s technology edge. That means even if they have good chip designs, they can hardly produce at scale. China’s best chip manufacturer uses less advanced machines, keeping yields relatively low. 

Chinese engineers say companies often get hooked on Nvidia’s software and tools, making them reluctant to switch to other vendors.

Since the Trump administration relaxed controls on the H20 chip in July, major Chinese tech companies have ordered at least 700,000 of the chips, according to people familiar with the matter. The strong demand has prompted Nvidia to change course and arrange capacity with its contract manufacturer to make new H20s to meet the demand, the people said. Nvidia initially planned to use only existing inventory to fulfill orders.

Nvidia declined to comment on its inventories or manufacturing plans. 

Some Chinese AI chip makers told officials in Beijing that U.S. export controls on chips had pushed Chinese companies to adopt homegrown alternatives faster and accelerate self-reliance. Trump’s reversal, which will allow Nvidia chips to once again flow to China, could stymie that pace of improvement, they said. Some have also expressed skepticism about the security of Nvidia products.

Nvidia CEO Jensen Huang at an AI summit. PHOTO: CHIP SOMODEVILLA/GETTY IMAGES

In late July, China’s cyberspace regulator summoned Nvidia representatives to discuss alleged “backdoor” security risks around the H20 chips, including the ability to track chip location and the possibility of a “kill switch” in the chips that would be operable under U.S. orders. Chinese officials have also raised concerns over proposed U.S. legislation seeking to add tracking capabilities for advanced chips sold abroad. 

Nvidia and others in the industry have lobbied against the bill. The White House has said it wants to study chip tracking but hasn’t weighed in on the legislation specifically.

Chinese authorities have demanded that Nvidia go through a cybersecurity review before resuming shipments of the H20, according to people familiar with the matter. The authorities have told Nvidia’s biggest Chinese customers, including Alibaba and TikTok parent ByteDance, not to buy Nvidia’s H20s until the U.S. company clears the review, the people said.

Nvidia has said there is no backdoor risk. “Cybersecurity is critically important to us,” a company spokesman told The Wall Street Journal on Tuesday, reiterating that Nvidia chips don’t have backdoors.

“The H20 is not a military product or for government infrastructure. China has ample supply of domestic chips to meet its needs. It won’t and never has relied on American chips for government operations, just like the U.S. government would not rely on chips from China,” Nvidia’s spokesman said. “We hope to resume sales to approved commercial customers soon. Banning the sale of H20 in China would only harm U.S. economic and technology leadership with zero national-security benefit.”

Write to Raffaele Huang at raffaele.huang@wsj.com and Robbie Whelan at robbie.whelan@wsj.com

The Global Chip Battle

U.S. Gave Chip Makers Billions. Now, the Hard Part.

China's Chip Goal: Cut Out the U.S.

Asia's Chip Giants Hustle to Maintain Their Edge

The New CEO Trapped in the U.S.-China Chip Battle

• Who/What/When/Where/Why: OpenAI's recent GPT‑5 release (2025) shifts focus from power users to the 700m+ rapidly growing free ChatGPT userbase on <a href="http://ChatGPT.com" rel="nofollow">ChatGPT.com</a> to enable broader monetization.
• Router centrality: GPT‑5 introduces a real‑time router that selects between models based on conversation type, complexity, tool needs, and explicit intent, continuously trained on signals like model switches and preference rates.
• Cost and performance impact: The router routes low‑compute queries to mini models to reduce costs and exposes far more free users to Chain‑of‑Thought thinking models (free exposure up ~7x first day; paying users up ~3.5x).
• Monetization thesis: Centralized routing enables classification of commercially valuable queries, paving a path toward monetizing free users via transaction take‑rates or affiliate models rather than intrusive display ads.
• Agentic purchasing capability: With routed higher‑compute allocation, ChatGPT could perform agentic actions (research, contact providers, book or checkout) for high‑value queries (e.g., legal services), generating referral/take‑rate revenue.
• Leadership and tone signals: Fidji Simo's hiring (product/monetization background) and Sam Altman's more open recent comments about monetization are presented as indicators of a push toward transactional revenue models.
• Partnerships and integrations: OpenAI already lists integrations and partners across finance, consumer, enterprise, and retail (examples: Stripe, Visa, PayPal, <a href="http://Booking.com" rel="nofollow">Booking.com</a>, Lowe’s, Salesforce, Intercom, Zendesk, Snapchat, Shopify, Instacart, Mercari), enabling checkout and tool workflows.
• Roadmap and competitive stakes: The path to a consumer SuperApp involves staged steps (affiliate fees, deeper partner plugins, agentic integrations); successful rollout threatens incumbent funnels (search/ads) and positions ChatGPT as a new purchasing channel challenging Google/Meta/Amazon.

---

To many power users (Pro and Plus), GPT5 was a disappointing release. But with closer inspection, the real release is focused on the vast majority of ChatGPT’s users, which is the 700m+ free userbase that is growing rapidly. Power users should be disappointed; this release wasn’t for them. The real consumer opportunity for OpenAI lies with the largest user base, and getting the unmonetized users who currently use ChatGPT infrequently in their day-to-day to indirectly pay is their largest opportunity.

Analysts focused on the model capabilities are missing the much larger context of network effects that ChatGPT is gaining quickly. In November of 2023, ChatGPT wasn’t even in the top 100 websites; now it is number 5. Larger than X/Twitter, Reddit, Whatsapp, Wikipedia, and quickly approaching Instagram, Facebook, Youtube and Google. On the top 10 list, every single property is much older than <a href="http://ChatGPT.com" rel="nofollow">ChatGPT.com</a>, and the sheer number of unmonetized users is staggering.

But that changes with GPT-5. OpenAI is laying the groundwork to monetize one of the largest and fastest-growing web properties in the world, and it all begins with the router.

The Router is the Release

On OpenAI’s release website, the second paragraph is about the “One United System,” specifically focused on the router. The wording is instructive.

GPT‑5 is a unified system with a smart, efficient model that answers most questions, a deeper reasoning model (GPT‑5 thinking) for harder problems, and a real‑time router that quickly decides which to use based on conversation type, complexity, tool needs, and your explicit intent (for example, if you say “think hard about this” in the prompt). The router is continuously trained on real signals, including when users switch models, preference rates for responses, and measured correctness, improving over time. Once usage limits are reached, a mini version of each model handles remaining queries. In the near future, we plan to integrate these capabilities into a single model.

The router serves multiple purposes on both the cost and performance side. On the cost side, routing users to mini versions of each model allows OpenAI to service users with lower costs. On the performance side, it will enable many users to use thinking aka CoT (Chain of Thought) reasoning for the first time. Over 99% of the of free users have yet to interact with a thinking model like o3, and for the average user, ChatGPT just got a huge upgrade. The number of free users exposed to thinking models went up 7x in the first day and the number of paying users up nearly 3.5x.

Source: OpenAI, SemiAnalysis

But the router is clearly a feature to the new service, and can likely see improvements or changes over time. It will continuously learn on preference rates and OpenAI promises it will improve over time. It just takes a single additional attribute to begin the path to monetization: the commercial value of the query.  

We believe that the Router is the groundwork for the next leg of ChatGPT’s story, and that’s monetization of free users.

The Router is Preparing for Free Monetization, Sam’s Subtle Tone Shift

Centralizing the control of the free user experience allows for many more future monetization paths. And this monetization path is one that has been hinted at subtly for a while. It all starts with OpenAI’s decision to hire Fidji Simo as CEO of Applications in May. Let’s look at her background because it’s telling.

Fidji was at Ebay from 2007 to 2011, but her defining career was primarily at Facebook. She was Vice President and Head of Facebook, and she is known for having a superpower to monetize. She was critical in rolling out videos that autoplay, improving the Facebook feed, and monetizing mobile and gaming. She might be one of the most qualified individuals alive to turn high-intent internet properties into ad products, and now she’s at the fastest-growing internet property of the last decade that is unmonetized. It’s an obvious story.

What’s more is Sam Altman himself has had a very direct tone shift in the last year.

“I will disclose as a personal bias I hate ads. I think ads were important to give the early internet a business model. I’m not totally against them, but ads plus AI are uniquely unsettling to me. I kind of think of ads as a last resort as a business model.”

But in recent interviews his tone has now shifted. There is clearly a lot of thought happening about how to best monetize free users more recently. This likely is in conjunction with Simo’s hiring.

“I am not totally against it… if you compare us to social media or web search where you can kinda tell that you are being monetized… we would hate to ever modify anything in the stream of an LLM… maybe if you click on something in there that is going to be there we’d show anyway, we’ll get a bit of transaction revenue and it’s a flat thing for everything, maybe that could work. It’s clearly possible to be a good ad driven company but there are obviously issues to it.

Compared to the previous conversation, when he was dismissive, his most recent thoughts clearly shows that Sam Altman is thinking about monetizing free users. He mentions a take rate a potential affiliate model. And as a response, the interviewer (Andrew Mayne of OpenAI) literally says, “I would love to do all my purchasing through ChatGPT because often times I feel like I am not making the most informed decisions,” as a response in the conversation. This is likely the direction OpenAI is taking.

The Router release can now understand the intent of the user’s queries, and importantly, can decide how to respond. It only takes one additional step to decide whether the query is economically monetizable or not. Today we will make our case for how ChatGPT’s monetized free end state could look like an Agentic super-app for the consumer. This is only possible because of routing.

We believe that display ads are unlikely. Perplexity has tried this, and it doesn’t seem to be going that well. Instead of inserting a paid feature into the query, we believe it’s more likely that they will pursue a take-rate based model.

Now let’s discuss, because it looks like an Agentic Assistant is a way that could align Sam Altman’s vision of AI being helpful, as well as monetizing via a transaction take rate.

Agentic Advice and Purchasing, Supply and Demand Dynamics

Let’s talk about Agentic purchasing and compare it to a search query today. Because LLMs have a core feature that Search does not, and that is scaling marginal costs. This is fundamentally different than the world search grew up in. Let’s examine “Aggregation Theory” work of Ben Thompson, because the core feature was that most technology companies had zero marginal costs to an additional user. There were some fixed overheads for running the largest search engine, but the incremental cost of another query was virtually zero. Agents and LLMs kill this concept.

Source: Arc-AGI

For the first time, the more you spend the better your result is because of CoT reasoning tokens and now marginal costs exist in software again. There is a somewhat direct relationship between more money, more compute, and a better answer. Nowhere is this clearer than in AI, in which you can spend variable costs to get a variably better answer or outcome.

So let’s apply marginal costs to a new purchasing experience. Let’s compare two queries, an information query and a commercial query.

1. A trivial information query like “Why is the sky blue?”
2. A highly commercial query, “What is the best DUI lawyer near me”

Before the router, there was no way for a query to be distinguished, and after the router, the first low-value query could be routed to a GPT5 mini model that can answer with 0 tool calls and no reasoning. This likely means serving this user is approaching the cost of a search query.

The search query on the other hand has a fixed cost. It would show a page ranking of websites, with a potential AI summary at the top. This is a fixed supply response to what could be a variably hard question. But now ChatGPT free (because of routing) can dynamically answer a harder question with a better answer, which is not how search is designed today. Below shows the value of a changing supply query to a harder question, the supply is fixed for Search.

Source: SemiAnalysis

So now let’s bring in the higher value query, the DUI Lawyer question. As you may know, this is an extremely valuable question. Today on search, this is one of the higher cost per click keywords, and it is plastered with ads. In a world of dynamic supply, ChatGPT can not only answer this question, it could realize this is a very valuable question and answer this question at the level of a human. It could throw $50 dollars of compute if there is a belief of high conversion, because that transaction is worth $1000s of dollars.

The router makes this possible. ChatGPT 5 could decide to allocate $50 to the query, create a plan, gather information about the incident, research local lawyers, consider who is likely to answer fastest, consider your budget, and then contact multiple lawyers on your behalf. It could even Agentically reach out to lawyers on behalf of the free user knowing that the conversion ratio of this query is even higher. This version of ChatGPT is highly helpful, aligns with the user’s query, and is a valuable referral to the seller of the services or goods. This wouldn’t be intrusive in the current format, and many free users would use this instantly.

This would apply to more than just services. Products that are highly likely to be purchased with agentic purchase would pay for referral fees, such as groceries, ecommerce purchases, flights and hotels. This could be a consumer SuperApp that would be an “agent” for day to day planning, purchases, and basic services.

The user wouldn’t pay via a cost of a subscription, but by transaction fees or ad take rates on purchase. The AI agent could still make the best response output but drive extremely high value business to a company almost instantly, in which the business would be willing to pay take rates on.

And what’s more you can see the glimpses of this in their model release notes. They highlighted Gmail and Google Calendar integration, as well as new tool use benchmarks for services such as Telecom, Retail, and Airlines. Imagine a world where very little customer acquisition cost is spent on ads, but rather asking a helpful AI assistant to set up the best internet plan in your neighborhood.

Source:OpenAI

And while this feels like speculation, this is already happening. Instacart added this feature to let agents checkout products in January of this year. Fidji Simo was at Instacart when this was implemented, and now she just joined OpenAI as head of product. The wheels are already in motion, this will be the future of free usage at ChatGPT.

AI labs such as Anthropic and OpenAI are even paying startups hundreds of thousands of dollars to spin up replicas of popular sites like DoorDash and Amazon to RL agents on successfully completing end-to-end transactions. It is not a question of if, but when this capability happens.

[

Jun 08, 2025

Scaling Reinforcement Learning: Environments, Reward Hacking, Agents, Scaling Data

Dylan Patel, AJ Kourabi

](https://semianalysis.com/2025/06/08/scaling-reinforcement-learning-environments-reward-hacking-agents-scaling-data/)

You can see the future. Imagine a world where you ask for new dinner recipes for the week, and ChatGPT gives you multiple options and orders the cart for you to check out. The fee would be paid on the completion of the purchase, and Search is completely cut out of the picture. Every thing that can be researched or planned in an AI app could be purchased for companies that adopt partnerships quickly.

Companies would flock quickly to this large new purchasing habit. Booking flights, purchasing items, buying food, etc. If there is a connection to a website and payment information, this could all be monetized for “free” use of the application but at transaction.

And while we are far from this future, the router is the necessary step to begin the sorting of high and low compute, and eventually commercial intent queries. None of this is possible without a single unified interface routing dynamic responses to users.

This would fulfill a vision for non-intrusive ads by Sam Altman, as well as continue to enable ChatGPT to be a trusted advisor to users for free. ChatGPT will become an agent to help users make one of the most important decisions in their day-to-day lives: buying stuff. OpenAI and Shopify are already working on a checkout integration today.

From Today to Agentic Purchasing End State

It’s no secret that we are very far from that future today. It would likely take a few steps before a true product launch, but it starts with a Router and likely many partnerships on the other side as connectors. Maybe in the beginning, ChatGPT gets affiliate fees for the currently recommended items that lead to a purchase. This would look like a typical affiliate marketing deal, and would have hard-to-measure success rates and lower take rates.

As the model becomes more agentic, it would likely need plug in to the service’s system to be able to make reservations, book flights, or schedule appointments. It would likely need heavy partnership.

Let’s examine some of the partnerships they have already made today. So far OpenAI has partnered with:

Finance Companies: Stripe, Visa, PayPal

Consumer Companies: Mattel, Booking.com, Lowe’s

Enterprise Software: Salesforce, Intercom, Zendesk

Consumer Internet: Snapchat, Shopify, Instacart, Mercari

Every company that can and will shift to a cheaper customer acquisition cost will be excited to move sooner. There will be low overhead, as less customer service, advertising, marketing and other functions will be needed as ChatGPT collapses that entire purchasing funnel into a helpful assistant.

OpenAI is Creating a Consumer SuperApp

OpenAI has clearly crossed the chasm with its web-scale presence. And every single large internet-scale property with this many users has been monetized for “free” by ads. Agentic purchasing with take rates can begin now after the implementation of the router.

We now turn our focus onto the hyperscalers, who are facing increasing competition in the consumer space, and which smaller companies are already benefiting from the shift of monetizable queries away from search to AI.

OpenAI is firmly knocking on the door of technology giants Google and Meta and even Amazon. Previous scares about AI have been focused on search query volume, not being replaced in the ad tech stack. ChatGPT can compete with dominant platforms for its place in the ecosystem, and to date this is push into purchasing is the most concrete example of OpenAI coming for advertising at large. If they were to first launch an aggressive Agentic checkout solution before Meta or Google, this would be seen as huge competitive shots at both companies.

A reminder that if we are talking about pure usage, only one company is growing users at a meaningful rate. It’s OpenAI.

Source: SemiAnalysis, SimilarWeb

In some ways by completely bypassing the top of funnel of search or pushing ads, it creates a third space for purchasing. Social media time continues to climb as human-to-human interaction is still the primary time consumption, so the research portion of consumer journeys would more heavily impact Google, but there is opportunity for smaller players. Certain smaller players are actually winning here as shown by the data below

Subscribe for full access to this article

With a SemiAnalysis subscription you’ll get access to newsletter articles and article discussions.

Model access not included – please reach out to sales@semianalysis.com for our institutional offerings.

Please verify your email address to proceed.

By subscribing, you agree to the Privacy Policy and Terms and Conditions.

• Who/What/When/Where/Why: Mark Guiducci, Vanity Fair's new editorial director, sent a memo on Tuesday announcing a refocus of the Condé Nast title to prioritize entertainment, celebrities, culture, money, politics and style.
• Editorial strategy: Vanity Fair will re-center coverage around core subjects and treat each story as a Vanity Fair story rather than by siloed vertical.
• Coverage changes: The publication will move away from news aggregation, reviews, and trade coverage.
• Website verticals: The company will gradually sunset siloed verticals, including phasing out The Hive (business, politics and technology) and eliminating labels like “Hive post” or “HWD post.”
• Hiring: Vanity Fair is hiring 13 new staffers, including a global creative director, two senior editors, three correspondents (Hollywood, Washington, Style), producers and a new social team.
• Layoffs: Several staffers have been let go, including chief critic Richard Lawson and Hollywood correspondents David Canfield and Anthony Breznican.
• Leadership appointments: Claire Howorth will become Deputy Editor overseeing stories across platforms; Daniel Kile will be VP, Global Content Strategy, with video, social, operations, audience development and events reporting to him.
• Design and events focus: Guiducci plans a magazine redesign starting with his first Hollywood Issue to influence Vanity Fair’s look across platforms and live events, from the Oscar Party to film festivals.

---

Vanity Fair’s new editorial director, Mark Guiducci, is putting his stamp on the Condé Nast title a little over a month after he took the new gig.

In a memo to staff Tuesday, Guiducci said Vanity Fair will be re-centering its coverage around entertainment, celebrities and culture, along with “money, politics and style.” As part of that, the publication will be scaling back certain areas of coverage, including “news aggregation, reviews and trade coverage,” Guiducci wrote in the memo, a copy of which was obtained by Variety.

Related Stories

[

](https://variety.com/2025/tv/news/bluey-studio-ludo-live-action-feature-allen-netflix-1236467802/)

'Bluey' Studio Ludo Shoots Live-Action Feature 'Allen' for Netflix

[

](https://variety.com/2025/tv/news/wwe-raw-netflix-top-10-ratings-1236466837/)

'Raw' Power: WWE Hits 6 Straight Months in Netflix Top 10

The move will include shutting down vertical sections of the VF website. That will include phasing out The Hive, which has focused on business, politics and technology. The pub launched The Hive in 2016 as its first mobile-focused site.

Popular on Variety

Visit Advertiser website

“We will no longer think of something as a ‘Hive post’ or a ‘HWD post,’” Guiducci wrote, referring to Vanity Fair’s Hollywood Daily vertical. “We will treat each story as a Vanity Fair story.”

Vanity Fair is hiring 13 new staffers under Guiducci’s leadership, including three new correspondents who will focus on Hollywood, Washington and style.

At the same time, several VF staffers have been let go including chief critic Richard Lawson and Hollywood correspondents David Canfield and Anthony Breznican, according to a source familiar with the matter.

Guiducci started as Vanity Fair’s global editorial director on June 30 after previously serving as creative editorial director at sister publication Vogue. Guiducci, 37, took over for Radhika Jones, who earlier this spring stepped down as the top editor at Vanity Fair.

Read the memo:

Hello everyone.

Over the past six weeks, we have talked a lot about focusing on the intersections between Vanity Fair’s core subjects — Hollywood, the arts, money, politics, and style — in modern ways, from newsletters to TikTok to new platforms that don’t yet exist.

Today, we will start working toward our new editorial strategy, gradually sunsetting our siloed vertical structure to focus on Vanity Fair as a whole. We will be moving away from news aggregation, reviews, and trade coverage. We will no longer think of something as a “Hive post” or a “HWD post.” We will treat each story as a Vanity Fair story.

This does not mean we are becoming less ambitious. The opposite is true. Transformations are not without difficulty, but I’m thrilled to say that we are hiring for a number of new positions. They include a global creative director to oversee visuals and design across all editions and platforms. They also include two senior editor positions and three new correspondents, each focusing on Hollywood, Washington, and Style, as well as producers and an entirely new social team. We will continue to add roles.

As we look ahead to the Hollywood Issue, my first, we are working on a redesign of the magazine that will influence the look of Vanity Fair on every platform, including our live events, from the Oscar Party to our presence at film festivals.

I’m also thrilled to announce that Claire Howorth will be stepping into the role of Deputy Editor, overseeing our stories everywhere that they may live. In this role, Claire will guide our day-to-day assignments, from quick turn reporting to longform narrative features to live events coverage.

I’m also happy to announce that Daniel Kile will become VP, Global Content Strategy. Video, social, operations, audience development, and events teams will report to Daniel, who will also work closely with departments across the building, from Revenue to Product and Technology. In this role, Daniel will also be collaborating with Simone Marchetti and our global HOEC as we look to grow Vanity Fair internationally.

My favorite thing about the past six weeks has been meeting each of you 1:1 and seeing firsthand the immense talent here at Vanity Fair. I also know that these weeks have been busy, often exhilarating, and I appreciate everyone who has stepped up.

Today and always, I want to encourage all of you to come to me with your ideas and questions. My door is always open. Most importantly, thank you for welcoming me to the team.

Mark

• Who/What/When/Where/Why: Perplexity, an AI startup, on Tuesday offered $34.5 billion to buy Google’s Chrome browser to challenge Google’s web-search dominance and potentially satisfy an antitrust remedy.
• Offer vs Valuation: The $34.5 billion bid is substantially higher than Perplexity’s estimated $18 billion valuation, and Perplexity said several large venture-capital investors agreed to back the transaction.
• Chrome Value Estimates: Independent estimates of Chrome’s enterprise value have ranged roughly from $20 billion to $50 billion.
• Judicial Context: U.S. District Judge Amit Mehta, after ruling Google illegally monopolized search, is weighing remedies including possibly forcing a sale of Chrome and is expected to rule on remedies this month.
• Perplexity’s Proposal Details: Perplexity told Sundar Pichai it would maintain and support the Chromium open-source project and initially keep Google as Chrome’s default search engine while allowing users to change settings.
• Google’s Response: Google has not signaled willingness to sell; CEO Sundar Pichai testified that a forced sale or mandated data-sharing could harm the business, deter investment, and create security risks.
• Market Footprint and Product Moves: Chrome has about 3.5 billion users and over 60% global browser market share; Perplexity, founded in 2022, has released its own browser called Comet to some users.
• Antitrust Case Background: The Justice Department sued Google in 2020; possible remedies under consideration include limiting default-pay agreements and requiring data-sharing, Google has proposed narrower fixes and plans to appeal, and analysts consider a forced sale unlikely though the judge has queried whether a sale could be cleaner.

---

Aravind Srinivas, co-founder and chief executive of Perplexity. Photo: David Paul Morris/Bloomberg News

Artificial-intelligence startup Perplexity on Tuesday offered to purchase Google’s Chrome browser for $34.5 billion as it works to challenge the tech giant’s web-search dominance.

Perplexity’s offer is significantly more than its own valuation, which is estimated at $18 billion. The company told The Wall Street Journal that several investors including large venture-capital funds had agreed to back the transaction in full.

Estimates of Chrome’s enterprise value vary widely but recent ones have ranged from $20 billion to $50 billion.

U.S. District Judge Amit Mehta is weighing whether to force Google to sell the browser as a means of weakening Google’s stranglehold on web search. Mehta last year ruled that Google illegally monopolized the search market and is expected to rule this month on how to restore competition.

The Perplexity offer could be an attempt to signal to the judge that there is an interested buyer, should he force a sale. In a letter to Sundar Pichai, chief executive of Google parent Alphabet, MMGOOGLMM Perplexity said its offer to buy Chrome is “designed to satisfy an antitrust remedy in highest public interest by placing Chrome with a capable, independent operator.”

Google hasn’t indicated a willingness to sell Chrome. In testimony this year, Pichai told the judge that forcing the company to sell it or share data with rivals would harm Google’s business, deter it from investing in new technology and potentially create security risks. Chrome has roughly 3.5 billion users worldwide and accounts for more than 60% of the global browser market.

Founded in 2022, San Francisco-based Perplexity recently released its own web browser, called Comet, to some of its users.

Two subsidiaries of News Corp, parent of the Journal, have sued Perplexity.

Perplexity told Pichai that, as part of the proposed acquisition, it would maintain and support Chromium, the open-source project that supports Chrome and other browsers. It also said that it would continue placing Google as the default search engine within Chrome, though users could change settings.

The Justice Department filed the antitrust case against Google in 2020. In addition to forcing a sale of Chrome, the judge is considering limiting Google’s ability to pay to be the default search engine on devices and browsers and requiring it to share data with rivals, among other things. In weighing potential remedies earlier this year, Mehta questioned how much new AI chatbots might be whittling away at the traditional search business, of which Google has 90% market share.

Google has proposed a narrower set of remedies that would modify its exclusive agreements with Apple, as well as Mozilla and Android, to allow for more competition. It has said it will appeal the judge’s ruling.

Analysts say the judge is unlikely to force the company to sell Chrome, though he gave little indication of which way he was leaning. During closing arguments earlier this year, he asked whether doing so would be “a little cleaner and a little bit more elegant” than other remedies aimed at improving search competition.

Write to Katherine Blunt at katherine.blunt@wsj.com