Alibaba Shares Rise After Unveiling Upgraded Flagship AI Model

Alibaba Group Holding Ltd. shares rose as much as 5.4% on Monday after the company launched a preview version of its flagship Qwen3.8 Max model, describing it as second only to Anthropic PBC’s Fable 5.

The Sunday release came only days after startup Moonshot AI unveiled a powerful new offering that’s roiled markets and triggered concern in the US about China closing the gap on global leaders like Anthropic and OpenAI. Qwen3.8 Max has 2.4 trillion parameters, joining Moonshot’s Kimi K3 in the heavyweight class. With 2.8 trillion parameters, K3 rivals top offerings and Alibaba is setting similarly high expectations. (…)

Alibaba plans to make the model open-weight soon, expanding access beyond the preview release. Interest in these made-in-China artificial intelligence systems and models is so high that Moonshot was forced to pause taking on new subscriptions late on Sunday to manage overwhelming demand.

While optimism around Alibaba is growing, other contenders in China’s hotly contested AI race have suffered a drop in the wake of the new Kimi release. Rival Zhipu declined nearly 30% on Friday and added a further 14% to the losses on Monday, after being one of the star debut stocks in Hong Kong for much of this year.

Hangzhou-based Alibaba, China’s e-commerce leader and one of its biggest investors in AI, recently scored another victory after Beijing approved Apple Intelligence, the software suite for iPhones, iPads and other Apple Inc. gear, which will use Alibaba technology in the country.

Top American AI Execs Sound Alarm on Chinese Models White House is divided on how to respond to recent advances in Chinese AI, has weighed crackdown measures

Silicon Valley and Washington are debating a multibillion-dollar question: Should American companies be able to use Chinese artificial-intelligence models?

OpenAI and Anthropic executives are sounding the alarm about the rise of cheap AI, particularly powerful new models produced in China, suggesting they will lead to a “dystopian” AI future and present unacceptable security risks without regulation.

Some analysts who study the AI industry say the two companies, which are preparing for public listings in the next year, just want to eliminate the competition.

The emergence of highly capable, open autonomous AI systems—including Moonshot AI’s Kimi K3 model and Alibaba’s Qwen 3.8 Max, which were released in recent days and viewed favorably by investors and users—has turned the AI race on its head once again. (…)

The debate over the new models, which are “open weight,” allowing users to download and customize them with company data and for specific tasks, coincides with division in the Trump administration about whether to take steps to limit the use of the models in the U.S.

“One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a ‘public good’ which will ultimately be provided by the state as a kind of ‘digital public infrastructure,’” Dean Ball, OpenAI’s head of strategic futures, said in an X post Friday. (…)

He also highlighted a central challenge with the AI race: Top-tier AI companies raise billions of dollars to pay for the vast computing resources needed to continue improving AI systems. If everyone uses AI systems that people largely don’t pay for, there would be no way to finance continued frontier AI development. (…)

Chief Executive Sam Altman has previously said OpenAI’s past approach of only developing closed tools was unwise.

Other top AI executives have warned that open models present huge risks, since developers and policymakers lack control over how they are used or modified.

Anthropic CEO Dario Amodei for years has warned about the risks of powerful and open AI systems, saying in a recent Bloomberg interview that having AI models with advanced cybersecurity capabilities that are free to download could be harmful. “It’s a serious concern,” he said in June.

Use of Chinese models, which are far cheaper than U.S. counterparts, is surging at U.S. companies, prompting some investors to question the staying power of top model-makers such as Anthropic and OpenAI.

The current U.S. open-source model frontier is starting to catch up with China’s. On Wednesday, Thinking Machines Lab—led by former OpenAI technology chief Mira Murati—released its first AI model as open weight. Nvidia’s Nemotron 3 Ultra is starting to see traction, and Reflection AI, an Nvidia-backed open-model developer, has close ties to the Trump administration and plans to release its first model later this year.

The market’s faith in Anthropic and OpenAI continuing to build more capable models that push the AI frontier has been at the heart of the boom, helping to justify trillions of dollars in spending on infrastructure in the coming years. The threat that new players will vastly undercut what they can charge for advanced AI pushed down some tech and AI company stock prices last week. (…)

Sacks and others have long seen calls for AI regulation by companies such as Anthropic as efforts to use new laws and policies to stymie competitors. (…)

The CEOs of OpenAI, Anthropic and Alphabet’s DeepMind have recently signaled they support increasing government oversight of AI as models become popular, fueling some criticism that the companies are trying to stifle their competition. Alphabet is the parent company of Google. Demis Hassabis, CEO of Google’s DeepMind lab, recently suggested developers of open models be included in discussions about AI regulation. (…)

Officials who have pushed for oversight of AI have worried that open models could pose cyber and biological-weapon risks if they continue advancing and don’t have to follow the same rules as top U.S. companies such as Anthropic and OpenAI, the people said.

The Trump administration is committed to promoting America’s open-source ecosystem and strengthening its security, a White House official said.

The new focus on the issue shows how the messy policies surrounding open models are challenging CEOs trying to cut their AI bills and policymakers who don’t want the technology used to harm the U.S.

“AI is increasingly synonymous with power and the dual-use concerns are real. But for American businesses and most of the world, being able to run cheap, high-quality models in a way they can control is going to matter a lot,” said Austin Carson, CEO of SeedAI, an AI-policy nonprofit. “If you know about open source, you’d know that you can’t win by exclusion.”

• Meet the Companies Shelling Out for Top AI Models

As some companies pump the brakes on AI spending by resorting to cheaper models, others are going all-in on the most advanced AI systems—even with their hefty price tags.

So-called frontier AI models, or the most capable systems made by companies like OpenAI and Anthropic, can be expensive to use partly because they require a lot of compute and process large numbers of tokens, AI’s basic unit of measurement. But these state-of-the-art models are considered the best because they can “reason” through complex, multistep problems and are capable of supporting a variety of tasks, including powering autonomous AI agents.

The calculus often is as much a business decision as an engineering decision. If paying a premium for a frontier model means a better product or an upper hand over rivals, many companies say it’s worth it. (…)

In other words, in the race to build the next, better product, you’ll get there faster with frontier models. (…)

The cost is probably not worth it for simple queries and tasks like summarization and editing, where the difference between frontier and cheaper models is negligible, tech leaders and analysts say. Indeed, there is an ongoing debate over whether AI models are becoming a widely-available commodity.

But for complex reasoning tasks like managing AI agents, advanced coding problems and multistep research, frontier models perform better—even if by a small percentage—and that can make all the difference in outpacing the competition. (…)

Other companies say they’re choosing frontier models because they need top-of-the-line capabilities. (…)

With the cost of AI rising, more companies are using smaller, cheaper models, and open-source or open-weight models. It has also become more popular to use cheaper models for less critical tasks—allowing companies to save on token costs.

At companies like Spotify, it’s an ongoing discussion whether frontier models are worth the cost. (…)

Developers tend to love using frontier models because “they simply work better,” said Philip Walsh, an analyst focused on software engineering at market research and IT consulting firm Gartner. But most companies are trying to find ways to make sure workers use more cost-efficient models or are building AI agents that can take advantage of frontier models for planning tasks, while relegating lower-tier tasks to cheaper models, he said. (…)

Boris Cherny, the head of Anthropic’s Claude Code, said the AI lab allows customers to put spending limits in place and opt for some of Anthropic’s lower-cost models. Customers can also “tune” how much thinking a model does—essentially asking a model for less intelligence, which uses fewer tokens, he said.

“It’s just keeping costs reasonable and predictable,” Cherny said. “But I think actually the far bigger opportunity is increasing return, and I think this is what customers are saying, too. The more tokens that they use in a useful way, the more return they get.”

It is a balancing act between capability, cost and data control. More powerful closed frontier models can be best for critical tasks but most users will lean towards lower cost open (customizable) models for less critical tasks or if data or model control is paramount.

As Global Semi Research explains

(…) what Kimi K3 really proves is not that model companies have no moat, but that the model itself is not the moat.

Raw model capability is becoming commoditized very quickly. A model can top the leaderboard today and be matched by competitors a few months later. Model capability still matters. It determines whether a company can sit at the table. But it is becoming harder for model capability alone to form a durable moat.

The real long-term value lies in the flywheel formed by the model, workflow, feedback data, customer relationships, and reinvested revenue. (…)

The more important questions are: how long can model capability leadership last? After open-source diffusion, who actually captures the revenue? Where do customer relationships and task feedback accumulate? And who can turn one model release into the starting point for the next iteration and the next stage of commercialization?

In many cases, the difference between the top model and the rest of the leading pack is no longer a generational gap. It is often a difference in benchmark design, task preference, and specific use case.

A single model lead is therefore more like an asset that depreciates quickly. (…)

Model capability is the ticket to the game. But a ticket is not a moat. What matters is whether a company can keep getting the next ticket, faster, cheaper, and more reliably than others.

The problem for Anthropic and OpenAi is that this discussion happens before their IPO which would have reduced their debt with a highly priced currency.

The problem for the US government is that this financial rebalancing has not happened. These two companies are currently too big and too critical for the US to fail.

• How Companies Are Managing AI Token Spend

(…) Chief information officers told The Wall Street Journal Leadership Institute they are deploying a number of strategies, including tried-and-true techniques sharpened during the rise of cloud computing—and the need to manage ballooning cloud costs—to keep their AI costs under control.

“With AI, you’re putting the credit card in the hands of the end user. If you have no control over that, or if the end user is not educated enough, they’re going to run up that tab,” said Chris Reed, a senior director of IT finance at online travel company Priceline.

Unlike in previous tech cycles, corporate adoption of AI rests on all employees—not just developers—picking up on the technology. AI is increasingly being billed by usage, and the price of tokens, the basic unit of AI computing, has been volatile. That all translates to higher costs for AI. (…)

Adding to the cost pressure is the shift from prompt-based chatbots to always-on autonomous AI agents, which consume vastly more tokens. And with larger, more sophisticated AI models, those costs are expected to climb sharply.

“It will be orders of magnitude higher than what we spend today,” said Greg Meyers, chief digital and technology officer of Bristol-Myers Squibb, adding that he expects “exponential” costs associated with AI agents as AI usage hits an inflection point. (…)

“If you factor in what we believe is the payoff here, we believe that it’s actually a pretty positive [return on investment],” he said. (…)

Enterprises already are using more AI than ever before, with many wrangling more AI agents than they can keep track of.

Compared with asking a chatbot a question, asking an agent to complete a task can require 50 times as much computing power, according to Jim Schneider, a senior equity research analyst at Goldman Sachs. Goldman predicts that AI agents will increase AI token consumption by 24 times over the next four years, and business AI agents will increase token consumption by 55 times by 2040.

Model providers OpenAI and Anthropic have said the costs of their tokens are going down, and both have considered drastic price cuts.

Even with less expensive tokens, however, agents are consuming more of them as they interact with other agents and work over long periods of time. While model prices fell roughly 50% from December 2024 to December 2025, tokens consumed grew 4.5 times in the same window, according to research from Bain and Co. (…)

“High AI usage isn’t necessarily a good or bad thing. It depends on the business outcome that’s attached—that’s the most difficult part to quantify,” said Priceline’s Reed.

That uncertainty is pushing companies toward another tactic: paying less per task. Rather than run everything on large, expensive models, some are swapping in smaller, older or open-source models. Running those models on Qualcomm’s own hardware saves the company even more, Tinic said.

Seemantini Godbole, chief digital and information officer of Lowe’s, said the company is putting guidelines and mechanisms in place to avoid “token wastage,” including using smaller and open-source models. (…)

China weighs tighter export controls on AI models and chips Beijing consults companies on ways to stop west acquiring its advanced technologies and star start-ups

(…) MofCom talked to AI companies including Alibaba, ByteDance and Zhipu on limiting the transfer of key data for the training of their models overseas, as well as allowing their model weights to be downloaded by foreign users, the people said. China would still let overseas customers access the models and services, however. (…)

MofCom has also sought views on possible restrictions that would prevent overseas chipmakers including Qualcomm and TSMC from producing advanced semiconductors based on designs developed by Chinese companies such as Huawei, Alibaba and ByteDance, according to the people.

Potential restrictions could also be imposed on the overseas acquisition of strategic technology groups in areas such as agentic AI, the people said. This is mainly to address a loophole that Beijing believes to have led to Meta’s $2bn acquisition of Manus, a deal that was subsequently ordered to be unwound by Chinese authorities.

The new measures could be incorporated into the next revision of China’s catalogue of technologies prohibited or restricted from export, the people said, reflecting Beijing’s growing confidence that it has established a global lead in some areas of AI. (…)

Trump Imposes Additional 50% Tariffs on Certain Canadian Goods Measure applies to products including wine, though energy and parts of other sectors are exempt

President Trump imposed an additional 50% tariff on certain goods from Canada, including wine, hockey sticks and cement, the White House announced Monday.

The White House said that the tariffs were a response to the country’s “discriminatory treatment of American products.” Some sectors and goods, including energy, potash and fish or critical minerals, will be exempt from the new tariffs, the White House said.

The White House said the tariffs were “designed to offset the burden and disadvantage” from what it described as Canada’s discrimination of U.S. goods, including autos. The White House said the U.S. opposed Canadian policies that require companies to invest in auto production in Canada, rather than the U.S., as well as bans some Canadian provinces have imposed on U.S. liquor products. (…)

The new tariffs, unlike earlier rounds of levies on Canada, will apply to goods that comply with the USMCA deal, the White House said. (…)

The new tariffs come after smoke from wildfires in Canada drifted into the U.S., blanketing cities including New York, Chicago and Washington. In a Truth Social post Friday, Trump threatened to impose steeper duties on America’s northern neighbor to compensate for the smoke’s impact.

“We are holding Canada responsible for the fact that they are not properly maintaining their Forests, and Brush therein, and the United States is being unnecessarily invaded by filthy, polluted, and unhealthy air, the quality of which is dangerous, and totally unacceptable!” Trump said. (…)

Trump told reporters that Canada needed to stop the wildfires.

“I told them, I mean, you got to stop these fires from coming in and you know poisoning our air,” Trump said. “Our air has been poisoned. Have a good relationship with Mark Carney, but you know we got to stop the fires up there. If we can help them, we’ll help them. But maybe they should pay us some damages or something, or we should do some tariffs.” (…)

• Canada issues air quality warnings over US wildfire smoke after Trump tariff threat Officials say smoke from US fires has crossed border

(…) Officials in the Canadian province of British Columbia warned in an alert on Saturday evening that smoke from US fires, spurred on by thousands of lightning strikes in the north-east, was creating smoky conditions as polluted air drifted north.

“Much of this smoke is originating from fires south of the Canada-US border in Washington and Oregon,” the provincial wildfire service said.

Twenty-two large fires are burning across Washington and Oregon, according to the Northwest Interagency Coordination Center, the fire coordination agency for both states. British Columbia has air quality warnings in place for two regions, Kootenay Lake and Cranbrook, located along the south-east border with the US. (…)

Embarrassing! Though some people just don’t care being embarrassed.

Retail Resilience Defies Doomsayers

The Richmond Fed’s analysis sums up the consensus view. Beware!

The latest retail sales data suggest that both consumer spending and the retail sector remain robust. (…) retail and food services sales as measured by the Census Bureau were up 6.9 percent year over year as of May. Furthermore, a weekly same-store retail sales metric published by Redbook Research — a leading indicator for the official Census measure — increased for the sixth consecutive week, reaching 11.5 percent year over year the week ending July 4. This is its highest reading since 2022 and an indicator of further near-term strength.

(…) Figure 2 shows that the retail trade services producer price index (PPI) has risen rapidly since the pandemic, with an increase larger than the increase observed in the PPI for nontrade services. This price index — which we’ve previously covered — measures the average difference between retail businesses’ selling prices and their acquisition prices, with increases indicating a larger gap between selling and acquisition prices.

Additional data from the Census Bureau’s Quarterly Financial Report reveal that retail businesses have been achieving healthy profit margins. Figure 3 plots the retail trade profit margin, which is computed as retailers’ after-tax profits as a percentage of sales. Retail profit margins rose to 5.8 percent in the first quarter of 2026, which was the highest share seen in data (outside of the pandemic) starting in the fourth quarter of 2000.

Source: Census Bureau via Haver Analytics

These positive developments show that retail businesses and U.S. consumer spending have kicked off summer in full stride, assuaging concerns of a near-term consumer pullback.

It seems that Amazon’s moving its widely popular Prime Day sale to June 23rd-26th from July 8th-11th in 2025 has been missed by many (it is not just Amazon as many other retailers run their own sales events around Prime Day for competitive reasons and to capture the halo effect).

The Census Bureau seasonal adjustments did not catch this important change in monthly patterns. It won’t catch July’s drop as well.

The same holds for inflation data in June (red line below).

For retail sales and goods inflation data, June was Amazon Prime Day month. July will be Payback Day.

Goldman’s Jan Hatzius:

The weaker-than-expected June employment report has brought our estimate of underlying US job growth down to 73k, from 130k a month earlier. And while we would not ignore a move in the unemployment rate given its proven value as a cyclical indicator, we expect the latest drop to 4.2% to reverse in coming months because it was driven by a suspiciously large drop in labor force participation.

Moreover, other signals such as the ongoing weakness in household job market assessments as well as depressed flows both into and out of employment still suggest that the labor market remains a tad cooler than normal.

This chimes with the slowdown in our GS wage tracker to 3.4%, below the 4% pace that would be consistent with a 2% inflation target assuming a 2% productivity trend. (…)

Amidst the higher US core PCE numbers this year, it is worth noting how favorable the inflation news has been almost everywhere else.

Core inflation in the G10 ex-US—using either the traditional ex food and energy definition or trimmed-mean measures—has continued to trend down this year and now stands at 2.1%, despite the energy price surge in March and April.

This means US core PCE is an outlier to the high side, not just relative to alternative measures of underlying US inflation such as core CPI or trimmed-mean PCE but also relative to other economies with similar levels of resource utilization.

This reinforces our view that core PCE overstates true underlying inflation, in part because of mismeasurement and in part because of idiosyncratic US shocks such as tariffs.

The better inflation news has effectively extinguished whatever chance there was of a rate hike at the July 28-29 FOMC meeting. Hikes at subsequent meetings are possible but would probably require significantly higher inflation and/or lower unemployment than we expect. (…)

The rebound in energy prices probably won’t cause the ECB to hike on July 23, but it has made us more confident in our call for a second hike in September. Beyond that, however, our views diverge from market pricing as we expect the Governing Council to hold off on additional hikes (and return the deposit rate to 2% in 2027). (…)

We expect policymakers to step up their easing rhetoric in the July Politburo meeting and draw on remaining fiscal buffers quickly to stabilize investment and growth. But our full-year 2026 growth estimate has drifted down to 4.6% and the dependence on exports means that China is vulnerable to any renewed global growth shocks from the Middle East or elsewhere.

Warsh at Congress:

Facing Congress, Fed Chairman Warsh did not provide any guidance on where the Fed is heading over the near term, but he did say three important things.

First, he pledged that the Fed will bring inflation down and that he had no tolerance for high levels, asserting the Fed’s responsibility for controlling inflation through monetary policy.

Second, he outlined an agenda for task forces that will be manned by prominent academics and business leaders to reform the central bank in order to regain public trust and ensure price stability, focusing on economic evaluation, communication, economic data quality, and its policy-setting process.

Third, he endorsed the so-called Roaring 2020s narrative, revealing his enthusiasm about the AI boom as a reason to be optimistic about growth and inflation. In his judgement, “the spending on data centers, software and infrastructure should boost productivity, raise the economy’s non-inflationary growth rate and help ease inflation pressures over time, offering support to the Fed drive to price stability”. (Hubert Marleau, Palos Wealth Management)

John Mauldin adds:

“If we get policy right — and we will — the inflation surge of the last five years will be a thing of the past.” (As Peter noted, the level of government spending is a key factor in inflation, and with a budget deficit at 5.6% of GDP, it makes it tough to get back to 2% sustainably.)

The Chairman then stated he was “doubling down” on the 2% inflation target — a general concept he’s been publicly critical of in the past — and here it’s worth reminding market participants that the Federal Reserve targets “the annual change in PCE” [read: headline, not “3mo super core” or any of these other silly dovish “cherry-picked” metrics]

Though maybe most stunningly and most importantly, the Chairman stated that “unfortunately, after 63 months of prices above target,” it is “our job and my commitment to take sticky prices and un-stick them”. (…)

Warsh sees his job as getting inflation back below 2%. (…)

I think market participants will eventually come to understand, if they don’t already, that Chairman Warsh is completely independent of the Trump administration. He is a fierce proponent of Federal Reserve independence.

Escalating U.S.-Iran conflict sends oil prices higher as world braces for tightening supply

(…) Vessel crossings in the strait hit a three-week low late last week as attacks on ships intensified and the U.S. reimposed its naval blockade. Crossings fell to eight on Thursday, according to Kpler data, a maritime-intelligence firm. (…)

Mr. Bertamini said markets are currently underpricing the tightness in global supply, a view he shares with other experts who are growing increasingly concerned about the depleting government stocks of crude oil. Emergency stocks have been strategically released over the past few months to stabilize prices and make up for the loss of Gulf supply – or about one fifth of the world’s oil.

In early March, 32 member countries of the International Energy Agency agreed to release 400 million barrels of oil from their emergency reserves to address the disruptions. Recently, the IEA said that its member countries had released almost three-quarters of the planned amount.

Prior to the conflict, the U.S. strategic petroleum reserve – the world’s largest publicly known emergency stockpile of oil – had about 415 million barrels of oil stored, more than half its capacity.

But the recent conflict has pushed reserves to their lowest levels since 1983. In March, President Donald Trump announced that the country would draw 172 million barrels to combat the world’s largest oil-supply disruption on record and stabilize oil markets.

About 317 million barrels of crude-oil stocks remain as of July 10, according to the U.S. energy information administration. (…)

“These are finite resources,” he [Amos Hochstein, former senior energy adviser to The White House] said at the forum, while also stating that continuing to draw below the 300-million-barrel mark poses serious risks to the structural integrity of the salt caverns where the stocks are stored. (…)

“We think we’re at the cusp of China returning,” Mr. Nuttall said.

If the world’s largest crude importer, China, and other countries with dwindling domestic stocks of crude and refined products increase their imports while Gulf oil remains choked, it could significantly push up oil and fuel prices. (…)

There’s more:

In a report last week, the International Energy Agency said:

• “Since August 2025, at least 100 strikes against Russian refineries have been recorded, with the pace of attacks increasing in recent months. In June alone, at least 10 strikes on refineries were reported.”
• “Almost every large refinery in the western part of Russia has been hit by drones. … Attacks continued through June and into July, with many refineries being hit multiple times.”

“In some regions, drivers have queued for days, fuel sales have been rationed, and stations—including those operated by major retailers—have temporarily run dry. Retail prices have risen sharply, with some independent stations reportedly charging 50% or more above normal levels,” Natasha Kaneva, head of global commodities research at JPMorgan, wrote in a note last week.

• “The disruption has now spread well beyond private motorists. Agriculture, public transport, utilities, logistics and small businesses are increasingly affected, marking a shift from consumer inconvenience to broader operational disruption across the economy,” she added.

Meanwhile, Russia’s traditional diesel customers — which include large emerging markets like Brazil — are scrambling to secure supplies of their own, with many turning to the U.S.

• Foreign buying is helping to push up prices for Americans. (Axios)

So far, the war in Ukraine (now longer than WWII) only had limited global supply chain repercussions. “No-cards-Ukraine” keeps finding smart cards. Perhaps borrowing from Iran’s playbook, Ukraine is now hitting Russia’s energy stack with repercussions soon to be felt worldwide, not only on diesel prices but also on supplies.

• Chinese helium ban threatens supplies to Europe Beijing’s export controls cut off a route for Russian flows, adding to pressures after disruption to Qatari production

Europe is facing an even tighter squeeze on helium supplies as China cuts off exports of the industrial gas that is vital for manufacturing microchips and the functioning of medical devices including MRI scanners. (…)

“China isn’t a source; it’s a conduit”, meaning the halt “pinches a re-export valve Europe had been leaning on”, said Seokjoon Kwon, a professor at Sungkyunkwan University in Seoul. The ban, which was announced as the fragile ceasefire between the US and Iran came under pressure, was an indicator that China was preparing for “renewed scarcity”, he added. (…)

Cliff Cain, commercial manager at London-listed Pulsar Helium, said the market situation had worsened over the past several months, with shortages affecting users such as aerospace and welding companies. (…)

Someone must be thinking this is such a complicated world…

Why Trump Is Going After Brazil’s Beloved Payment System The Trump administration imposed 25% tariffs on Brazil, citing its PIX payment system

(…) few things have provoked more outrage here than Trump’s attack on PIX, the country’s beloved instant-payment system that Washington cited as a key justification for its decision this week to impose a 25% tariff on many Brazilian goods.

From coconut sellers to billionaires, more than 90% of Brazilian adults—more than 140 million people—regularly use PIX, a government-run program that allows users to transfer money in seconds on cellphones at no cost. In less than six years since its creation, PIX now handles more transactions in Latin America’s biggest economy than credit and debit cards combined. (…)

The Trump administration disagrees. Washington said Wednesday that, as of next week, it would impose a 25% tariff on some 3,000 goods from Brazil. It is the first country hit with duties under the Trump administration’s new strategy of using Section 301 of U.S. trade law to punish alleged unfair trade practices.

Among a host of grievances, including anticorruption enforcement and barriers to U.S. ethanol imports, Washington argues that PIX has become so ubiquitous in Brazil it unfairly disadvantages U.S. payment companies such as Visa and Mastercard. There is also growing concern in Washington that Brazil and other countries are seeking to reduce their dependence on the dollar.

The Trump administration says Brazil has tilted the market toward PIX in several ways, including by requiring financial institutions with more than 500,000 active accounts to offer it. Unlike state-backed PIX, private providers, many of them American, must also cover fraud prevention, technology and shareholder returns through fees, making it nearly impossible for them to compete. (…)

“This is the first example and it won’t be the last,” she said, calling it a warning to countries seeking greater control over their payment systems, including the European Central Bank as it develops the digital euro. (…)

Whether buying a Porsche 911, paying bills or giving money to a beggar at a stoplight, the payment is made the same way. With PIX built into banking apps, users can send money using a cellphone number, tax ID, email address, PIX code or QR code. While larger businesses may pay a small fee, it is generally lower than card-processing charges, allowing many retailers to offer customers discounts for PIX payments.

It also reduces the need to carry cash in a country plagued by robbery. (…)

Polls suggest Washington’s attacks have boosted da Silva ahead of October’s election against Bolsonaro’s son, Flávio. Many Brazilians blame the Trump-allied Bolsonaro family for the tariffs, while da Silva has cast them as an attack on Brazilian sovereignty. Brazil consistently imports more from the U.S. than it exports.

The Trump administration is using tariffs to protect outdated and costly US payment companies evolving like comfortable snails while enjoying 65-70% ebitda margins and 30-50% return on capital charging 21% interest.

What do you think this administration will do with AI?

DeepSeek was an early 2025 surprise but Kimi K3 is a confirmation that China, even using inferior chips, is firmly in the AI race with the US, not only on costs but also on performance. In May, Anthropic CEO forecasted that Chinese models would “probably come close to US models in 6-12 months”. It took 2 months.

Time will tell but the US AI world is facing daunting odds:

• A small number of large US companies (Anthropic, OpenAi, Alphabet, Meta, SpaceX) are racing with closed, high cost models against several, decentralized, open, low cost Chinese models that have already proven, even with inferior technology, that they can match on performance (potentially lead) and at much lower costs. The very definition of disruption. Open-weight models allow users to download, customize and run them locally on their own systems while retaining their proprietary data instead of giving it away free to the closed American companies which would then  monetize them.
• The two leading US stalwarts are currently private companies with no other sources of revenues/cashflows, critically fighting (spending) to remain in the race.
• Corporate AI work does better with the smartest models, hence the expensive race to the top, but most of the AI revenues will come from what Global Semi Research calls the “Flywheel”. Mozilla CTO Raffi Krikorian told Axios that “For many routine tasks cheaper models are fast enough, capable enough and can cost up to 50 times less.”

Raw model capability is becoming commoditized very quickly. A model can top the leaderboard today and be matched by competitors a few months later. Model capability still matters. It determines whether a company can sit at the table. But it is becoming harder for model capability alone to form a durable moat.

The real long-term value lies in the flywheel formed by the model, workflow, feedback data, customer relationships, and reinvested revenue.

In many cases, the difference between the top model and the rest of the leading pack is no longer a generational gap. It is often a difference in benchmark design, task preference, and specific use case.

This does not mean model capability is unimportant. Quite the opposite. Model capability is the ticket to the game. But a ticket is not a moat.

• Companies offering open, customizable “sub-models” sitting locally and trained on locked proprietary data will feed from a globally trained model but will act as the local flywheel propelling various AI applications. As GSR says, “The model is not the moat. The Flywheel is.” Most Chinese models are designed as flywheels. Nvdia’s Nemotron is a flywheel.
• The US has a (diminishing) technology lead but China has the enduring engineering manpower, ample low cost power and a huge tech savvy and tech hungry population allowing for rapid and smart build up of the AI ecosystem.
• American companies have experienced how AI can help them grow, compete and control costs. They need and want more, but costs are paramount in a world where competitors will be using low cost Chinese models and/or applications. On OpenRouter, Chinese models already occupy the top five spots by weekly token usage.
• Artificially protecting US AI could well end up like the solar, battery and EV industries, meaningfully hurting the American economy currently running mainly on AI fuel.
• Some American companies are not waiting for a government umbrella. Thinking Machine (former OpenAI CTO Mira Murati), Nvidia (Nemotron) and SpaceXAI (Grok Build) smartly offer open-weight customizable models to feed corporate flywheels.
• David Sacks, a White House AI adviser, told Axios: “I believe the U.S. will win the AI race if we stick with President Trump’s pro-innovation, pro-infrastructure, pro-energy and pro-export vision. The danger is in abandoning that approach in favor of bureaucratic controls fueled by hysteria, fear, and companies seeking regulatory capture.” The “pro-export vision” is questionable but Sacks is right to warn of “companies seeking regulatory capture”.
• Maybe, someone should take Xi to his words last week: “We should seize this rare historic opportunity to encourage open-source, openness, and collaboration.”

The secret Trump administration battle to fight Chinese AI

The Trump administration is showing signs it could ban cutting-edge Chinese AI models — a momentous move that could lock in dominance by OpenAI and Anthropic.

Parts of the administration have tried to implement de facto bans on foreign open-source models before, knowledgeable sources tell Axios. Last week’s rise of Chinese model Kimi is reigniting those efforts. (…)

The administration wouldn’t need to impose an outright ban to get U.S. companies to drop those Chinese platforms.

• “What’s actually happening is slower and more durable,” one source familiar with government discussions said, citing procurement rules, Entity List threats and public pressure campaigns aimed at U.S. companies using Chinese models.
• Instead of a ban, another source familiar with government discussions described a push to highlight potential backdoors and lack of security with Chinese models, and the governance issue that brings.
• It’s an offensive approach where the administration encourages a more innovative U.S. open-source ecosystem, the source added.

Powerful U.S. alternatives are still limited at best compared to cheaper and more practical Chinese models. (…)

• While not outright bans, the approaches under consideration would have a chilling effect on Chinese open-source tech and U.S. companies that rely on it, the source close to the administration said.
• The source described leading AI labs or their allies approaching the administration every 3-5 months with an idea to ban open-source models. (…)

Meanwhile, exploding compute demand keeps meeting very tight capacity as The Information narrates this real world story:

Earlier this week, I wrote about startups moving some of their workloads from Amazon Web Services and other traditional cloud providers to young challengers such as Nebius. That’s in part because startups are having trouble getting the Nvidia chips they need on AWS, sometimes because those chips are only available in too expensive packages.

Capacity can be so scarce that an Nvidia representative told one small AI startup it should rent spare GPUs from a facility associated with the Qatari government, an executive at the startup told us.

And even neoclouds like Nebius don’t have the capacity to serve all potential customers.

For instance, as we reported in our piece, The Biological Computing Company, an AI startup that signed up to use Nebius servers for cost reasons six months ago went to AWS in the past month or two, according to Alex Ksendzovsky, the startup’s CEO. He said Nebius didn’t have the capacity the startup needed, and AWS also dropped its GPU rental price.

Dan Lawrence, general manager of the Americas at Nebius, said that for every Nvidia AI server chip in its facilities, there are four to five companies that would gladly use it. That’s forcing Nebius to pick and choose which ones get the goods.

“We have capacity meetings three times a week where we talk about the customers that are coming and which ones we want to pick,” Lawrence said.

The supply-demand imbalance likely influenced Nebius when it decided to hike prices earlier this year, and “we are still selling out across all chip types at the higher prices,” Chief Revenue Office Marc Boroditsky told analysts in May. (Server and chip component shortages are also pushing up prices that neoclouds and other firms pay for Nvidia server racks, as we explained in this article.)

Nebius’ willingness to cater to numerous small customers stands in contrast to other neoclouds that overwhelmingly prefer to rent out large clusters of graphics processing units to a handful of mega customers. But Nebius also has struck some big deals with Microsoft and Meta.

Lawrence says around 75% of Nebius’ business comes from companies that have already maxed out the servers they can rent from traditional clouds, while the other quarter are moving all their cloud workloads to Nebius.

Lawrence said Nebius’s startup customers are generally AI model makers, robotics companies or sell coding generation tools, such as Cursor and Cognition. He said while Nebius does give out credits to help startups get started, its strategy to attract new business isn’t centered around freebies.

“Rather than offer free compute, we find well-funded startups that can scale and grow,” he said, and later added: “We don’t consider ourselves a low-cost provider.”

Vibe-Coding in the Cloud 

Beyond GPUs, AWS is also facing new competition from an older class of cloud software startups that help businesses develop websites or apps, including Render and Vercel, which are both eight or more years old. These firms can make it easier for nontechnical customers to work with AI tools, compared to traditional cloud providers, some companies and people who work with them say.

“AWS has always been stronger at providing the building blocks,” said Randall Hunt, CTO at Caylent, which helps customers use AWS. “They give you the Legos and you get to build whatever your mind can imagine, whereas Vercel is much more of the ‘Hey, here’s the exact template to follow and this is how you should build.’”

That’s why Trusted Health, a nurse staffing company with around 200 employees, is shifting around 75% of its AWS spending to Render before the end of the year, said Marcelo Silveira, a vice president of architecture. Silveira said the switch will help make it easier for nontechnical staff to use AI tools, without relying on the help of Trusted Health’s engineering staff. Those engineers can now spend more of their time on AI-related projects, rather than working on cloud-related tasks, he said.

For its part, AWS says it offers a bevy of tools to help startups navigate its tech, including AWS Startup Advisor, an AI assistant, and offers its own AI that can help nontechnical customers and coders alike develop apps and sites.

• The AI revolution takes on the world’s most cyclical industry Massive investment plans lead to investor fears of a new boom and bust in memory chips

(…) Samsung Electronics is down by a third from its June high, despite stronger than expected quarterly guidance; SK Hynix had a successful US stock offering, but its South Korean shares are off nearly 40 per cent; and Micron has dropped more than 30 per cent.

The share price volatility illustrates the high-stakes race playing out between demand for memory chips from AI data centres on one hand and the colossal investments being made to increase supply on the other. (…)

Kwon Seok-joon, a professor at Sungkyunkwan University in Seoul, said expansion plans from the big memory chipmakers could push the industry into oversupply by 2028 if AI demand disappointed. (…)

Memory chips have emerged as an important bottleneck in AI data centres, with demand running ahead of supply and big short-term profits at the main producers.

But DRam is a highly cyclical industry, prone to booms and busts. (…)

In recent weeks, all three companies have announced some of the largest investments the industry has ever seen, including what the South Korean government dubbed a “Great Leap Forward” — a combined investment by Samsung and SK Hynix that aims to double South Korea’s DRam output within five years.

Although most of the new facilities are unlikely to come online before 2030, the expansion would involve investments of more than Won2,000tn ($1.5tn) over the next 15 years, stoking fears of another boom and bust.

As well as these additions to supply, the sustainability of DRam demand has come under question, with investors wondering whether so-called hyperscalers can sustain their aggressive AI spending. (…)

DRam shortages have been so great that customers have signed multiyear supply agreements with Samsung, SK Hynix and Micron, reflecting a focus on securing supply. (…)

Kwon of Sungkyunkwan University said oversupply was less likely in HBM because the chips were highly customised. Conventional DRam, however, could face excess capacity from 2029, he said.

The biggest uncertainty is China. ChangXin Memory Technologies is preparing for a $9.8bn listing that could fund further capacity expansion, helping it break the longstanding dominance of the three main players.

Morgan Stanley estimates China will account for about 30 per cent of net DRam wafer additions through 2028, second only to South Korea.

“China will be the decisive variable,” said Kwon. “Korean companies say they will adjust their investments depending on market conditions. But they will find it harder to control supply if CXMT expands more aggressively than expected.”

EARNINGS WATCH

From LSEG IBES:

49 companies in the S&P 500 Index have reported earnings for Q2 2026. Of these companies, 89.8% reported earnings above analyst expectations and 4.1% reported earnings below analyst expectations. In a typical quarter (since 1994), 67% of companies beat estimates and 20% miss estimates. Over the past four quarters, 80% of companies beat the estimates and 16% missed estimates.

In aggregate, companies are reporting earnings that are 12.3% above estimates, which compares to a long-term (since 1994) average surprise factor of 4.4% and the average surprise factor over the prior four quarters of 7.5%.

Of these companies, 84.0% reported revenue above analyst expectations and 16.0% reported revenue below analyst expectations. In a typical quarter (since 2002), 63% of companies beat estimates and 37% miss estimates. Over the past four quarters, 73% of companies beat the estimates and 27% missed estimates.

In aggregate, companies are reporting revenues that are 3.2% above estimates, which compares to a long-term (since 2002) average surprise factor of 1.3% and the average surprise factor over the prior four quarters of 2.2%.

The estimated earnings growth rate for the S&P 500 for 26Q2 is 26%. If the energy sector is excluded, the growth rate declines to 22%.

The estimated revenue growth rate for the S&P 500 for 26Q2 is 12.1%. If the energy sector is excluded, the growth rate declines to 11.3%.

The estimated earnings growth rate for the S&P 500 for 26Q3 is 27.6%. If the energy sector is excluded, the growth rate declines to 25.6%.

These 49 companies (incl. 19 Financials, 10 consumers, 5 IT) reported aggregate profits up 55.5% on revenues up 14.2%.

Trailing EPS are now $302.61, up 5.3% vs June 20. Full year 2026e: $341.72. Forward EPS: $371.54e. Full year 2027e: $404.03 (+18.2%). Note that hyperscalers’ mark-to-market Q1 profits are included ~$16).

Goldman Sachs numbers:

Median stock earnings growth rates are a lot more subdued:

Here’s how the 4 stocks David and I regard as the AI flag bearers are valued vs their last 8-year history (charts and data from Koyfin):

• ASML is the monopoly making the EUV machines needed for advanced chip manufacturing. Its 31.6x forward EPS is just below its 8-yr median (range 26-40). No growth at ASML = no growth in advanced chip production.

• TSMC is the largest chip fab in the world. Together, Apple, Nvidia, Qualcomm, AMD, Broadcom, MediaTek, Intel, and AWS account for well over 60% of TSMC’s revenue. Its 19.4x P/E is also just below its 8-yr median (range 15-23).

• NVDA clearly has the best AI stack in the world: a dominant full‑stack GPU + software ecosystem for training and inference, a highly efficient CUDA platform for programmers, a key provider of various hardware to optimize data center AI infrastructure (rather than just selling chips) and aggressive expansion into “physical AI” (e.g. robotics, autonomous vehicles). The stock is a good reflection of the AI scare, trading at 20.3x forward EPS, lowest since 2018. Its P/E is 53% below its 8-yr median (range 27-47 and below what has been a good low valuation level since 2018 (27). Three possibilities: 1- investors are unjustifiably scared (of AI, of NVDA or both), 2- there is a major re-rating of what has been so far the AI poster child or 3- investors don’t believe the company can meet the current consensus of $9/sh this year, +88%, or $12.83 next year (+43%).

• Alphabet is arguably the most end-to-end Western AI company: it has a frontier model, proprietary chips (TPUs), data centers and widely used software applications its AI can boost as well as being financially very solid. At 27.3, its P/E is hanging on the high end of its range (20-30).

US day traders flock to ‘the most dangerous product in crypto’ Trump administration has opened American markets to highly leveraged perpetual futures

Retail traders are piling into so-called perpetual futures that were only cleared to trade in the US in May despite consumer advocates branding them “the most dangerous product in crypto”.

On certain corners of TikTok, baby-faced investors flaunt the sports cars and villas they claim to have bought with their winnings on these highly leveraged derivatives, which trade all day every day of the year and allow big bets with only a small cash stake. (…)

The head of the Commodity Futures Trading Commission hailed the decision as a “watershed moment” for US capital markets, as part of the Trump administration’s embrace of financial “innovation”.

But for critics, the move encourages yet more speculative, adrenaline-fuelled trading — following the boom in prediction markets — that exposes small traders to losses and makes markets more volatile in times of stress. (…)

“The demand for short-term retail access to risk is massive, people want to gamble,” Davies added.

(…) perps differ from traditional futures contracts in that they never expire and involve no physical delivery of an underlying asset. Traders simply take an up or down view on the price of assets, such as bitcoin. (…)

Offshore venues offer perps with up to 40 times leverage on everything from energy prices to the value of private companies ahead of their initial public offerings. (…)

In traditional markets, investors facing losses will receive an urgent margin call to post extra cash or collateral by a certain deadline to avoid being liquidated.

Perp traders face different risks. Losing bets are instantly liquidated — sometimes with little warning — when they fall below a certain threshold. And winning trades can be “automatically deleveraged” as a last resort to prevent crypto exchanges going insolvent during crises. 

Without the delays introduced by margin calls, the sudden closing of positions can dump more assets into a falling market, further depressing prices and cause more liquidations.

More than 1.5mn crypto traders were liquidated within 24 hours of Donald Trump’s threat on October 10 of fresh tariffs on China, which sent the price of bitcoin crashing about 10 per cent. The global crypto market shed $1.2tn, or 25 per cent, of its value over the following six weeks.

• Gen-Z Traders Go for Broke in Pursuit of a New American Dream Lottery-like meme stocks and options can seem like a way to beat high home prices, inflation and AI job threats.

(…) The share of stock volumes of retail investors has doubled over the past 15 years, Bloomberg Intelligence estimates. Once quaintly dubbed mom-and-pop, they’re also a major driver behind the record option volumes, especially in short-term contracts, where bets look cheap but are far more volatile. Cryptocurrency remains a $2.5 trillion-plus asset class even after multiple high-profile scams and selloffs. Sports betting is effectively legal across all 50 states, and if you want to wager on which nicknames the US president will use on his enemies, or how often Elon Musk posts on X each week, you can do that too.

Social media and the gamification of trading have played a decisive role. But to Lukhey and others, riskier bets are a rational response to the state of the US economy, in which wealth inequality is worsening, the dream of homeownership is increasingly out of reach, and artificial intelligence is expected to replace many high-income jobs. In that context the point is less entertainment than advancement. (…)

Some recent studies suggest this kind of risk-taking may be especially popular among people who are just narrowly cut off from homeownership. (…) Their findings suggest that once renters give up on buying a home, they consume more, work less and take on more financial risk.

“It seems like they’re trying to gamble their way into housing once they start to realize that the traditional method of working hard, saving for it, making a safe investment is no longer guaranteeing you the path to the American dream,” says Yoo, who’s 30. “What’s going to happen when all this generation gives up as a whole?” (…)

To Timothy Fong, a clinical professor of psychiatry at the University of California at Los Angeles, who treats addictive behaviors such as gambling disorders, the rise of degens stems less from the need for financial stability than from the desire to enjoy the moment and emulate the luxurious lifestyles that have become ubiquitous on social media. (…)

Academic evidence on retail trading is almost uniformly discouraging. Individuals tend to trade too frequently, incur high transaction costs and fall prey to hype and over­confidence. Research by Brad Barber, a professor emeritus of finance at the University of California at Davis, shows retail investors often buy into surging assets at their peak. A 2014 paper he co-authored found less than 1% of day traders are consistently profitable. Even in newer arenas such as prediction markets, early evidence by other academics points to a similar pattern: a preference for long-shot bets, often accompanied by costs that erode gains. (…)

“When you have stuff on TikTok and trading narratives and momentum and stuff like that, you can kind of just trade without needing to understand complicated cash flows or have any sort of degree,” he says. “It’s a momentum market.” (…)

Big Mo’s better to stay around. A sabbatical at this time would not be pleasant:

Source:  Topdown Charts

EMBARASSING!

Trump threatens new tariffs on Canada over wildfire smoke

U.S. President Donald Trump is threatening new tariffs on Canada over forest fire smoke that, he said, has “invaded” American cities in recent days, calling it “Willful Negligence” on the part of Canadian authorities.

Mr. Trump, in a post to his Truth Social site on Friday, said the smoke has created an “incalculable” expense for the U.S., and the “cost of this pollution must of necessity be added to the TARIFFS Canada is currently paying.”

“We are holding Canada responsible for the fact that they are not properly maintaining their Forests, and Brush therein, and the United States is being unnecessarily invaded by filthy, polluted, and unhealthy air, the quality of which is dangerous, and totally unacceptable!”

“We”, the people? Or the royal We?

Certainly not the American people I have met and known all my life, including during my investing career and the 43 years I spent most of winters in the US.

Pete Hoekstra, the very undiplomatic U.S. Ambassador to Canada since April 2025, said last Wednesday, before the president he “represents to the best of my ability” wrote the above, “This challenge knows no borders. The United States will continue to coordinate closely with Canada, just as we have for more than four decades of shared wildfire emergencies.”

Right after Trump’s Friday post on Truth Social, the embarrassed mouthpiece felt a need to adjust his previous empathic, neighborly views:

“Businesses are closing, because they don’t feel it’s safe for their workers to go into work. You know, bees are not pollinating. It is (also) affecting the tourism industry. Not taking the president of the United States serious — do that at your own risk.” (Or his own?)

Canadian police and military have already helped to rescue stranded American campers in response to a request from Minnesota Gov. Tim Walz, he [Ontario Premier Doug Ford] said. And Canadian resources are regularly offered to help respond to emergencies in the U.S., including linemen who helped to restore power in Georgia and North Carolina in the fall of 2024. [And 1300 Canadians who help fight to 2020 wildfires in California, Oregon and Washington states].

”If there’s some politicians out there chirping away, well, maybe what you should do rather than complain is send support. Send help,” Mr. Ford said. “Because we have done the exact same thing for our American friends. And that’s what you’re supposed to do.” (Globe & Mail)

Speaking of tourism and embarrassment, one week ago, Suzanne and I were having dinner on a lakefront terrasse 10 minutes from the US border. About half of the guests were Americans enjoying Canadian cuisine at a 30% discount. At a nearby table, we overheard a lady confess to her friends that, when travelling abroad, she presented herself as Canadian. True story!

From Mark Twain, who, in 1873, minted the term The Gilded Age while also, perhaps with long-term vision, supposedly saying that “history rhymes”:

• My kind of loyalty was loyalty to one’s country, not to its institutions or its officeholders. The country is the real thing, the substantial thing, the eternal thing; it is the thing to watch over, and care for, and be loyal to; institutions are extraneous, they are its mere clothing, and clothing can wear out, become ragged, cease to be comfortable, cease to protect the body from winter, disease, and death.
• Loyalty to the country always. Loyalty to the government when it deserves it.
• Each man must for himself alone decide what is right and what is wrong, which course is patriotic and which isn’t. You cannot shirk this and be a man. To decide against your conviction is to be an unqualified and excusable traitor, both to yourself and to your country, let men label you as they may.
• But in this country we have one great privilege which they don’t have in other countries. When a thing gets to be absolutely unbearable the people can rise up and throw it off. That’s the finest asset we’ve got — the ballot box.