US Producer Prices Unexpectedly Drop, First Decline Since April
The producer price index decreased 0.1% from a month earlier and July’s figure was revised down, according to a Bureau of Labor Statistics report out Wednesday. From the year before, the PPI rose 2.6%.
The report suggests companies refrained from outsize price increases last month despite higher costs from President Donald Trump’s tariffs. While the step back follows a sizable advance in July, many firms have been wary that steep markups could push customers away at a time when economic uncertainty continues to weigh on spending decisions.
Goods prices excluding food and energy rose 0.3%. Services costs fell 0.2%. Within services, margins at wholesalers and retailers fell 1.7%, matching the biggest drop in data going back to 2009 and reversing an outsize increase in July. (…)
“It does look like retailers have been eating tariff costs in recent months,” Stephen Stanley, chief economist at Santander US Capital Markets LLC, said in a note. “Firms have consistently said that they have held the line as long as they could, but that they would need to begin selectively hiking prices going forward.” (…)
Look at the core goods line on this GS chart. Also, core intermediate producer prices increased by 0.4% MoM.
U.S. Population Growth Will Slow Even More, CBO Says Deaths now forecast to exceed births in 2031, two years faster than thought
U.S. population growth will slow to a crawl over the next few decades as fertility rates decline and net immigration shrinks because of stricter enforcement, the Congressional Budget Office said Wednesday.
Deaths are now projected to exceed births in 2031. Just eight months ago, CBO had projected that threshold wouldn’t be crossed until 2033.
By 2055, the U.S. population will be about 367 million, up from 350 million today. In January, CBO had projected a 2055 population of 372 million. From 1975 through 2024, U.S. population growth averaged 0.9% annually. By the early 2050s, according to the latest projections, population growth will effectively be zero.
The demographic shifts are being driven largely by the major changes in U.S. immigration policy during President Trump’s second term. (…)
In July, Trump signed a tax-and-spending law that gave immigration authorities more resources to locate and detain people who are in the country illegally.
The law alone will cause a 320,000-person decline in population by 2035, mostly from deportation and partly from people leaving the country voluntarily, according to CBO.
Net immigration in 2025 is now projected at 400,000, down from two million in the January projections. That is mostly because of declines in a category that includes people who enter the country illegally and those awaiting certain hearings in immigration courts. In 2026, CBO now forecasts net immigration at 600,000, down from 1.6 million.
The changes to immigration and fertility estimates are linked. Slower immigration means fewer foreign women of childbearing age, and those women tend to have higher fertility rates than the rest of the population.
By 2055, the fertility rate is projected to be 1.57 births per woman. That is down from the January projections and below today’s level of 1.6. it is also below the so-called replacement rate of 2.1 births per woman, the level that is roughly needed for a generation to replace itself in the population without immigration.
Data: Congressional Budget Office. Chart: Axios Visuals
China Mulls Helping Local Governments With $1 Trillion of Bills
China is preparing to tackle the significant backlog of unpaid bills owed by local governments to the private sector, according to people familiar with the matter, an amount of arrears some have estimated at over $1 trillion.
The government is considering asking state lenders and policy banks including China Development Bank to lend to local authorities so they can make the payments in arrears, the people said, asking not to be named because the matter is private.
The amount of money under discussion would plug at least 1 trillion yuan ($140 billion) of debt owed to private companies in the first phase of a longer-term initiative, one of the people said. Officials aim to complete the task by 2027, according to the people.
President Xi Jinping warned in a February speech made public last month that the government’s delayed payments to companies risks undermining people’s trust in the authorities. Underscoring the importance Beijing is placing on the issue, China’s top leader said unpaid bills could “cripple” affected businesses in the embattled private sector and was hurting “society at large.” (…)
Local government-related entities in China are estimated to owe 10 trillion yuan, or about $1.4 trillion, to corporates and civil servants — equivalent to 7% of the country’s gross domestic product last year, according to economist David Li Daokui’s estimate. (…)
Chinese banks are struggling with their profitability after being enlisted over the past few years to help prop up the economy with cheap loans, driving their margins to a record low. At the same time, their balance sheets are under pressure from a growing pile of bad loans. In the first half of this year, the five biggest commercial banks set aside allowances for losses on loans of 3.51 trillion yuan, up almost 6% from the end of last year.
Units of state-owned China Overseas Holdings Ltd. and China Resources Holdings Co. are among the bidders indicating interest in Evergrande Property Services Group Ltd., according to the people, who asked not to be identified discussing private matters. (…)
The Chinese government is slowly finding the ways and means to address its housing and local gov. debt problems. Inevitably, the solutions are at the top.
Mexico Plans 50% Tariff on Chinese Cars Before US, Canada Talks
Mexico plans to impose tariffs of as much as 50% on cars and other products made by China and several Asian exporters, aligning the country more closely with US protectionism as President Claudia Sheinbaum prepares for talks over North America’s free-trade deal.
Higher tariffs would apply to a list of more than 1,400 categories of products coming from countries with which Mexico has no trade agreement, Economy Minister Marcelo Ebrard said on Wednesday, describing them as part of efforts to protect Mexican industry. China, South Korea and India are among the exporters that would be hit under the proposed levies, which must be approved by Congress.
The import taxes would also affect items such as auto parts, steel, toys and furniture, with rates of 10% to 50% depending on the category.
“We are going to take it higher to up to 50%, which the World Trade Organization allows us to do. Why? Because the prices at which they are arriving in Mexico are below what we call reference prices,” Ebrard said on the sidelines of an event in the state of Mexico, referring to the car levy. “The main objective is to protect jobs.”
Mexico has become the biggest destination for cars from China, much to the chagrin of its northern neighbor as President Donald Trump wages a trade war on the Asian country. The move serves to appease Mexico’s largest trade partner, but also raises the specter of a broadening economic conflict as Chinese producers seek markets beyond the US. (…)
“It’s a protectionist measure, very much in Trump’s style, that suggests creating a common bloc against China. It’s to be expected ahead of the review of the USMCA,” said Gabriela Siller, director of economic analysis at Mexican bank Banco Base. “It allows Mexico to appease Trump, and in the process collect more money. But there is a cost, and it will be felt by both consumers and producers.” (…)
Mexico has replaced Russia as the top destination for exports of Chinese cars, which rose by nearly a quarter in the first half of 2025 compared with the same period a year earlier, according to the China Passenger Car Association. (…)
Rosales said that the tariff hike would affect investment made by domestically owned distribution companies that have installed more than 800 points of sale for vehicles offered by Chinese brands. This network represents more than 60 billion pesos ($3.2 billion) in investment and more than 32,000 direct jobs, he said. (…)
Honda Debuts $18,300 Electric Kei Car in Japan to Take on Nissan
(…) The car can travel 295 kilometers (183 miles) on a single charge, besting the 180-kilometer range of Nissan Motor Co.’s Sakura, the nation’s most popular EV. (…)
EVs are likely to account for just 3.4% of new vehicle sales in the country in 2025, according to BloombergNEF, making Japan’s transition to cleaner transport significantly slower than the likes of the US, Europe and China.
Still, automakers are wagering that consumer tastes will transform over the longer-term. In addition to Honda, local brands Suzuki Motor Corp. and Toyota Motor Corp. will jointly launch a kei EV this year. Meanwhile, China’s BYD Co. plans to launch a battery-powered kei for Japan next year, the Financial Times reported. (…)
- VW vows to defend Europe crown in fightback against Chinese EVs Competition in region’s electric vehicle market intensifies as German carmaker launch new models and software
(…) German carmakers’ latest comeback signals a further intensification of competition in Europe’s battery vehicle market, where Chinese groups have rapidly expanded their presence. The market share of Chinese brands reached a record 5.7 per cent in the UK and European car markets during the second quarter, rising to 10.7 per cent in the EV market, according to Schmidt Automotive Research.
VW, with its new models, a stronger cost base and its software partnerships with Rivian in the US and Xpeng in China, believes it has the armoury to fight back. Europe’s largest carmaker is by far the dominant player in the region’s EV market, with a 30 per cent share in August. This was up from 23 per cent a year before, while market shares for Mercedes-Benz, BMW and Tesla have all declined, according to Jefferies. BYD, meanwhile, grabbed a 3.8 per cent share in the European EV market, up from 2.5 per cent. (…)
At Munich the Wolfsburg-based group presented a quartet of entry-level EVs, which will go on sale next year starting at €25,000: a new Škoda, Cupra and two new Volkswagen models. (…)
[BYD] plans to bring its ultrafast charging technology to European models from next year and begin producing all of its EVs in Europe within the next three years.
“This is a huge game changer,” said Li, referring to its new battery charging system able to add a range of about 470km in five minutes. (…)
Leapmotor, which has leveraged its capital tie-up with Stellantis to expand its dealer network, says its affordable pricing — its electric B10 compact SUV starts at €29,900 — is a differentiating factor.
“We are almost very close” to price parity between petrol and battery-powered vehicles, said Tianshu Xin, who heads the joint venture between the Chinese EV start-up and Stellantis. (…)
European car executives say it will be difficult for Chinese brands to build cars in Europe at competitive prices as in their home market due to the higher labour and energy costs.
But Li at BYD, which will open plants in Hungary and Turkey, said the company learned from building cars in Thailand and will leverage its cost-effective manufacturing technology. (…)
In the USA:
Automakers reconsidering EV strategies amid weaker sales, additional headwinds: Morningstar
Several car manufacturers are rethinking their commitments to electric vehicle technology as sales in recent quarters have slowed or failed to meet expectations leading some automakers to abandon plans to build certain models while others have lowered annual EV production ranges. (…)
Ford Motor Company recently said it would cancel its planned three-row all electric sport-utility vehicles along with other strategic electrification updates. And General Motors has lowered its planned 2024 EV production to range from 200,000 units to 250,000 units, down from an initially planned range of 250,000 units to 300,000 units, according to the report.
GM also said it will delay the expected launch of the first EV model for its Buick brand. (…)
AI CORNER
Jack Ma-Backed Ant Showcases Its First Humanoid Robot
The company’s unit Shanghai Ant Lingbo Technology Co., also known as Robbyant, demonstrated its R1 humanoid model at the 2025 Inclusion Conference on the Bund in Shanghai. The robot can serve as a tour guide, sort medicine at pharmacies, provide medical consultation or perform basic kitchen tasks.
Ant is the latest big name to delve into humanoid robots, a nascent field fought over by the likes of Tesla Inc. and dotted by up-and-comers like Hangzhou neighbor Unitree Robotics. China, which already has a higher density of robots per human on its factory floors than the US and Japan, is preparing humanoids to move into increasingly complex roles.
Unlike other companies that focus on hardware development, Ant is honing in on developing brains for robots. Ant views humanoids as a strategic gateway to popularizing AI chatbots and assistants, betting that large AI models — still in the early stages of transforming society — will soon reshape how humans interact with machines, according to Zhu Xing, chief executive officer of Robbyant.
“If humanoid robots are used in homes, they won’t just help with everyday tasks, they’ll act like super-smart brains, tapping into cloud-based AI to assist with even more things,” said Zhu. “It makes sense for Ant to be doing this because our goal is to make people’s lives easier, whether it be in payments, finance or digital public services.”
While Ant is best known as the fintech company behind the Alipay digital payments system, it’s been investing heavily in artificial intelligence to ensure its competitiveness in a new era led by the likes of ChatGPT and DeepSeek. The company is developing its own large language model, BaiLing, and testing ways to train it with cheaper, made-in-China semiconductors.
The R1 is built using parts from Chinese suppliers including Ti5 robot for the joint modules and Galaxea AI — which is backed by Ant — for the chassis, according to people familiar with the matter. The company is also in discussions with Unitree and Shanghai-listed Orbbec Inc., the people added, asking not to be named as the information is private. (…)
“One of the most important prerequisites will be the AI model and not necessarily the manufacturing or engineering. China has formidable competitors in this space.”
Ant’s large model enables its R1 robot to deal with end-to-end planning of complex tasks. The company said its AI helps the R1 plan and execute jobs like preparing and serving a meal. It can in theory also learn new recipes and how to use different tools from woks to stoves. That’s because its spatial perception system can recognize relationships between objects such as tables and appliances. (…)
- AI’s $344 Billion ‘Language Model’ Bet Looks Fragile Fortunes are riding on one AI technique, even as cracks start to show.
This year the world’s four largest tech firms will spend $344 billion on AI, mostly on data centers used to train and run so-called large language models (LLMs) like ChatGPT that can process text, audio and visual content. The technology is largely underpinned by the same technique of predicting tokens that appear next in a sequence.
Their spending isn’t all in vain of course. Personal-use chatbots are already growing quickly, with some AI startups starting to break even and businesses still in the early stages of boosting themselves with generative AI. Large language models represent the first AI technique to achieve mainstream adoption at enormous scale: More than 700 million people use ChatGPT each week, for instance. (…)
DeepSeek’s model was an LLM, but its method signaled that all the resources being poured into AI research today drove a tide that could raise other boats. Through its history, AI has moved forward by blending past insights with new ideas, and the pursuit of super-intelligent machines may demand no less.
Much of that exploration now happens at places like Covariant, a Bay Area startup that’s building software to help machines perceive their surrounding space rather than sift through patterns in data. Companies focused on robotics and drones, drug discovery or climate modeling, tend to be those who have naturally stayed away from the language-model obsession because their tech needs to react to physical world conditions in real time. (…)
Now, some cracks are appearing in the large-language model thesis, from the eye-watering costs to the prospect of diminishing returns. The latest models from OpenAI or Google are only slightly better than the older ones, even as more money is poured into their development. Hallucinations haven’t gone away, muddying the path to adoption for companies in healthcare or legal analysis.
A recent study in Nature also shows that the social reasoning abilities of language models — being able to figure out what people really mean in conversation — depend on an extremely small set of model features, and that tiny tweaks can break them. That raises fundamental questions about reliability. Somewhat relatedly, OpenAI admitted last month that ChatGPT’s safeguards for vulnerable people could break down during long conversations. That disclosure came after the bot gave self-harm instructions to a teenager. (…)
Yann LeCun, Meta Platforms Inc.’s chief AI scientist, has long argued that large language models are a “dead end” for smarter machines because they don’t understand their physical surroundings or plan ahead. They’re just “token generators,” he warns.
LLMs aren’t going away, but the history of markets shows the dangers of becoming infatuated with a single solution. Investors and businesses should stay alert for technical breakthroughs and be ready for the ground to shift. In technology it can — and often does — before anyone expects.
- Nebius Raises $3.75 Billion for AI Plans as Debt Deal Boosted
- Chinese Data Center Firm Zdata Seeks $500 Million Private Loan
- Vantage Data Centers Raises $1.6 Billion From GIC, Abu Dhabi
- Alibaba Leads Chinese Tech Funding Spree With $3.2 Billion Deal
(…) The huge appetite for cash among Chinese tech giants is a sign of the bruising competition in the sector, where companies are piling billions of dollars into cloud computing, AI, and even food delivery.
Earlier this week, another Chinese tech giant Baidu Inc. raised 4.4 billion yuan ($618 million) from a dim sum bond offering, following a 10 billion yuan issuance in March. Tencent Holdings Ltd. is considering its first public debt offering in four years with a sale of offshore yuan bonds as early as this month. Meituan is also exploring a potential dim sum bond offering. (…)
The Hangzhou-based company [BABA] said earlier this year it will spend $53 billion over three years on AI infrastructure such as data centers in an ambitious bid to become a leader in artificial intelligence. (…)
AI Wants More Data. More Chips. More Real Estate. More Power. More Water. More Everything
It looks easy enough: Ask ChatGPT something, and it responds. But pull back the curtain, and you’ll find that every ChatGPT prompt and Microsoft Copilot task consumes vast resources. Millions of human beings engineering, correcting and training models. Enough terawatt-hours of electricity to power countries. Data center megacampuses around the world. Power line networks and internet cables. Water, land, metals and minerals. Artificial intelligence needs it all, and it will need more.
Researchers have estimated that a single ChatGPT query requires almost 10 times as much electricity to process as a traditional Google search. Your typical search engine crawls the web for content that’s filed away in a massive index. But the latest AI products rely on what are known as large language models, or LLMs, which are fed billions of words of text—from the collected works of William Shakespeare to the latest forecasts of the Federal Reserve. The models detect patterns and associations and develop billions and billions of so-called parameters that help them mimic human behavior. Using these models, ChatGPT and the like create new content—hence the term generative AI.
The resource-intensive nature of AI will create winners and losers. Those with the most resources will have the most advanced AI systems. (…)
On September 24, 2024 I posted Power Play in the AI CORNER section of the Daily Edge post:
While David and I seeked AI/LLMs investment opportunities, one of our main findings is that the world will be hard pressed to supply the amount of energy needed to feed data centers over the next 10 years. (…)
At present, data centers worldwide consume 1-2% of overall power, but this percentage will likely rise to 3-4% by the end of the decade. In the US and Europe, this increased demand will help drive the kind of electricity growth that hasn’t been seen in a generation. Data centers will use 8% of US power by 2030, compared with 3% in 2022 with a 60/40 split between gas and renewables in meeting new demand growth for this infrastructure. (…)
Between 2023 and 2033, thanks to both the expansion of data centers and an acceleration of electrification, Europe’s power demand could grow by 40% and perhaps even 50%, according to Goldman Sachs Research. At the moment, around 15% of the world’s data centers are located in Europe. (In 2023, 44% of EU electricity was generated by renewables, compared with 21% in the U.S. WSJ)
The U.S. own energy demand will create competition for natural gas between new electricity demands from data centers and the LNG exports Europe depends on. Factset reckons that “with limited ability to move gas from areas of supply to demand centers, the East Coast competes for the same gas in the Louisiana Gulf Coast that serves LNG exports. With the needs of both areas set to rise, the stage is set for competition between the needs of East Coast data centers and the U.S.’ LNG exports, which has become a core pillar of Europe’s supply.”
GS analysts expect incremental data center power consumption in the US will drive around 3.3 billion cubic feet per day of new natural gas demand by 2030, which will require new pipeline capacity to be built.
Utilities in the U.S. East Coast are calling for a combined buildout of 21 GW of new solar facilities over the next five years, which would double the region’s current capacity. If the pace of solar buildout lags, however, more gas will need to be imported into the region. (…)
- One year later, all the numbers above have increased significantly. AI adoption and money grab is much larger than most expected. Those with the most resources will have the most advanced AI systems. And the winners will be in the applications, not in the LLM plumbing.
