What Surging Bond Yields Mean for Consumers and Markets A selloff in government bonds has pushed the 10-year Treasury yield close to its 2026 high
(…) The most immediate reason yields are rising is the resumption of fighting between the U.S. and Iran. The resulting rebound in oil prices has renewed worries about the inflation outlook and spurred a shift in interest-rate expectations, as investors bet that the Fed will need to raise rates to bring inflation down. (…)
Factors other than the Iran conflict have also kept inflation hotter than expected, including surging demand for artificial-intelligence services. That has also supported the case for rate increases. (…)
Right now, rising borrowing costs appear to be an inconvenience to individual borrowers rather than a serious threat to economic growth. Even with the rise in yields, big tech companies have continued their historic borrowing spree to fund the construction of artificial-intelligence data centers. Consumers have kept on spending on everything from auto parts to electronics.
Higher yields can also threaten stocks both by increasing borrowing costs for companies and giving investors an attractive alternative to riskier investments. Here, again, the fallout has been modest, with stock indexes remaining close to all-time highs. But stocks have still underwhelmed of late: The S&P 500 dropped 0.1% Wednesday and is roughly flat this month. (…)
Since the 2008-09 financial crisis, the 10-year yield has only once reached 5%: in October 2023. Back then, its sharp rise dented stocks and sparked hand-wringing about the potential economic fallout.
Fortunately for markets, the yield stayed above 5% only for part of a morning, with that round number leading to a surge of demand from investors. It was back below 4.9% by that afternoon and under 4% by the end of the year.
When bond yields rise, the financial environment shifts in ways that structurally disadvantage stocks. Decades of market evidence and economic theory highlight four primary transmission channels through which higher yields can depress equity valuations.
1. The Present Value Discount Channel: Future cash flows are discounted back to the present using a rate tied to risk-free treasury yields. Growth stocks are more susceptible because their cash flows reside farther in the future.
2. The Asset Allocation Competing Channel: Investors constantly compare the Risk-Free Rate of Return against the Equity Risk Premium (the extra return expected for taking on stock market volatility). When risk-free 10-year Treasuries offer attractive, guaranteed return of capital, income-seeking investors (like pension funds) reallocate capital out of equities and into bonds.
3. The Corporate Cost of Capital Channel: Most corporations rely on debt to finance operations, expansion, and stock buybacks. As existing debt matures, companies refinance at higher interest rates. Increased interest expenses eat into corporate net profit margins, leading to lower earnings per share.
4. The Valuation Compression Channel: Even if a company maintains its earnings, its stock price can suffer simply because investors refuse to pay a premium multiple when less risky cash yields are high.
So far this year, 10Y yields have jumped from 4.0% to 4.6% but they have been stabilizing since May and are only back to their “pre-financial-repression-period” yields (so have real yields btw). Meanwhile, core CPI declined a little from 2.8% to 2.6% but earnings have exploded.
Investors could be reassured by stable inflation (so far) and Kevin Warsh statements to keep it low and by rising profits. But wars in the Middle-East and Ukraine are shaking expectations about economic growth (profits) and inflation (rates).
Last May, Goldman Sachs analyzed how equities have reacted to changes in nominal and real bond yields since 2000. Rising yields are clearly a handicap for equities.
Currently not a friendly trend, is it?
(…) With Treasury yields out to the 10-year trading well above 4% and longer maturities sitting above 5%, investors are getting better compensated for holding bonds, according to Chen, who noted that the market offers “increasingly compelling” valuations. (…)
“Current yield levels provide a substantial cushion against further rate selloff,” wrote Chen, who helps co-manage BlackRock’s $18 billion BlackRock Total Return Fund. BlackRock estimated that 10-year Treasury yields would need to jump roughly 70 basis points from current levels before generating negative total returns over a one-year horizon.
BlackRock, the world’s largest asset manager with $15.3 trillion in assets under management, outlined four potential 12-month return scenarios for the Bloomberg US Treasury Index.
A Fed on hold is seen generating a 6.4% return, while 50 basis points of rate cuts should boost that to 7.2%, the note said. Central bank hikes of 1 percentage point are seen clipping gains to 2.5% next year, while a recession and 1.5% of easing would boost Treasury gains to 11.6%. (…)
U.S. Surges Forces Toward Middle East, Giving Trump Options to Expand Iran War
In the past week, special-operations forces have deployed to the region from their U.S. bases, according to flight-tracking data and U.S. officials. Squadrons of jet fighters have been staged across the Middle East, and bomber aircraft at bases in the U.S. and U.K. are on high alert to ramp up operations, according to one of the officials.
Additionally, more than 150 medics have arrived at the Landstuhl Regional Medical Center in Germany in recent days, another official said. The hospital is the primary location for treating troops injured in combat in the Middle East. (…)
“The big idea here is to give the president and the secretary [of defense] the most flexibility, and the force the most options to present to them,” Votel said. (…)
The troop surge could also be designed to give Trump more capacity to target Yemen’s Houthi militants, who threatened this week to impose a naval blockade on Saudi Arabia. The president said Tuesday that he would “take care of things” if the Houthis targeted ships in the Red Sea.
Trump launched an airstrike campaign against the Houthis last year, which ended after nearly two months in a simple ceasefire, leaving the Houthis battered but not broken. They continued to target some ships but have mostly stayed on the sidelines since the Iran war began in February. They lashed out last week at Saudi Arabia, which they accused of bombing an airstrip to prevent an Iranian plane from landing at San’a airport.
On Wednesday, a tanker sailing through the Red Sea off the coast of southern Saudi Arabia was struck by an unknown projectile, sparking a fire, according to U.K. Maritime Trade Operations. The Houthis’ military spokesman said in a televised statement that the group had targeted two Saudi oil tankers with ballistic missiles, cruise missiles and drones. (…)
Returning to all-out war also risks further draining U.S. munitions and exacerbating disruptions to global oil supplies, as Iran continues to attack commercial ships transiting the Strait of Hormuz and as the Iran-backed Houthis threaten shipping in the Red Sea. (…)
Trump on Wednesday threatened to bomb a bridge or power plant each time Tehran attacks a vessel traveling through the strategic waterway. Current and former military officials caution it isn’t lawful to strike an adversary’s infrastructure for political purposes, The Wall Street Journal previously reported.
Iran’s foreign minister, Abbas Araghchi, said in a social-media post that “any aggression against Iran, including our infrastructure, will compel a powerful and decisive response.” (…)
“Every time that the Trump administration has plussed up militarily, they’ve used it,” said Dana Stroul, former deputy assistant secretary of defense for the Middle East during the Biden administration. “The conundrum of the moment is that there’s just no military solution here with this regime intact. You cannot eliminate every missile and every drone, and limited air power and sea power alone are insufficient to change the regime’s fundamental approach.”
- Oil Nears $100 as Houthi Attacks in Red Sea Amplify Supply Risks Saudi exports from the Red Sea have been a major workaround since flows from the Persian Gulf fell to a trickle.
This is happening when US strategic oil reserves and gasoline inventories are near a multi-decade low.
Mercedes risks US sales ban under Senate China bill
Mercedes-Benz faces a possible ban on selling cars in the US after a Senate committee passed a bill that penalises groups with Chinese investment, in another sign of mounting rivalry between the US and China.
Mercedes, a German automaker that is 20 per cent owned by Chinese companies, had lobbied the Senate commerce committee to dilute the Connected Vehicle Security Act, which would also ban the import of battery systems from China’s CATL, the world’s biggest battery maker.
The bill bars automakers from selling connected vehicles — which use hardware and software for communications — in the US if they are more than 15 per cent owned by Chinese groups. It still requires approval by the full Senate before being reconciled with any version passed by the House of Representatives. (…)
At a meeting to debate the legislation, Ted Cruz, the Texas Republican who chairs the committee, criticised the decision, saying General Motors had lobbied for the measure in an effort to push Mercedes out of the US. His comment came after the panel rejected two amendments that would have eased the restrictions. (…)
Chris McGuire, another former White House official, of the Council on Foreign Relations, said he was very concerned about efforts to weaken the language on battery-management systems, which he said posed a big security threat.
“The electronics inside modern batteries decide when it charges, how hot it gets and whether the safety limits hold. If a foreign adversary controls those electronics, it could use that access to push the battery past its limits until it catches fire, effectively turning car batteries into car bombs,” he said. (…)
CATL said vehicle manufacturers, rather than the group, maintained access to and control over all battery-management system data flows. “They inherently do not have the capability to transmit data externally, nor do they contain independent channels to access vehicle or user information.”
AI CORNER
Intel, AMD sign long-term server CPU deals with Chinese clients as prices surge, sources say
U.S. chipmaking giants Intel and Advanced Micro Devices are signing longer-term purchase commitments with Chinese server customers for data-centre processors as prices surge, two people familiar with the talks said.
The move highlights a broader consequence of the AI boom: demand has spread beyond AI accelerators to memory, networking gear and server processors, giving suppliers greater leverage to seek long-term purchase deals.
AI data centres require not only Nvidia-style graphics processors (GPUs) but also large numbers of central processing units (CPUs) to support servers, storage, networking and inference workloads.
The agreements under discussion typically lock in purchase volumes but not prices, the people said. Most cover about a year of supply, although Intel and AMD have discussed commitments of two years or longer from some customers, one of the people said.
The shift echoes trends in the memory-chip market, where the AI-driven shortage has pushed buyers toward longer-term supply commitments.
The sources declined to be identified because they were not authorised to speak to the media. Intel and AMD did not respond to requests for comment.
The talks mark a shift for server CPUs, which have been easier to obtain than AI accelerators or memory chips.
Tighter CPU supply could raise costs and slow deployment for Chinese cloud providers and internet companies expanding AI services.
Server CPU prices are still climbing in China, with month-on-month increases topping 10% for some products, one of the sources said. Prices of some CPU products have risen more than 40% in China since the start of the year, the source added.
Reuters reported earlier this year that Intel and AMD had notified Chinese customers of lengthy waits for server CPUs, with Intel lead times reaching as long as six months for some products.
The CPU shortage will be among the key topics likely to be addressed on Thursday when Intel reports its quarterly results.
CEO Lip-Bu Tan told analysts in April that demand “continues to run ahead of supply,” especially for Xeon server CPUs. He also cited a multi-year deal with Google as one of several long-term contracts Intel signed in the first quarter.
AMD, due to report in early August, already raised its server CPU market forecast to more than $120 billion by 2030, citing strong demand related to agentic AI workloads.
China is one of the world’s largest server markets, fuelled by rapid construction of data centre racks, AI computing clusters and national computing infrastructure.
The buildout has intensified competition for Intel and AMD processors, even as Chinese buyers face separate U.S. restrictions on access to the most advanced AI GPUs.
The 9th World Artificial Intelligence Conference (WAIC) took place in Shanghai July 17-20. We did not attend but I gleaned these info here and there.
The event drew an estimated 400,000 visits, +14.3% YoY.
Over 1,100 Chinese and international exhibitors showcased 4,486 products.
Representatives from about 80 countries and international organizations attended the event.
Official reporting from the event highlights a stark decline in Western presence compared to previous years.
Procurement deals signed during the event are expected to total 20.36 billion yuan, or roughly $3.01 billion, an increase of 25% YoY.
The conference featured the formal inauguration of the Shanghai Institute for Physical AI and Robotics; the launch of the 2026 Shanghai International Embodied Intelligence Competition; the rollout of the UniAI initiative, which connects Shanghai with cutting-edge AI projects; and a 2026-2030 “one-stop” financial support platform for innovative startups.
Foreign Ministry spokesperson Lin Jian said the conference set a direction for global AI governance and established the World Artificial Intelligence Cooperation Organization (WAICO), which will be headquartered in Shanghai. State media described the organization’s mandate as ensuring the orderly development of artificial intelligence (29 countries signed the founding WAICO charter, but not a single G7 nation joined.)
Chinese President Xi Jinping told the conference that artificial intelligence development should not be dominated by a single country and called instead for international cooperation, describing AI progress as requiring what he termed a “symphony of international cooperation” rather than a solo effort.
Xi also said countries should avoid overstretching national security concerns in the AI field at the expense of others’ security.
Xi also called for governments to establish laws, regulatory frameworks, technological monitoring systems, and early warning and emergency response mechanisms to keep artificial intelligence under human oversight, describing the approach as people-centric.
He also announced that China will provide developing nations with 5,000 opportunities in AI training and seminar programs over five years, and will develop international AI application cooperation centers with ASEAN, the League of Arab States, the African Union, CLACS, the Shanghai Cooperation Organization, and BRICS.
Poe Zhao of the analysis publication Hello China Tech said the conference’s central theme this year was the shift from standalone AI models toward systems capable of being deployed at scale across everyday applications.
Several companies outside China, including Siemens, have adopted Chinese open-source AI models, citing performance, lower cost, and greater customization options compared with closed systems offered by US companies such as OpenAI and Anthropic.
Key showcases included a record 261 Chinese foundation models and major hardware demonstrations like Huawei’s public unveiling of a supercomputing cluster built entirely without American chips.
Huawei self-reports that the 8,192-chip configuration delivers 6.7 times more computing power than Nvidia’s NVL144, according to Huawei’s own Connect 2025 technical specifications. These figures have not been independently verified. Huawei itself has acknowledged it cannot build processors that challenge Nvidia’s GPUs in single-unit performance — the brute-force scaling strategy compensates by connecting 56.8 times more NPUs than the NVL144 has GPUs.
The remaining performance gap is primarily in software, not silicon. CANN (Compute Architecture for Neural Networks), Huawei’s counterpart to Nvidia’s CUDA ecosystem, continues to narrow the gap in library coverage and developer tooling — but has not closed it, as Spheron’s comparison of Ascend 950 and Nvidia hardware documents.
The gap is closing: DeepSeek confirmed its models run on Ascend clusters at production scale — the clearest market validation yet that Huawei’s stack can support frontier AI workloads outside Nvidia’s ecosystem.
The Atlas 950 SuperPoD is scheduled for commercial availability in Q4 2026.
Nvidia was the most talked-about company in the infrastructure halls without actually having a massive sales pavilion. Domestic Chinese hardware giants like Huawei, Biren Technology, and Moore Threads explicitly designed their WAIC product debuts to show they could bypass US trade blocks.
Alibaba Group Holding’s chip design unit, T-Head, announced that it will open-source its proprietary software stack, marking its latest effort to streamline developer operations and challenge the dominance of Nvidia’s CUDA ecosystem.
T-Head announced that it is making the full technical stack of SAIL – the foundational software architecture for the unit’s Zhenwu series of AI chips – freely available to international developers.
The move is part of a broader campaign by Chinese AI chipmakers, including Huawei Technologies and Moore Threads Technology, to promote open, collaborative software ecosystems as an alternative to Nvidia’s dominant CUDA toolkit – the industry standard for writing software for graphics processing units.
The vast majority of artificial intelligence programmers globally remain reliant on Nvidia’s specialised software, which effectively locks them into using the firm’s hardware. By offering alternative frameworks, Chinese technology firms seek to bolster self-sufficiency amid the broader US-China tech rivalry.
T-Head stated that its own open-source initiative was designed to lower the barrier for international developers seeking to adopt its hardware, adding that programmers could adapt the SAIL stack to mainstream AI frameworks in less than seven days.
The strategy also aligns with T-Head’s efforts to accelerate commercialisation of its Zhenwu AI chips, as domestic players race to fill the market void left by Nvidia.
The hardware push underscores how Alibaba is developing full-stack AI capabilities, spanning microchips and servers to AI models and applications. At WAIC, Alibaba’s Qwen AI team also unveiled its first pair of AI-powered earbuds, marking the company’s latest push into smart wearable gadgets.
Beyond hardware, the Hangzhou-based e-commerce and cloud giant launched Meoo Team, a platform designed to help businesses manage resources and asset ownership while creating their own AI applications.
The conference’s Embodied Intelligence Pavilion brought together 161 companies and 314 exhibits, with more than 300 physical robotic units operating on site across five production-process demonstrations.
Unitree Robotics unveiled an unmanned robot factory — a facility operated entirely by autonomous robotic systems, with no human workers in the production loop.
SenseTime’s SenseMart Go featured humanoid robots independently managing a functioning retail store, with visitors able to experience a complete shopping process by scanning a QR code.
A live automated new energy vehicle production line operated by humanoid robots covered five production processes including battery module assembly and interior installation. These are not laboratory demonstrations — they are production-capable systems presented to procurement buyers at a commercial trade event.
The industry ecosystem is rapidly diversifying into an interconnected network of full-stack startups, traditional robotics vendors, cloud large-model giants, and industrial co-builders, moving away from isolated, vertically integrated silos.
The upstream supply chain is seeing an absolute explosion in dexterous hand vendors and mechanical design variations, signaling rapid innovation.
China’s humanoid robot industry is scaling at a rate that has no Western parallel. The country’s annual humanoid robot production is expected to reach 100,000 units in 2026, per China’s Ministry of Industry and Information Technology, with AI adoption among large-scale industrial enterprises already exceeding 30%, according to official NDRC figures published by People’s Daily.
Despite the pre-training foundations, fully “out-of-the-box” capabilities have yet to materialize, and any commercial deployments and real-world demonstrations still heavily rely on specific environmental trajectory collection and on-site fine-tuning.
In industrial application, logistics sorting and warehouse material handling stands out as one of the most commercially-ready applications, where some leading manufacturers’ logistics sorting and palletizing solutions have achieved decent accuracy and tempo, enabling long-term and continuous automated operations with the potential to move from PoC (proof of concept) to volume adoption
In household chores (e.g. breakfast making, clothes folding, and item organization), most applications still appear immature. In live demos, tasks involving refined and soft-object operations (such as putting bread into a toaster, folding clothes while keeping them neat upon placement, and cleaning messy desktops) generally have low success rates, with limited gripper ranges of motion, insufficient perception of soft-object deformation, and frequent errors or item drops during operations, indicating that current technology is still far from commercialization.
Internet/LLM giants are exploring cross-embodiment AI, i.e. “one brain, multiple bodies”, which means a centralized AI that can operate and share learned skills across a wide variety of physical bodies and hardware, leveraging their advantages in computing infrastructure, multimodal large model APIs, and cloud training platforms.
On the other hand, vertical industry companies, represented by automakers/auto component suppliers, electronics manufacturers, and pharmacy chains, are trying to transform from robot purchasers into co-builders, by binding with humanoid body makers or model vendors and opening up their production lines, warehouses, retail stores, and hazardous operational environments to jointly build embodied AI training hubs with data feedback loops.
China’s National Development and Reform Commission announced that the country’s AI-related industry output exceeded one trillion yuan — approximately $147 billion — in 2025, with growth projected above 30% in 2026. AI penetration across China’s key industrial sectors has crossed 80%, and AI-equipped smartphones and PCs are on track to outsell non-AI devices for the first time this year, according to official NDRC figures published by People’s Daily.
The trillion-yuan figure matters as more than a benchmark. It signals that China’s AI industry has crossed the threshold at which domestic demand alone — independent of export revenue — is large enough to sustain a self-reinforcing investment cycle.
That scale creates structural resilience: even if Western export controls close off additional technology flows, the domestic market is large enough to fund continued Ascend chip development, CANN software ecosystem maturation, and deployment at scale across Chinese manufacturing, logistics, and services.
(…) The event that will define WAIC 2026 in policy circles happened before the conference officially opened. On July 16, representatives from 29 nations gathered at the Shanghai World Expo Exhibition and Convention Center and signed the founding agreement of the World Artificial Intelligence Cooperation Organization.
Chinese Foreign Minister Wang Yi signed on behalf of China. Russian, Indonesian, Pakistani, Kazakh, and Laotian representatives signed alongside delegations from ten African nations, twelve other Asian countries, and Belarus, Serbia, Cuba, Brazil, and Venezuela. United Nations Secretary-General António Guterres attended the signing ceremony — lending the occasion UN institutional visibility without formally committing the organization.
No G7 economy — the United States, United Kingdom, Germany, France, Japan, Canada, or Italy — signed on. Nor did Australia, South Korea, or any EU member state.
The rivalry this creates is now embodied in two distinct institutional structures.
- Pax Silica, launched by Washington in late 2025, has signed up 35 countries and focuses on supply chain security — ensuring allied nations have access to trusted AI hardware and infrastructure.
- WAICO focuses on governance and standards, seeking to shape the rules of AI worldwide.
A person familiar with the US position told reporters that Kazakhstan is the sole country appearing on both membership lists, noting that minimal overlap indicates most Pax Silica countries declined to back China’s rival initiative, as reporting on the Pax Silica-WAICO rivalry has documented.
The rivalry is not just institutional — it reflects incompatible premises about legitimacy. Pax Silica operates through a small circle of like-minded democracies; WAICO is open to any sovereign state, with no values or regime-type test for entry.
That design directly contrasts with the G7’s Hiroshima AI Process. Academic analysis from an arXiv paper mapping the emerging governance landscape put it precisely: WAICO enters the field with the strongest claim to a distinctive position and the least developed means of acting on it — noting that who pays, who decides, and how the promise of openness is reconciled with effective rule-making all remain undefined.
French officials have been among the most vocal critics, characterizing WAICO as an attempt to normalize authoritarian AI governance models and undermine the Hiroshima Process, according to analysis of the Western response to WAICO.
India did not sign on as a founding member, with Indian analysts cautioning that democratic nations should remain vigilant about WAICO’s governance implications. Russian Deputy Prime Minister Dmitry Grigorenko offered a less ambiguous framing, casting WAICO as a direct response to American tech dominance: countries should establish transparent rules governing extraterritorial technologies.
The capacity-building pledge is the structural mechanism by which WAICO converts from a governance body into an AI adoption pathway. Countries that receive Chinese AI training, deploy MAZU (China’s AI-powered meteorological warning system), and build on Chinese open-source model infrastructure become technologically aligned with China’s AI stack — and are substantially more likely to adopt domestic regulations consistent with WAICO’s standards rather than with the EU AI Act. The governance split and the technology transfer package are not separate initiatives. They are the same initiative.
Waymo robotaxis show lower crash rates: IIHS
An independent study finds Waymo’s robotaxis are involved in fewer crashes than human drivers — a key milestone for earning public trust.
The research from the Insurance Institute for Highway Safety offers some of the strongest evidence yet that driverless vehicles can outperform human drivers on safety.
IIHS researchers compared crash rates of human-driven cars and Waymo robotaxis in San Francisco, Phoenix, Los Angeles and Austin, four cities where robotaxis are prevalent.
Waymo vehicles were involved in 68% fewer “police-reportable crashes” per mile traveled than human drivers in those cities.
- Waymo had 85% fewer single-vehicle crashes than human drivers.
- Injury crashes were 81% lower on a per-mile basis.
The data generally aligns with Waymo’s own published research, and with industry-wide messaging about the safety benefits of autonomous vehicles that don’t get drunk, drowsy or distracted. (…)
Waymo is the only major robotaxi operator that voluntarily releases driverless mileage data, allowing IIHS to estimate its safety performance. (…)
Saudi Nuclear Enrichment Is a Mistake Trump made the hard decision to fight in Iran but flunks the easier one.
The WSJ Editorial Board:
Why does Saudi Arabia need to enrich uranium on its home soil? The answer isn’t complicated: It doesn’t unless it would like to establish the rudiments of a nuclear-weapons program as a contingency. The more difficult question to answer is why the Trump Administration is going along with it. (…)
The stakes are high. The risk of nuclear-weapons catastrophe increases each time nuclear enrichment spreads. That’s why the bipartisan U.S. consensus has been to stop the spread of the most dangerous processes and capabilities to enemies and allies.
The next best way to prevent nuclear proliferation is with inspections and monitoring, but the Trump deal weakens those, too. The tried-and-true method is the International Atomic Energy Agency’s Additional Protocol. The United Arab Emirates signed up for it, along with the gold standard of no domestic enrichment, as a condition of its 2009 nuclear deal negotiated largely with George W. Bush. But now the Saudis have been allowed to reject the Additional Protocol and receive U.S. support all the same. (…)
Expect Turkey and Egypt to come calling, and we know how this President loathes to say no to Turkey’s Recep Tayyip Erdogan. And why would the U.A.E., which has been a far better ally to the U.S. than Saudi Arabia during the war with Iran, now be content to follow the rules alone? It will be far more difficult to prevent nuclear enrichment capability from spreading across the volatile region. (…)
Yesterday:
