+852 3594 6776

Financial BulletinView More>>

“Blue Wave” Won’t Hinder Data Center Construction! White House Official Reassures Market, Trump Says 100% Supports SI

As U.S. tech stocks came under pressure amid concerns over slowing data center construction, Kevin Hassett, director of the National Economic Council, said on Friday local time that despite the “blue wave” in the midterm elections (i.e., Democrats sweeping Congress), with voters overwhelmingly favoring restrictions on tech infrastructure construction, he expects data center construction to proceed as planned.The Democratic Party, which opposes data center construction, swept the House and Senate in Tuesday’s midterm elections. Democrats also scored major victories in Texas and other key U.S. data center hubs. Gina Hinojosa won the Texas gubernatorial race, with her campaign platform including an immediate moratorium on data center construction. This raised investor concerns about AI trades and led to a historic sell-off in U.S. tech stocks on Thursday.Hassett acknowledged on a program that ballots in state midterm elections included dozens of proposals aimed at restricting data centers, which are a key component of President Trump’s push for the United States to win the global artificial intelligence (AI) race.According to Ballotpedia, an American election information encyclopedia website, at least 46 of about 50 such proposals on 2026 ballots received majority voter support, calling for restrictions on data center development.“We respect the local governments that made that choice,” Hassett said. “But the United States is vast, and data center projects already scheduled for the coming years will not be affected by these referendum proposals.”“As for subsequent projects, you know, the United States has enough space for data centers to continue growing, just as predicted before the election,” Hassett said.Hassett said that he expects data centers to continue expanding in states where governors are more friendly, and hinted that even states with opposing attitudes will eventually change their stance.“That is exactly what will happen,” he said. “The benefits brought by data centers will become so obvious that people will decide to change their views on the regulatory approach.”Hassett also attributed Thursday’s tech-driven market plunge to “uncertainty over what political changes mean for the future of data centers.”“We now have a new majority that could step in and create chaos, and I think the market is anxious about that,” he said.He also said the rest of the U.S. economy is very strong, and even if data center construction slows, the economy is still expected to achieve robust growth.Trump Says 100% Supports SIU.S. President Trump strongly supports AI and data center construction, even attempting to reverse its seemingly unpopular status by renaming the technology “Super Intelligence” (SI for short).In a Truth Social post on Friday, Trump made clear that despite the midterm defeat, his support for the emerging industry has not wavered.“As President, I 100% support Super Intelligence (SI) and will work with the many states that want to become rich by having them,” Trump wrote. “They are gold, and only foolish leaders would say ‘no’ to the wealth they bring!”Reprinted from Cailian Press

Details+
Jensen Huang's Wealth Drops Out of World's Top 10

According to Bloomberg, Nvidia's stock plunged more than 9% on Thursday (November 6 Beijing time), dragging CEO Jensen Huang's personal wealth out of the world's top 10.Huang's personal net worth fell by $15.1 billion (about RMB 100.7 billion) to $150.9 billion (about RMB 1,006.3 billion). The loss dropped him from No. 9 to No. 13 on the Bloomberg Billionaires Index.Other tech tycoons with significant wealth declines include SpaceX CEO Elon Musk and Oracle co-founder Larry Ellison, whose fortunes fell by $79.7 billion (about RMB 531.5 billion) and $10.1 billion (about RMB 67.4 billion), respectively.Because Musk's wealth had already shrunk by $33.5 billion (about RMB 223.4 billion) the previous day, his net worth fell below the $800 billion (about RMB 5.3349 trillion) mark. In other words, after the midterm elections, Musk's wealth has shrunk by more than $100 billion.Source: Cailian Press

Details+
A 'Blue Wave' Sweeps Congress: What Lies Ahead for the AI Data Center Industry?

As Democrats ride a "blue wave" in the midterm elections, sweeping both chambers of Congress and reclaiming control, what changes lie ahead for the data center industry? As is well known, data centers have been a focal point of this midterm election cycle. Democrats, like many Americans, have strongly opposed the massive and rapid expansion of AI infrastructure, and they have secured victories in two key data center hubs: In Texas, Democrats scored a major victory, winning the Senate seat and the governor's office, ending Republicans' three-decade-long electoral dominance in this "deep red state." In another key swing state — Ohio — Democratic candidate Amy Acton comfortably won the governorship. She has long advocated for "conditional moratoriums" on data center projects that fail to meet requirements related to cost, union labor, transparency, environmental protection, community benefits agreements, and development standards. However, some analysts do not believe data center growth will stall, given how critical data centers are to the booming artificial intelligence (AI) industry. Instead, they predict investment will continue, but costs will rise — and data center construction will shift to state and local governments that have reached agreements with developers or are more friendly to data center development. Bloomberg Intelligence analysts said data center projects now appear to face greater policy risk, much of it stemming from state and municipal authorities, in the form of moratoriums, permitting restrictions, utility rate changes, and increased reporting requirements. "We do not expect these policies to materially reduce overall U.S. data center spending, but they could delay, shift, or increase the cost of individual projects," they said in a report. Analysts at investment banks Evercore ISI and Raymond James offered similar views. "Democratic gubernatorial victories in Texas, Ohio, and Georgia — states that host a large share of U.S. data center projects — mean new moratoriums and restrictions on data center development," said Evercore ISI analysts. "However, the impact of any moratorium or ban will depend on the specifics, such as whether exemption clauses exist and whether projects facing prolonged disruption can relocate to other states." Raymond James analysts said: "For AI trades, we see the biggest risk not as an immediate collapse in capex, but a significant and possibly underappreciated repricing of the cost, timing, and location of U.S. builds." Gary Cohn, vice chairman of IBM and a former economic adviser during President Donald Trump's first term, also predicted that people will move to places more welcoming to investment rather than see a sharp decline in development. "Data centers can be located anywhere," Cohn said in an interview Wednesday morning. "Take West Virginia... it's got plenty of natural gas. You can build lots of generation behind the meter, totally off the grid, and you can put data centers in West Virginia." Cohn added: "Many regions have realized that data centers aren't that bad — they eventually bring tax revenue, and we can raise teacher and police salaries. Every state and every region has to evaluate for itself. It's going to be a very localized issue. I believe some states will put up real obstacles, and others will lower the bar." 'One of the Highest-Risk Sectors' Of course, many analysts believe data center developers, their investors, and related industries will face headwinds ahead. Lale Akoner, global market strategist at eToro, said that after the midterms, "one of the highest-risk sectors" is the data center industry. "Since July, data center stocks have already priced in some of the potential shock, but the risk does not appear to be fully reflected in share prices," she added. "The knock-on effects of a slowdown in data could impact other industries — construction, electrical manufacturing, and utilities could all face pressure from cross-partisan political resistance affecting data center builds," she added. Billionaire entrepreneur and major Republican donor Elon Musk also commented that America now looks far less friendly than before. "Space data centers now seem quite attractive," Musk posted on social media. Impact on Related Stocks On Wednesday, shares of four of the five "hyperscale data center" companies rose. Goldman Sachs strategists said that hyperscale operators like Alphabet, Amazon, Meta, and Microsoft that already own data centers could benefit if future projects face restrictions. The strategist also noted that delays or cancellations of some data centers could slow down the "picks and shovels" trades associated with these facilities. Cohn also offered his own theory for the rally in hyperscale data center operator stocks. "If you're a company like Amazon, Google, or Meta that owns data centers and has already built them, and now data centers are scarcer than before, then the data centers you're building are actually more valuable today than they were yesterday," he said. "Also, look at those companies — if they're going to spend less on data centers, they go back to being cash-flow-positive companies," he added. Reprinted from Cailianshe

Details+
How Did the Largest Hack in History Happen? Binance's Defenses Fell in a 'Fake Interview'!

Binance on Tuesday published the results of its investigation into last week's theft of $6.3 billion in crypto assets. The investigation showed that the hackers lured one of its software engineers into installing malicious code through a carefully staged fake job interview. "This was not an isolated single intrusion, nor was it a simple phishing event. The attackers combined multiple technical methods, including... fake job interviews, malware downloads, and other approaches," Binance said Tuesday. The largest cryptocurrency heist by dollar amount in history comes as advances in AI are fueling a surge in such hacker attacks in the crypto space, while also exposing the security weaknesses of blockchain technology. But Binance's latest findings highlight the critical role that human error played in the hackers' breach of the exchange's systems. How did the heist happen? Binance revealed that in June this year, an engineer on its wallet development team had downloaded a tampered software program. Although Binance's internal systems detected it in time and replaced the device, the hackers had already obtained information about the engineer's "job functions and details of several colleagues on the wallet infrastructure team." The real breach of defenses occurred in September — a second Binance engineer received a private message on a job-hunting platform from someone claiming to be a project manager at a company (later confirmed to be entirely fictitious). The engineer participated in the so-called job interview as arranged; during the interview, the interviewer not only showed their social media account but also "displayed personal life photos and videos on the platform to gain trust." Binance said this tactic was "more sophisticated and stealthier than previously seen social engineering attacks." Subsequently, the two parties moved to a social chat app for a second-round interview. The other party claimed to be the CEO of that company and asked the engineer to download a "interview test file," which contained hidden malicious code. This download did not trigger security alerts, allowing the attackers to gradually penetrate and control parts of the exchange's core systems through Binance's internal AI tools. "The malware employed sophisticated obfuscation and evasion techniques, including replacing components of legitimate development tools with malicious versions to execute commands and establish communication with attacker-controlled external servers," Binance said. After the malware was implanted, the attackers used the compromised engineer's laptop and a Binance AI tool to take control of the laptops of the first engineer and another (third) engineer. Among them, the first engineer had system code release permissions, providing the hackers with an ideal opportunity. On the day of the incident, October 29, the attackers remotely logged into the third engineer's computer and "triggered a malicious script to launch the attack." Closing the barn door Binance said it has since rebuilt its wallet systems and AI tools and is working with external cybersecurity experts to strengthen its security systems. The company said it has set aside sufficient funds to cover user losses and has restored customers' withdrawal functions. Binance also said it would keep details of some security measures confidential "to prevent malicious actors from using this information to plan future attacks." Binance co-CEO He Yi said, "Although we successfully intercepted the vast majority of threats, a single lapse can come at an extremely painful cost — this situation should never have happened." Reprinted from Cailianshe

Details+
US Midterm Elections Voting Today! Lessons from History: Where Does the Stock Market Go Next?

On November 3 local time, the U.S. midterm elections will mark the official voting day. The U.S. stock market has been uneasy ahead of the midterms, but historically, regardless of whether the election results in a "red wave" (a big Republican win), a "blue wave" (a big Democratic win), or a split Congress, stocks tend to perform strongly after the midterms.The S&P 500 has tended to perform well after both midterm elections and presidential elections. According to a media analysis of S&P 500 returns following elections from 1980 to 2024, the index rises an average of 0.2% within one day and one week after the election, and an average of 1.1% after one month.Extending the time horizon to one year, the S&P 500 gains an average of 13.7% after midterms and 13.8% after presidential elections. Returns were positive in 91% of cases. Moreover, after all 11 midterm elections from 1982 to 2022, the S&P 500 rose every time.However, short-term returns are less stable, especially after presidential elections, where average returns one day and one week later are even slightly negative."There is always some anxiety before an election," said Ryan Detrick, chief market strategist at Carson Group. "Once the election is over, the market breathes a big sigh of relief, and it tends to do quite well."Markets prefer divided governmentHowever, markets typically prefer divided government: one party controls the White House, and the other controls one chamber of Congress.In the 10 midterm elections and 5 presidential elections without unified party control, the S&P 500's average gain one year later was 15%. Fourteen of those 15 had positive one-year returns. By contrast, in the year after elections with unified party control, the index gained an average of 11.5%.Detrick noted that if there is a huge red wave or blue wave, the market actually doesn't like it that much, because it means one side has too much power.Democrats controlling Congress and Republicans controlling the White House — a possible outcome of this midterm — is also positive. A media analysis found that after three midterm elections (1986, 1990, and 2006), when Democrats controlled both chambers and a Republican was president, the S&P 500 fell an average of 0.5% on election day, but rose an average of 3.4% one month later and 10.9% one year later.Of course, what really determines the direction of the stock market are factors more important than elections.In the year after the 2000 presidential election, the S&P 500 plunged 22.1% due to the dot-com bubble bursting, the worst post-election performance since 1980. Conversely, the S&P 500 surged 38.3% in the year after the 2020 election, recording its best post-election rally, when the market rebounded sharply from the COVID-19-induced sell-off.Société Générale believes that investors should remember: Federal Reserve policy and the corporate earnings cycle matter more than elections in the end.For example, after the 2018 midterms, U.S. stocks fell as Democrats seized control of the House during Trump's first term. A more important factor was that the Fed tightened monetary policy that year with four rate hikes; the S&P 500 fell 6.2% that year, but rose 11.7% one year after the election.Detrick said the biggest variable now is whether artificial intelligence (AI) and the tech stock decline will persist. As long as the AI wave continues and the economy keeps growing, "this bull market could remain healthy and strong."But if Democrats sweep both chambers, it would almost certainly put pressure on AI infrastructure expansion.Meanwhile, the global bond market remains tense. On the geopolitical front, a U.S.-Iran war could escalate after the midterms."Seasonal patterns are not inevitable outcomes; the Iran situation, AI risks, and France's debt problems will not simply disappear," said Jim Reid, head of global macro and thematic research at Deutsche Bank, in a report.If history is any guide, the S&P 500 is likely to rise after the midterms. But the current market theme remains AI, with the Fed, bond markets, and geopolitical conflicts still the decisive forces shaping market direction.Reprinted from Cailianshe

Details+
Is AI Spending Burning Through US Tech Stocks? Analysts Warn: History Says the Nightmare May Just Be Beginning!

Think last week's tech stock market was bad enough? Some analysts are warning that if history is any guide, the tech sector's suffering may just be beginning! Last week, US tech stocks tumbled broadly. The Nasdaq Composite fell about 2% for the week and was down 5.3% for October, its worst monthly performance since March 2025. The S&P 500 technology sector dropped 4.24% on Friday, the worst-performing of the 11 major sectors that day. Keith Lerner, chief investment officer and chief market strategist at Truist Advisory Services, noted that since 2023, the S&P 500 technology sector has seen six pullbacks of 10% or more. Of those six declines, four fell between 14% and 27%. He believes this pullback will be no exception. He further said: “The question now is no longer just about strong AI demand. Wall Street is increasingly demanding evidence that massive investments can translate into durable profits and cash flow.” Last Friday, the AI sector suffered a heavy blow, with more than $1 trillion in market value wiped out, as investors worried that big tech's heavy spending on AI would worsen cash flow. Negative cash flow news also emerged one after another during this earnings season. Alphabet, the parent company of Google and a cloud-services giant, reported last week that its third-quarter free cash flow swung from about $24.461 billion in the same period last year to negative $9.065 billion. Subsequently, Amazon also reported negative free cash flow in its earnings released on Thursday evening. A week earlier, Meta disclosed its first negative free cash flow since 2012. However, none of these top tech companies said they would slow down their AI spending; on the contrary, they all said they would maintain or even increase investment. Chris Zaccarelli, chief investment officer at Northlight Asset Management, said the market is “much more sensitive than usual,” adding that investors would rather sell first and ask questions later than stay trapped and keep absorbing losses. Dan Ives, a well-known tech analyst and partner at Yorkville Ives, said, “The disruptive impact of AI was on full display during tech earnings season. There is still much uncertainty over which companies and industries will be the disruptors and which will be disrupted.” He further noted, “This is not the time to panic-sell all tech stocks and head for the bunker... Although last week was tough, the underlying signs of enterprise AI buildout are actually more solid than a week ago... thanks to what we heard from hyperscale data center operators, as well as Meta's and SpaceX's results.” “That said, tech companies raising capital through bond markets, equity offerings, ATMs and other means has also added further pressure to the sector, exacerbating last week's market turmoil,” he added. From Cailian Press

Details+
Polish PM Accuses President and Opposition of Undermining Rule of Law

On October 31 local time, Polish Prime Minister Tusk issued a statement on Poland's judicial dispute in a televised address. Tusk said that regrettable events occurred outside the Constitutional Tribunal building on October 30, and he does not want to see similar scenes again. He stated that the police are responding to various aggressive behaviors in a calm but firm manner. In his address, Tusk accused the Polish president and opposition parties of undermining the rule of law in Poland. He said that the government, together with the parliamentary majority, had attempted to promote reforms of relevant institutions and proposed restoring the rule of law through resolutions and legislation, but these efforts were obstructed by the president and opposition parties. Tusk stressed that compromise is not possible on certain issues. Source: Cailian Press

Details+
Trump reportedly plans to resume strikes on Iran the day after midterm elections | Global Markets

Overnight, all three major U.S. stock indexes closed lower, and most major European stock indexes fell.All three major U.S. stock indexes closed lower, with the Dow Jones Industrial Average down 1.09%, the S&P 500 down 1.82%, and the Nasdaq Composite down 2.86%. On a monthly basis, the Dow rose 1.7% this month, the Nasdaq fell 5.3% this month, and the S&P 500 fell 0.93% this month.Apple's stock fell 1.2% on Friday, lowering its market value to $5.03 trillion; Nvidia fell 4.2%, closing with a market value of $4.93 trillion. This was the first time since the end of July that Apple once again overtook Nvidia to become the world's most valuable company.Federal Reserve official Logan said the Fed still needs to raise interest rates at least twice, by 25 basis points each time. After raising rates by 50 basis points, a pause would be reasonable to determine what further action is appropriate.Chile's Copper Commission raised its average copper price forecast to $6.18 per pound for 2026 and $5.95 per pound for 2027.Trump reportedly plans to resume strikes on Iran the day after the midterm elections. The U.S. notified Gulf states to be prepared for strikes as early as November 4.International oil prices rose on Friday. WTI crude for December delivery rose 2.1% to settle at $104.12 per barrel. Brent crude for December rose 2.4% to settle at $117.64 per barrel; the January Brent contract rose 2.4% to settle at $113.43 per barrel.Market News[U.S. Military denies Iran's claim of destroying U.S. maritime drone]The U.S. Central Command said in a social media post on October 30 that Iran's Islamic Revolutionary Guard Corps claimed earlier that day that it had destroyed a U.S. maritime drone in the Strait of Hormuz, but the claim was false. On the morning of October 30 local time, Iran's military navy said its forces had destroyed and sunk an enemy unmanned surface vessel operating in the eastern Strait of Hormuz. Iran also said that two tankers operating in restricted and unsafe areas east of the Strait of Hormuz ignored previously issued warnings and regulations, struck mines, sustained damage, and caught fire. Iran warned adversaries that any action, infiltration, or provocation within areas under the jurisdiction of Iran's armed forces would be met with a strong response from Iran's navy.[UKMTO reports two vessel attack incidents]The United Kingdom Maritime Trade Operations (UKMTO) reported a vessel attack incident about 40 nautical miles south of Iran's Lavan Island. The report said a vessel was struck by an unidentified projectile on its port side, causing a fire on board. The crew has been confirmed safe, and no environmental pollution has been reported. In addition, UKMTO also reported a delayed report of a vessel attack in the Strait of Hormuz that day. According to verified sources, a tanker was struck by an unidentified projectile while leaving the Strait of Hormuz.[Security Council holds seventh straw poll for UN Secretary-General selection]The UN Security Council held its seventh straw poll on October 30 for the selection of the next UN Secretary-General. After the vote, the council's rotating president for the month, Greece's Permanent Representative to the UN, Barta, said the results would be communicated to candidates through the countries that nominated them. The straw poll is held behind closed doors, with only representatives of Security Council members present, and results are not officially recorded or made public. The votes are "encourage," "discourage," or "no opinion." In the Security Council straw polls, a candidate must receive at least nine encouraging votes and no discouraging vote from any of the five permanent members to be recommended by the Security Council and appointed by the General Assembly.[JPMorgan: U.S. Treasury bill demand may not keep up with supply by 2027]JPMorgan strategists Jay Barry, Pankaj Vohra and others said in a report that U.S. Treasury bill issuance is expected to continue expanding—the market will need to absorb $991 billion in net T-bill supply by 2027—so the key question is whether demand can keep up with supply. "We see some headwinds on this front," the strategists wrote. They expect the pace of growth in money market fund assets to be roughly comparable to that seen during periods of larger inflows in 2024 or 2025, unless there is a major market disruption similar to the 2020 pandemic shock or the regional banking stress in 2023.[Blackstone tells investors Anthropic may not immediately obtain investment-grade credit rating after IPO]People familiar with the matter said the private equity giant told investors participating in Anthropic's latest $63 billion chip financing deal that it expects Anthropic to receive a high-tier junk rating from major credit rating agencies after its IPO. Sources said Blackstone told investment partners in recent months that it expected Anthropic to obtain an investment-grade rating by the end of the year. That statement is contrary to its earlier guidance and could make it harder for the AI lab to raise funds for its massive spending plans.[Five major tech companies, including Microsoft, slash 80% of buybacks due to AI spending]According to data from the London Stock Exchange Group (LSEG) and company financial reports, Microsoft, Google parent Alphabet, Amazon, Meta, and Oracle returned only 9.3% of their operating cash flow to shareholders through dividends and buybacks in the first three quarters of this year, the lowest level in at least 15 years, compared with 27.4% in the same period last year. In the first three quarters, the five companies' buybacks plunged 83% year over year, from $81.8 billion to $13.6 billion[Kremlin spokesman: No plans for Putin-Trump call at present]Russian Presidential Press Secretary Dmitry Peskov said on October 30 that there are currently no plans for a call between Russian President Vladimir Putin and U.S. President Donald Trump. Peskov also said the Kremlin is not aware of the possible three-way call among Putin, Trump, and Ukrainian President Volodymyr Zelensky that Trump previously mentioned.[Fed's Musalem: Further tightening needed, and sooner rather than later]Federal Reserve official Musalem said U.S. inflation is elevated and broad-based and may persist. Monetary and financial conditions remain accommodative. Third-quarter data confirm strong underlying momentum in the economy. Further policy tightening is necessary, and it should be done promptly.[Binance's He Yi: Net affected funds around $3 billion-plus, over 2% of user assets and reserves]Binance co-founder He Yi said that in the hot wallet theft incident, the net affected funds amounted to more than $3 billion, accounting for more than 2% of user assets and public reserve assets. He Yi said user funds are SAFU, and the suspension of withdrawals was to ensure asset safety, thanking industry practitioners for their assistance. Binance will continue to increase security investment and strengthen efforts to combat hackers.[Trump says U.S.-Iran war will soon end]According to CCTV News, U.S. President Donald Trump said on Friday that the U.S.-Iran war will soon end, but did not reveal specific plans. Meanwhile, the U.S. continues to strengthen its military deployment in the Middle East, and the commander of U.S. Central Command has briefed Trump on options for a new round of military strikes on Iran. With the November 3 U.S. midterm elections approaching, the U.S. government is both pressuring Iran through military deterrence and seeking to restart diplomatic negotiations.Reprinted from Cailian Press

Details+
Amazon Earnings Call: High AI Spending Scares Market, Even AWS's Trillion-Dollar "Pie" Doesn't Look Appealing?

After the market close on Thursday U.S. time, Amazon reported its Q3 2026 earnings, with both overall revenue and operating profit rising. AWS cloud business grew 39.3% year-over-year, marking six consecutive quarters of accelerating growth.During the earnings call, management laid out an extremely optimistic target for future growth: AWS maintaining growth above 40% in 2027-2028, and hitting a trillion-dollar revenue scale within 5-10 years.In addition, executives responded to core questions including AWS's trillion-dollar revenue logic, margin pressure, balance between internal and external supply of Trainium chips, Bedrock's growth prospects, robot deployment, fresh grocery e-commerce growth, and capital expenditure payback cycles.However, after hearing the company's expectation that "capital expenditures will increase significantly in 2027", market concerns about the company's AI investment prospects inevitably intensified further, and the stock fell 7.3% in after-hours trading.1. Core Earnings Highlights1. Overall Performance: Total revenue of $201.9 billion (+12% year-over-year), operating profit of $28.1 billion, and operating margin of 13.9%, a record high. Q4 guidance: revenue of $240-247 billion, operating profit of $28.5-33.5 billion.2. AWS: Revenue of $46 billion, annualized $184 billion, backlog of $578 billion, margin 38.7%; AI inference revenue accounted for more than half of AWS AI revenue.Bedrock surge: spending by Fortune 500 companies grew 7x year-over-year, token volume grew 17x year-over-year, with multiple mainstream large models integrated.Orders for self-developed chips Graviton/Trainium increased 20x year-over-year. Trainium is planned for external sales at the end of 2027.3. Retail & Advertising: Retail unit sales grew 15% year-over-year from a high base; fresh grocery and daily necessities grew at 3x the rate of other categories, significantly driving cross-category consumption. Ad revenue was $19.6 billion, up 12% year-over-year, affected by timing disruptions; Q4 growth is expected to recover to nearly 20%. External AI agents have not yet impacted the advertising business.2. QA Highlights (Abridged)1. AWS Capacity and Retail Costs: Quarterly capacity deployment naturally fluctuates, with a large amount of capacity still coming online in the future; Q4 fuel and line-haul freight cost pressures are higher than Q3, and seller surcharges can only cover part of the costs.2. AWS Trillion-Dollar Revenue Logic & RobotsTrillion-dollar logic: large-scale compute orders are locked in early; enterprise AI inference is still in its early stages, with declining inference costs unleashing demand; new scenarios such as security and AI agents continue to open up incremental growth, and many applications will be transformed into agent architectures in the future.Robots: there are now over 1 million warehouse robots, with safety improvements; repetitive jobs are gradually being replaced by machines; new warehouses deeply integrate robots, with large-scale deployment only after meeting cost-efficiency, while also exploring last-mile delivery.3. AWS Margin & Fresh Grocery BusinessMargin: benefiting from efficiency and self-developed chips, but AI business margin is lower than traditional cloud business, and component price fluctuations create pressure.Fresh grocery: users of same-day fresh delivery buy daily necessities at 7x the rate of ordinary users, and non-fresh consumption doubles, representing the core gap in retail growth.4. Bedrock and Balancing Internal/External Chip SalesBedrock (fully managed generative AI product): multiple models will coexist in the future, positioned as an aggregator of various large models, aiming to grow to the same scale as EC2 (Amazon Elastic Compute Cloud product).Chips: constrained by wafer capacity quotas; external sales provide near-term monetization, while internal use on cloud instances can generate cash flow for years; Trainium external sales will start at the end of 2027, with the internal/external supply ratio adjusted dynamically each year.5. Capital Expenditure and Payback Period: 2026 capex is approximately $220 billion, with 2027 spending continuing to rise; the payback period for infrastructure will continue to shorten.6. Impact of Inference Service Model on Margin: Bedrock primarily runs on self-developed Trainium chips; owned/long-term leased data centers plus self-developed hardware bring structural cost advantages, benefiting profitability.3. Management's Core Outlook1. AWS: Growth of >40% in 2027-2028, moving toward trillion-dollar revenue2. Retail: focusing on fresh perishables to drive overall consumption; robots to reduce costs and improve efficiency in the long term.3. Advertising: Q4 growth to recover; external general AI agents pose no material threat for now.Amazon Q3 2026 Earnings Call Full Transcript (AI-assisted translation, some content abridged)Dave Fildes (Vice President of Amazon Investor Relations): Hello everyone, welcome to Amazon's Q3 2026 earnings call. Joining today to answer questions are our CEO Andy Jassy and CFO Brian Olsavsky. I'll turn it over to Andy.Andy Jassy (Amazon CEO): Thanks, Dave. This quarter we delivered revenue of $201.9 billion, up 12% year-over-year, and operating profit of $28.1 billion, up 61% year-over-year. Amazon delivered another strong quarterly performance.On AWS, our strategy is taking shape. Despite a high base from last year when we significantly expanded capacity, AWS still delivered 39.3% year-over-year growth, accelerating for six consecutive quarters, the highest growth rate in nearly five years.Backlog reached $578 billion, with triple-digit year-over-year growth, and the customer base is highly diversified. We deploy an AI accelerator approximately every 14 seconds, and capacity is snapped up by customers as soon as it comes online. Core compute and storage revenue posted the fastest growth in years.Our core business has maintained excellent operating margins over the long term, and AI segment margins are improving rapidly, already outperforming the core business at the same stage of development. We achieved these results despite rising component costs such as memory and high depreciation from large-scale capital expenditure.The payback period for capital expenditure continues to shorten, and we are more confident than ever that investments in AWS will generate revenue and free cash flow far exceeding expectations from a few years ago, with very attractive returns. A strong flywheel effect has formed within AWS. To borrow the flywheel logic from Jim Collins in "Good to Great": completing A naturally drives B; completing B drives C, and so on in a continuous cycle.In AWS's case: leading features, security capabilities, and performance drive customers to deploy applications and data on AWS; production inference workloads tend to sit alongside other data and applications, which in turn generates AI demand for AWS; AI reinforcement learning, post-training, and tool use consume significant core cloud resources, driving growth in the base business; and base business growth in turn pushes us to continuously iterate and enrich AWS's overall product capabilities. This is completely different from a business model that simply provides GPU hosting or short-term data center leasing. It is a long-term platform strategy that helps customers achieve innovation, ensure stable operations, and build reliable partnerships for years of mutual growth. Regardless of hardware supply tightness, this model delivers highly attractive long-term economics.Let me give a few examples of AWS's progress. In the base business, about 85% of global IT spending still goes to on-premises data centers. We are in a 10-20 year cycle where this landscape will be reshaped by cloud computing. In the past, the market generally believed that critical core workloads such as mainframes and storage area networks would never move to the cloud. But now, with dedicated cloud alternatives built on AWS, plus AI-driven migration tools that significantly reduce migration difficulty, such workloads are moving to the cloud at scale. Enterprise customers can reduce capital expenditure, lower operating costs, accelerate innovation, improve developer productivity, and access data more easily to implement AI. Over the past 20 years, we have built an efficient operating system, and core business margins are strong.As the core business continues to grow, we keep reinvesting in foundational capabilities and AI features. For example: customers are generating ever-larger volumes of data and need analytics and AI to unlock its value, but moving data across storage and analytics engines is cumbersome and costly. Through the acquisition of DuckLabs, DuckDB will become the common SQL dialect across AWS storage and analytics services. This modern analytics engine is fast, easy to use, and universal. Customers write a SQL query once and can run it directly on any AWS analytics service. For the first time in data analytics, there is zero migration and zero adaptation cost between query engines. This is significant, and it also helps AI agents perform queries efficiently.In AI, the inference service Amazon Bedrock continues to evolve, with explosive growth. Fortune 500 companies' spending on Bedrock has grown nearly 7x year-over-year, and total token volume has grown nearly 17x year-over-year. Bedrock has become the fastest-growing billion-dollar business in AWS history. Recently, we integrated OpenAI models into Bedrock, and usage has climbed rapidly; we also added Bedrock-hosted Kimi, GLM, and SpaceXAI models. When talking about AI today, security is unavoidable, and this is an important reason customers continue to choose Bedrock. Bedrock is designed to protect customer data privacy: input and output data is not shared with model providers or used to train foundational models.When customers fine-tune foundational models, the platform generates a dedicated private copy for their exclusive use; data is encrypted in transit and at rest. Frontier models are becoming more capable, and software vulnerabilities that were previously hard to find are now being exposed at unprecedented speed. We launched AWS Continuum, which performs vulnerability discovery, prioritization, validation, and remediation at machine speed. Companies such as Capital One, IBM, Infor, Luminate, and Robinhood have already started using it. In Q3, we announced that we would embed this capability into coding tools, including Anthropic's Claude Code, OpenAI's Codex, and AWS's own Kiro. This new security service has received more customer intent than any similar product in our history. AI agent deployment is accelerating, with many agents built on the AgentCore component.Tens of thousands of customers currently use AgentCore; in Q3, the number of intelligent agents deployed on the platform exceeded the total of the previous three quarters combined. In addition to customers building their own custom agents, the vast majority of enterprises also use out-of-the-box agent services. Several of our products are very popular: Kiro for code development, Amazon Connect for contact center agents, and AWS Transform for migration. These products iterate quickly and their customer base continues to expand. The AI office assistant Amazon Quick is also growing rapidly. Large enterprises such as Mondelez, EchoStar, 3M, Southwest Airlines, and DXC Technology have adopted it. It is built specifically for the office environment and never compromises customer data security.Self-developed chip business maintained triple-digit growth year-over-year; revenue commitment orders for Graviton and Trainium each grew more than 20x year-over-year. AI continues to benefit Graviton chips: model post-training, reinforcement learning, and agent tool calls mostly run on CPUs. Graviton is a top-performing CPU with a price-performance advantage of 30-40% over competitors. In Q3 alone, customers using our self-developed chips saved over $2 billion in total. We just open-sourced TorchNeuron, allowing developers to run most of the PyTorch code they are already using directly on Trainium chips without rewriting code. Even as new-generation chips are released, demand for previous-generation chips remains strong. After Graviton5 was launched, backlog for previous-generation Graviton instances doubled quarter-over-quarter. After the first half, spot pricing for short-term capacity of older Trainium instances saw a premium.Demand for Trainium is extremely hot, and many customers are asking us to sell the chip externally. We plan to begin external shipments at the end of 2027.Overall demand for AWS capacity is unprecedented. Enterprise deployment of inference at scale is still in its early stages, but inference revenue already accounts for more than half of AWS AI revenue and continues to rise. Model training also contributes significant revenue. 2026 capital expenditure is expected to be approximately $220 billion, with the vast majority going to AWS; capital expenditures will increase significantly in 2027, and we can clearly see substantial returns on investment, as mentioned on the last call. AWS annualized revenue has reached $184 billion, but there is still enormous room to grow.Thanks to continuous innovation, new foundational and AI products, enterprises are willing to deploy core and AI workloads on AWS, combined with large-scale annual capacity additions. We expect AWS to grow at more than 40% year-over-year in both 2027 and 2028, and to become a trillion-dollar revenue business within 5-10 years. The industry inflection point has just begun. Our strategy is working and should continue to deliver. On the retail side, I think some market participants underestimate the progress this business has made in recent years. It is difficult to find another retail business of this hundred-billion-dollar scale that, including advertising, achieves double-digit operating margins while maintaining its current growth rate.So far this year, product unit sales have grown nearly 15% year-over-year on an extremely high base, after surpassing 30 billion units last year. Over the past twelve months, gross merchandise value has increased by more than $90 billion net. This growth rate is significantly ahead of retail peers, with quarterly incremental GMV roughly twice that of the second-largest U.S. competitor, and we continue to gain retail market share. Over the past few years, operating margins in both North America and international segments have improved by nearly 10 percentage points each, allowing us to reinvest in optimizing customer experience. Nearly 80% of retail transactions still occur in physical stores; in the long run, e-commerce's share will continue to rise.We are building a high-efficiency ultrafast fulfillment network to drive this shift, delivering excellent customer experience while generating substantial free cash flow and returns. Faster delivery, fulfillment cost optimization, fresh groceries and pharmacy, and AI shopping are all progressing in tandem. We are ramping up same-day delivery networks, Amazon Now, and Prime Air ultrafast delivery, with continuous improvement in delivery speeds. In Q3, global delivery performance hit an all-time high, completing over 1.5 billion same-day or next-day deliveries for Prime members. Amazon Now is expanding rapidly worldwide with strong market feedback, now covering more than 450 towns and cities, with order volume growing 10x year-over-year.Unit fulfillment costs are on track to decline for the fourth consecutive year. We can continue to price at or below competitors while cost-effectively supplying fresh groceries and daily necessities. Faster delivery is driving the fresh grocery business, including perishables and daily necessities, with gross merchandise value approaching $250 billion over the past twelve months. In Q3, fresh groceries and daily necessities grew at three times the rate of other retail categories, accounting for about half of all unit sales. Customers using same-day fresh delivery buy daily necessities at 7x the rate of ordinary customers; active fresh customers order from non-fresh categories such as home and apparel about four times as often as ordinary customers, and spend more than double.Amazon Pharmacy is also growing rapidly. The user base grew more than 60% year-over-year, with growth accelerating in every quarter of 2026; repeat customers grew more than 50% year-over-year. In the J.D. Power rankings, Amazon Pharmacy ranked first among mail-order pharmacies for the third consecutive year. The product experience continues to improve, and prescription delivery is now available within three hours in more than 350 cities. Over the past twelve months, more than 400 million customers used the AI shopping assistant Alexa Shopping, with interactions surging 570% year-over-year. In the U.S., customers using the feature spend 50% more per order. The tool can research products, compare prices, review a full year of price history, generate personalized gift recommendations, and watch for promotions and discounts on the user's behalf.Although we will also partner with general external agents, we believe Alexa Shopping will become the preferred agent for customers shopping on Amazon. When browsing products and have questions, there is a shopping assistant at hand that can place orders directly using your order history, shipping address, and payment information, which is extremely convenient for customers. Next, let's talk about Amazon Advertising. Q3 ad revenue was $19.6 billion, up 12% year-over-year.There were some timing-related disruptions this quarter, but overall business momentum remains positive. Three factors affected this quarter's growth:First, most countries moved the Prime member event to Q2, shifting much of the transaction and ad revenue that would have been in 2025 Q3 to 2026 Q2. This benefited Q2 and weighed on Q3's year-over-year growth, causing an 850 basis point quarterly disruption to ad growth.Second, Q2 2026 included the first season of NBA on Prime, which did not exist in Q2 last year; the new NBA season starts in Q4, so Q3 had no related revenue.Third, in Q3 2025, we adjusted our search engine marketing approach to drive more traffic to Amazon's on-site search, improving customer experience while increasing monetization efficiency. That policy began boosting ad growth in Q3 last year, and Q3 this year entered a period of high base comparison. We expect Q4 ad growth to return to nearly 20%.The market naturally wonders whether external general AI agents will significantly impact Amazon advertising. The answer is no. At this stage, very few users complete purchases through external agents, and the experience is not as good as directly visiting a retailer's platform to search or one-click order. The ad team completed several innovations in Q3, with effects expected to roll out over the coming months. The core product, Amazon Ads Agent, integrates the full advertising capability into an AI-first unified platform. Leveraging trillion-level shopping, browsing, and streaming behavior signals, it provides insights, answers questions, and automatically optimizes and executes across the entire advertising process. We also launched full-funnel campaigns in the product, merging sponsored ads, display ads, video, and streaming TV ads into a single campaign.Advertisers input basic information such as budget and duration, and Amazon AI automatically plans, executes, and continuously optimizes. Several other business lines are also showing strong momentum. Prime Video has grown into an important business segment with annualized revenue in the tens of billions and is now independently profitable. Hit series include "Kelly", "The Summer I Turned Pretty", "The Boys", "Off Campus", "Reacher", and "Fallout"; the film "Project Hail Mary" had strong box office; sports content includes Thursday Night Football (best start in history in 2026), Prime's NBA broadcast, NASCAR, the Masters, and UEFA Champions League soccer. We just launched the strongest hardware lineup in our history, with 34 new products across Echo, Kindle, Alexa tablets, Fire TV, Ring, and Bee.The business is still in its early stages, but the start has been strong, and we expect more users to experience it during the holiday season. Amazon Leo is on track to launch initial services this year. We are about to enter the most important period of the year: the re:Invent developer conference and the year-end shopping season. Thank you to all employees for their relentless innovation in improving customers' lives. Amazon is building an unprecedented business landscape. No company in history has ever reached trillion-dollar annual revenue, and Amazon has the potential to have two trillion-dollar businesses: AWS and online retail. This is a rare opportunity, and we are highly energized about the future. I'll turn it over to CFO Brian.Brian Olsavsky (CFO): Thanks, Andy. Let's start with the overall financials. Worldwide revenue was $201.9 billion, up 12% year-over-year; excluding currency impacts, growth was 13%. Worldwide operating profit was $28.1 billion, with an operating margin of 13.9%, a company record. Let's start with North America and international segments. A reminder: this year, most major countries held Prime events in Q2, while in 2025 they were all in Q3. Excluding the impact of the event timing shift, 2026 Q3 year-over-year growth would be 400 basis points higher. Q3 customer engagement remained high, continuing the trend from the first half.North America revenue was $112.9 billion, up 6% year-over-year. Excluding currency impacts, international revenue was $43.1 billion, up 8% year-over-year. Worldwide paid unit sales grew 12% year-over-year. Prime remains the foundation of the business. Hundreds of millions of Prime members enjoy a wide selection, low prices, and the fastest delivery ever, with positive feedback. Over the past year, monthly orders for fresh same-day delivery among Prime members have more than tripled. Now let's look at profitability. North America operating profit was $8.7 billion, with a margin of 7.7%; international operating profit was $1.6 billion, with a margin of 3.7%. On the fulfillment network, teams faced transportation cost pressure from fuel inflation and rising line-haul freight rates.Despite these headwinds, through inventory optimization, shorter transportation distances, and package consolidation, overall unit fulfillment costs declined again this quarter. Shipping costs rose 11% year-over-year, while unit sales increased 12%. Excluding fuel and line-haul inflation, shipping costs grew only 8%, 400 basis points below unit growth. Robots and AI continue to deliver measurable productivity gains and improved employee safety, with effects expected to amplify further. In 2026, deployment of robotic arms such as Cardinal and Sparrow will double. We are also developing foundational robot models that simplify programming and teach complex physical tasks, with the goal of deploying practical robots at scale. Looking ahead, there remains significant room to improve global fulfillment network productivity and reduce unit fulfillment costs while further improving delivery speed.Quarterly operating margins will fluctuate; our core goal is continuous long-term optimization of unit fulfillment costs.Now let's turn to AWS: Revenue was $46 billion, up 39.3% year-over-year, driven by both the base business and AI services, with sequential quarterly revenue growth of $3.8 billion. AWS annualized revenue reached $184 billion. Customers continue to move to the cloud and expand usage of foundational cloud services; companies wanting to fully unlock the value of AI are accelerating their cloud migration. AI investment and base cloud service consumption are strongly correlated. As customers increase AI investment, underlying compute, storage, and network resource consumption rises in tandem; as more AI workloads enter production, this linkage will strengthen further.AWS operating profit was $17.8 billion, with a margin of 38.7%, benefiting from high growth and efficiency gains. Investments in self-developed chips and networking hardware, along with software and process optimization, help us optimize capacity and build a more cost-effective infrastructure. As we have repeatedly mentioned, AWS operating margins will fluctuate due to the scale of investments in a given period. Customer adoption of AI products continues to rise, with AI revenue growing triple digits year-over-year, and Bedrock is the core growth engine. On capital expenditure, Q3 cash capex was $75.9 billion, mostly directed to AWS and AI, consistent with strong customer demand. We continue to invest heavily in AI, which we believe is a massive opportunity that will drive long-term revenue and free cash flow and improve returns on invested capital.Now let's provide Q4 guidance. Q4 net sales are expected to be between $240 billion and $247 billion. Based on current exchange rates, currency movements will create a year-over-year headwind of about 60 basis points. Q4 operating profit is expected to be between $28.5 billion and $33.5 billion. One factor affecting guidance: Amazon Leo commercial services are scheduled to launch initial services this year; related production and launch costs will be capitalized and deferred to Q1 next year, later than I expected on the prior call. As a result, most Leo-related expenses in Q4 will still be recognized as current-period expenses. We are ready to serve customers during the holiday season while supporting the countless businesses building on AWS.Andy Jassy (Amazon CEO): Thank you to the global Amazon team for their hard work and customer obsession. Continuously improving the customer experience is the only reliable path to creating long-term value for shareholders. Let's move to Q&A.Moderator: Thank you. We will now open the Q&A. The first question comes from the Bank of America analyst.Analyst: Thank you. Andy, the cloud business growth acceleration is very impressive, but the sequential revenue increase has slowed. What is driving that from the customer side, token costs, or other factors? And how do you view Q4 capacity deployment? Brian, how much impact will fuel costs have on Q4 guidance? Thank you.Andy Jassy (Amazon CEO): Let me address the growth question first. We are very pleased with the 39.3% growth rate. Achieving this at an annualized revenue of $184 billion is remarkable, with six consecutive quarters of acceleration and near a five-year high. We are adding capacity at scale every quarter, but the amount of new capacity each quarter is not uniform. Looking at this year's capacity additions and the forward plan of roughly 8-10 quarters, there is still a significant amount of capacity coming online.Growth is driven by multiple factors. In AI, our full-stack product capability is key: you can build models from scratch with SageMaker AI, or securely and cost-effectively access various top frontier models through Bedrock.Customers can use AgentCore and Strands to build and securely run AI agents at scale, or directly use out-of-the-box agent services, which is what most enterprises choose: Kiro for coding, Amazon Connect for contact centers, AWS Transform for migration, and the fast-growing office AI assistant Amazon Quick. Customers are adopting quickly. The scale of new capacity each quarter will fluctuate.Even at our current scale, we are still guiding to more than 40% annual growth in 2027 and 2028; a trillion-dollar revenue within 5-10 years is a rare and major opportunity, and we are fully committed to it. Self-developed chips also help: enterprises deploying inference at scale place high importance on price-performance. Trainium offers a significant price-performance advantage over competitors and is very attractive. The widespread adoption of AI drives reinforcement learning, post-training, and tool use, boosting demand for foundational compute; Graviton CPUs offer 30-40% better price-performance than competing x86 products. These multiple factors are jointly driving high growth.Brian Olsavsky (CFO): On fuel and line-haul freight, I can't give a precise number, but Q4 pressure is higher than Q3. Earlier this year, we introduced a fuel surcharge for FBA third-party sellers, which covers some but not all of the fuel costs; the remaining pressure is reflected in retail inbound and outbound logistics costs, with overall levels higher than Q3.Moderator: The next question comes from the Morgan Stanley analyst.Analyst: Thanks for taking my questions. Two questions. Andy, you are very optimistic about AWS's prospects, and you rarely give long-term targets like a trillion-dollar revenue in 5-10 years. Is this trillion-dollar target derived from top-down calculations or bottom-up reasoning? Where does your confidence that this revenue scale can be achieved come from? Second question, for retail fulfillment, what is the core robotics milestone in 2027?Andy Jassy (Amazon CEO): On AWS, looking at the current annualized revenue growth and our projections for 2027 and 2028, simple math shows rapid revenue expansion. Large-scale capacity at megawatt and even gigawatt levels is pre-booked by customers and cannot sit idle. For these large orders, we have good visibility into demand for 2027, 2028, and even parts of 2029. At the same time, we still reserve substantial on-demand capacity (EC2, Bedrock) flexibility to ensure customers can use resources flexibly.Enterprises are still in the very early stages of deploying inference at scale in production. On one hand, internal expertise in AI fine-tuning and post-training is still insufficient; on the other hand, inference cost remains a barrier. Continuously lowering inference cost is a major AWS mission, and it is a key reason we developed our own Trainium chips. Inference costs have already fallen substantially over the past two years, and they will continue to decline significantly in the future.At the same time, new AI use cases are constantly emerging. In security, for example, new models can find vulnerabilities that are very difficult for humans to discover. Some market views believe that security-related token consumption will account for a significant share in the next year or two. Add code development, consumer and office AI assistants (Quick), and these are just the first batch of large-scale scenarios. In the future, almost all enterprise production applications will be transformed to support intelligent agents, enabling rapid iteration and better defense against security risks. The AI opportunity is enormous. While it is hard to predict precisely over 5-10 years, the trillion-dollar revenue target is not unrealistic imagination.Now on robots. Amazon's fulfillment network now has over 1 million robots, delivering real improvements in cost, productivity, and safety, with warehouse incident rates at an all-time low. We have made great progress over the past 4-5 years, but both our deployed products and solutions under development are still in their early stages. Fulfillment centers will continue to employ many people, but a significant number of repetitive manual jobs will be given to robots. We are investing in many forms of robots. In the future, we will also explore last-hundred-feet delivery solutions. We evaluate robots comprehensively on productivity, safety, cost, and delivery speed, and only deploy at scale when economic criteria are met. New same-day warehouses are designed with deep robot integration. What robots have achieved so far is just the tip of the iceberg compared to their potential in the next five years.Moderator: The next question comes from the Evercore ISI analyst.Analyst: Two questions. First, thank you for sharing the long-term growth outlook for AWS. I'd like to ask about margins. There will be quarterly fluctuations, but over the next two years, can AWS margins hold current levels, staying solidly above 30% or even moving toward 40%? Second, on fresh perishable fulfillment, I see it rolling out quickly around me. How many Prime households does it currently cover? And roughly how much consumption lift does it bring to users — is it 20%, 50%, or 100%?Brian Olsavsky (CFO): We don't provide guidance on AWS margins, but I can comment. The 38.7% operating margin this quarter is robust. This result is no accident; it comes from ongoing efficiency gains, capacity optimization, tight management of fixed costs, and the increasing proportion of self-developed chips that Andy mentioned.But there are also disruptive factors: AI is expanding rapidly, and AI segment margins are lower than the traditional base business, so the mix shift creates a blended effect. Although AI margins at the same stage are already better than the early base cloud business, the absolute level is still lower. Component prices also create volatility. With multiple factors at play, we continue to focus on cost, efficiency, and scaling new products.We are also very optimistic about the fresh grocery business. Amazon's retail fresh business has long focused on shelf-stable goods, canned goods, personal care, and pharmacy; Whole Foods, as a premium fresh format, with the new small urban Daily Shop model, is growing revenue and profit faster than the rest of the fresh segment and will continue expanding over the next two years. We have always believed that to grow fresh, we must make perishables work. Comparing consumer total spending with actual spending on Amazon, the largest gap is in fresh groceries, and perishables are the gateway to fresh consumption. Daily necessities unit sales are growing at three times the rate of the rest of retail and account for half of total units. Customers using same-day perishable delivery buy daily necessities at 7x the rate of ordinary customers; fresh customers order from non-fresh categories such as home and apparel about four times as often, and spend double. Perishable fresh business will significantly close the gap between us and customers' total shopping needs.Moderator: The next question comes from the Goldman Sachs analyst.Analyst: Two questions related to AWS. First, what key lessons have been learned from Bedrock's development so far, and how will it continue to scale? Second, with Trainium planned for external sales, how do you balance the commercial opportunity of external sales versus internal cloud compute demand?Andy Jassy (Amazon CEO): On Bedrock: looking back at over two decades of experience building technology products, there is never a single dominant tool. This is true in databases and analytics engines; in the early days of AI, people thought TensorFlow would dominate, but now PyTorch has a larger footprint. The same applies to large models. In the next 2-3 years, there will be 6-8 mainstream models of comparable strength, each specializing in different areas such as reasoning, coding, image, low latency, and low cost. The vast majority of enterprises will use multiple models simultaneously, and even a single generative AI application will call multiple models.What customers really want is a service that aggregates a large number of frontier models, offers excellent price-performance, and has robust security and access controls. Customers can freely combine and switch models to adapt to business changes. That is Bedrock's positioning. It includes closed-source models from Anthropic and OpenAI, as well as open-source models like Kimi and GLM. The Western market will also expect more local open-source large models, which will gradually emerge. Bedrock has enormous market potential and an opportunity to grow to the same scale as EC2.Balancing internal use and external sales of self-developed chips: there are actually only two companies globally that have both a large-scale cloud business and chip capabilities strong enough to sell externally. High-end chips all come from the same wafer foundry, and annual capacity is subject to quota limits. Once you get the quota, you need to decide how much to keep for the cloud and how much to sell externally. External sales can quickly recognize short-term revenue; but if used internally to run EC2 and Bedrock instances, the chips generate revenue and free cash flow over a 5-6 year lifecycle, providing higher total long-term returns, so we need to weigh this carefully.Our chip strategy differs from that of another peer. We plan to start external sales of Trainium at the end of 2027, and we will re-evaluate the balance each year. The external chip business and the cloud infrastructure business are completely different businesses; the price-performance advantage from self-developed chips is a huge strategic advantage whether used to lower internal cloud costs or to develop external sales.Moderator: The next question comes from the JPMorgan analyst.Analyst: With AWS growing at 40%+ over the next two years, how much capital expenditure is needed? Also, what is the latest on the payback period for server, network, and data center infrastructure investments?Andy Jassy (Amazon CEO): Let me address the payback period first. We continue to shorten the payback period for infrastructure. AI is different from traditional core hardware models, and the cycles vary by category. AWS has historically been good at extending hardware useful life and shortening payback time, and we are still making steady progress.The scale of capital expenditure will be very substantial. AI is the biggest technological change in our lifetime. To maintain 40% growth on an annualized base of $184 billion and move toward the trillion-dollar goal, we must continue investing to consolidate our leadership. If the transformation could be spread over 25 years, the capex pressure would be smaller, but that is not the reality. We have the capacity to support this investment. This investment will permanently reshape Amazon's revenue and free cash flow, and we are committed to it to maintain our industry leadership.Moderator: The final question comes from the Barclays analyst.Analyst: A follow-up on Bedrock. The vast majority of Bedrock workloads run on Trainium chips. As the business shifts more from bare-metal sales to inference-as-a-service, how will that affect AWS revenue, especially margins?Andy Jassy (Amazon CEO): That's right. The vast majority of Bedrock currently runs on Trainium. We do this largely because we are very satisfied with the price-performance it offers, which allows us to lower prices for customers as much as possible while also achieving desirable margins.If you think about margins in cloud computing and look at the cost structures of different providers, you will find a few things worth noting.First, their data centers: how much do they own or long-term lease, and how much is colocation or new cloud? New cloud operating margins are about 3,500 basis points lower than long-term leased data centers.Then you need to look at the hardware they use to run servers. If you use the same hardware as everyone else, your cost structure is similar to everyone else's. If you have access to different hardware, you have an advantage.If you look at our situation, the majority of our capacity is owned or long-term leased, which gives us operating margins much higher than companies that primarily use colocation or new data center models, or a large portion of their capacity in colocation or new data centers.We use our own chips. This not only provides price-performance advantages, but also, because we own them, it is more cost-effective for us. I think this gives us some structural cost advantages. When we pass those advantages back to customers, I believe it is very helpful for customers and beneficial to AWS's overall economics.Moderator: Thank you for joining our call today and for your questions. A replay of this call will be available on our Investor Relations website for at least three months. We appreciate your interest in Amazon and look forward to speaking with you again next quarter.Reprinted from Cailianshe

Details+
Fed Holds Steady! "New Bond King" Publicly Questions: Why No Rate Hike, What Are You Waiting For?

Jeff Gundlach, founder of DoubleLine Capital and known as the "New Bond King," publicly questioned the Federal Reserve's decision to hold rates steady on Wednesday, saying he simply cannot understand why the Fed has not hiked further. "What are you waiting for? You just upgraded your assessment of the economy," Gundlach said, referring to Fed Chair Kevin Warsh's remarks about the strong macroeconomic environment at the press conference following the latest rate decision. "Headline CPI inflation is heading toward the 4% level. The commodity index is rising, and the commodity market is in a bull market. I don't understand why the Fed is so reluctant to act," Gundlach said. Gundlach's comments came before the Fed announced early Thursday morning Beijing time that it would keep its federal funds rate target range unchanged at 3.75% to 4%, in line with broad market expectations. After the Fed's 25-basis-point rate hike in September, Gundlach had already said the Fed should have hiked more. At the time, he believed the Fed should have raised rates by 50 basis points in one move, rather than gradually increasing in smaller steps across multiple meetings. "I just think the Fed is a bit behind the curve right now," Gundlach said Wednesday. He noted that the Fed's overnight rate of 3.75%-4% is significantly below the roughly 4.7% yield on 2-year Treasuries. He argued that this spread indicates the Fed needs to hike further. Despite the Fed's hold this time, according to the CME FedWatch tool, the probability of a 25-basis-point rate hike at the December meeting stands as high as 85%. The Fed said in its statement that it would determine appropriate policy based on incoming information to ensure inflation returns to its 2% target; however, Warsh reiterated at the press conference that future policy decisions cannot be determined at this time. "Unless oil prices fall significantly, I think they have to hike in December," Gundlach said. U.S. oil prices remained above $100 per barrel on Wednesday. Gundlach also said that after the Fed's latest decision, his investment strategy remains unchanged: 30% of the portfolio is allocated to stocks—he particularly favors the S&P 500 Equal Weight Index to avoid overexposure to what he calls potential artificial intelligence "problems"; 30% to bonds; 20% to commodities; and the remaining 20% kept as "dry powder." Warsh also said Wednesday that the Federal Open Market Committee will still hold eight monetary policy meetings per year. However, in 2027, one monetary policy meeting will focus on broad monetary policy strategy, with no vote on the federal funds rate and no monetary policy statement; in 2028, two monetary policy meetings will be dedicated specifically to discussing monetary policy strategy. Gundlach said the move was not surprising and expressed support for immediately reducing the number of rate decision meetings per year to six. "I think eight meetings a year is really unnecessary. It feels like these meetings come too fast... At least give us a few months to let more economic data come out and be assessed," Gundlach said. Reprinted from Cailianshe

Details+

SG International Group

One-stop world-class wealth management platform

SG International is an independent, all-in-one and global wealth management platform to provide total solutions to achieve your financial goals

You can enjoy our personalized services of a boutique organization and the global resources of the largest and most respected financial firms in the world through SG International. Our financial advisors are able to find solutions that address to your specific needs in terms of strategic investment approach, designing portfolio and macro capital management services.

Shanggu Advantages

  • Investment management - Trinity

    Trinity service, to provide you with more professional and convenient investment services

    1. ·Investment bank
    2. ·EAM (Enterprise asset management system)
    3. ·Insurance
  • Professional

    Built on professional personal wealth management experience.

    Our wealth managers provide a comprehensive solution for customers through banks, trusts, global investment platforms and insurance solutions.

  • Integrated offshore financial platforms

    Professional team makes diversified products and various types of platforms more in line with personal needs, asset security worry free.

    1. ·Private banking platform
    2. ·Family trust platform
    3. ·Securities platform
    4. ·Individual: insurance platform
    5. ·Fund platform
    6. ·Personalized service
    7. ·Investment banking platform
    8. ·Business: Law Package

Shanggu Business

Securities

The full range of securities trading platform, more professional analysis and guidance.

Financial Consultant

To invest as a leading and provide a full range of corporate financial services.

Asset Management

Provide two services for private banking and family offices, asset analysis and management for individuals and families

Fund Customized Services

Analyze assets to customers and provide professional asset manageme

Insurance

Provide a comprehensive solution to meet customer needs. Provide the most suitable choice of insurance products and companies to meet customer needs.