MONEY MASTER

Intel’s shares fell more than 11% after Q4 earnings, not because the quarter was weak, but because management made it clear that the recovery investors were hoping for will take longer to materialize. The numbers showed progress in some areas, but guidance and commentary reinforced a familiar message:  Intel’s turnaround is real, but it remains uneven and back-loaded . 1. Guidance, Not Q4 Performance, Drove the Selloff On the surface, Q4 was not a disaster: Revenue of $13.7bn came in above expectations Gross margins exceeded consensus Non-GAAP profitability improved meaningfully year over year However, the market was not focused on what Intel just delivered — it was focused on  what Intel cannot yet deliver in early 2026 . Q1 guidance pointed to: Sequential revenue contraction A sharp step down in gross margins Another GAAP loss That outlook effectively reset expectations for a near-term earnings inflection. 2. Server Demand Is Strong — But the Earnings Impact Is Gradual Intel...

Alibaba Group Holding Ltd  is preparing to spin off and potentially list its AI chipmaking unit  T-Head , tapping into rising investor appetite for companies seen as local alternatives to  Nvidia Corp  in the fast-growing AI accelerator market. According to people familiar with the matter, Alibaba plans to first  restructure T-Head into a partially employee-owned business , before exploring an initial public offering. While the timing and valuation remain unclear, interest in Chinese chipmakers has surged amid Beijing’s push to reduce reliance on US technology. The move comes as Alibaba steps up its ambition to become a  leading AI powerhouse , investing heavily in infrastructure, chips and consumer-facing AI services. CEO  Eddie Wu  has pledged more than  US$53 billion  toward AI and infrastructure, with room to scale further. Why T-Head Matters T-Head plays a strategic role in Alibaba’s ecosystem, supplying chips that support its ...

Amazon and Nvidia have deepened their AI alliance, integrating Nvidia’s accelerated computing technologies directly into Amazon Web Services’ (AWS) next-generation custom silicon — marking a major escalation in the global AI infrastructure arms race. At AWS re:Invent, the two tech giants unveiled expanded support for  Nvidia NVLink Fusion , a new interconnect architecture that links Nvidia GPUs with AWS-designed chips such as  Trainium4  and future  Graviton  processors. AWS is building its future AI platforms using  NVLink Fusion + Nvidia MGX rack architecture , creating a unified high-performance fabric capable of powering the next wave of large-scale AI models. Advanced GPU Access & Sovereign AI Clouds AWS is also broadening access to  Nvidia’s Blackwell GPUs , which now power AWS’s new  AI Factories  — dedicated sovereign AI cloud clusters designed for global enterprises needing local, regulation-compliant compute. Nvidia says this ma...

 Amazon has officially stepped deeper into the  AI arms race , launching  Trainium 3 , a new chip built to challenge Nvidia’s dominance in the GPU market. The latest hardware — available through Amazon Web Services — promises  4x faster model training  than its predecessor while using the  same energy footprint . Each cluster of Amazon’s new “UltraServers” can run up to  144 Trainium 3 chips , enabling large-scale LLM training at hyperscaler level. The move positions Amazon directly against  Google and Nvidia , both of which dominate today’s AI infrastructure landscape. Google’s lead in the AI model race has reportedly pushed  OpenAI CEO Sam Altman  into a “code red” mode. Crypto Miners Become AI’s New Powerhouses The AI boom has created a problem even Big Tech struggles with:  massive energy and space requirements . Enter crypto miners — companies already running large-scale, power-heavy data centers. After the  2024 Bitcoin h...

US President Donald Trump and Chinese President Xi Jinping held their first conversation since last month’s tariff truce, covering a wide slate of geopolitical and economic issues — from trade and Taiwan to Russia’s war in Ukraine — while Trump confirmed plans to visit Beijing in April. Trump described the call as “very good,” saying both sides discussed agricultural purchases, including soybeans, and stepped-up cooperation to curb illegal fentanyl shipments. He added that he invited Xi for a state visit to the US next year. But Trump’s public readout avoided the topic that Beijing highlighted most: Taiwan. According to China’s Foreign Ministry, Xi stressed that the island’s “return” is central to the post-World War II order and urged the US to maintain the positive momentum gained after their October meeting in South Korea. Beijing Pushes Taiwan Issue; US Stays Silent Xi’s remarks came as tensions rise between China and Japan, a key US ally, following comments by Japan’s new Prime Min...

The US Justice Department has charged four individuals in an alleged scheme to illegally export Nvidia’s high-end AI chips to China — a case that has intensified calls in Washington for strict, end-to-end tracking of advanced semiconductors. According to prosecutors, two US citizens and two Chinese nationals conspired to bypass export controls by funnelling Nvidia A100, H100 and H200 GPUs through Malaysia and Thailand using fake contracts and falsified paperwork. Authorities say the group successfully shipped 400 A100 GPUs into China between October 2024 and January 2025 before US law enforcement intercepted further attempts involving 10 supercomputers and dozens of top-tier GPUs. Congress Pushes for Chip Security Act The revelations prompted immediate political reaction. John Moolenaar, chair of the US House Select Committee on China, urged Congress to quickly pass the bipartisan  Chip Security Act , legislation he introduced earlier this year. “China recognises the superiority of...

Taiwan Semiconductor Manufacturing Co (TSMC) reported a slowdown in monthly sales growth for October, reigniting debate over whether the artificial intelligence (AI) boom driving global chip demand is beginning to lose steam. The world’s largest contract chipmaker posted a  16.9% year-on-year revenue increase  last month — its  slowest pace since February 2024 . Analysts expect TSMC’s overall sales to rise around  16% in the current quarter , still a solid expansion but notably slower compared to recent surges. Despite the moderation,  TSMC shares have climbed about 37% year-to-date , supported by relentless AI-related demand from its biggest customers. Market Correction Fears Resurface Last week’s sudden selloff across Asian tech stocks has rattled investor confidence, underscoring growing caution that the record-breaking rally in AI and semiconductor counters may be peaking. Several Wall Street executives have warned of an impending correction, with  Mich...

US President  Donald Trump  said he did not discuss authorising exports of  Nvidia Corp’s Blackwell AI chips  to China during his meeting with President  Xi Jinping , cooling investor hopes that the powerful processors might be cleared for sale to the world’s largest semiconductor market. Blackwell Chips Off the Table Speaking aboard Air Force One, Trump said he and Xi discussed  Nvidia’s access to China in general , but confirmed that  Blackwell accelerators were not part of the conversation . “We did discuss chips,” Trump said. “They are going to be talking to Nvidia and others about taking chips. But we’re not talking about the Blackwell.” The clarification followed Trump’s earlier remarks suggesting he might revisit US export restrictions, which briefly fuelled market optimism and helped Nvidia shares rally — pushing its market value to  US$5 trillion  for the first time. Policy Stance: No Change in AI Chip Controls US Trade Representativ...

Advanced Micro Devices (AMD.US) skyrocketed  24%  on Monday, sparking a surge in  bullish call-option activity worth millions of dollars after unveiling a  multi-year AI chip-supply deal with OpenAI  — one that could generate  tens of billions in revenue  for the chipmaker. AMD shares closed at  US$203.71 , after touching an intraday high of  US$226.71 , marking its biggest one-day gain in over a year. The rally triggered aggressive options buying, including a  US$2.27 million block trade  for calls with a  US$240 strike price  expiring in four days — more than  22 times  existing open interest. Under the deal,  OpenAI  will deploy  6 gigawatts  of AMD’s Instinct GPUs beginning in  2H 2026 , while AMD issued a  warrant for up to 160 million shares  tied to delivery milestones. The first tranche vests upon shipment of the  initial 1GW of MI450 chips . AMD CFO  Jean Hu ...

 Key Takeaways Stock surge : Baidu shares jumped  16% in Hong Kong , adding  US$6.4b in market value , biggest gain since March 2022. Analyst upgrades : Arete flipped from  sell  to  buy ; Citi and Goldman also turned bullish on Baidu’s AI growth prospects. Chip & cloud catalysts : Mass production of Baidu’s Kunlun AI chips and new China Merchants Group partnership expected to boost revenue. AI model upgrade : New Ernie X1.1 shows “significant improvements,” even surpassing some DeepSeek benchmarks. Baidu Rides AI Hype Baidu’s shares surged on Wednesday as analysts highlighted the company’s  AI chip and cloud opportunities . The rally came after  Arete Research  upgraded Baidu from  sell  to  buy , reversing its lone bearish stance since May. Options Activity Jumps Investor enthusiasm spread to the derivatives market, with  170,000 contracts traded  — more than triple the 20-day average — showing bets on further u...

 Key Takeaways Alibaba secures China Unicom as client  for its  T-Head AI accelerators , marking early adoption of its in-house chips. Data center deployment:  Chips will power Unicom’s new facility in Qinghai, alongside rivals MetaX and Biren. Competitive edge:  Alibaba’s chip reportedly offers  better memory specs than Huawei’s Ascend 910B , though Huawei is readying an upgraded 910C. Strategic investment:  Alibaba committing  ¥380B (US$53.5B)  over three years into AI infrastructure, aiming to reduce reliance on Nvidia. Deal Significance The deployment at China Unicom’s Sanjiangyuan data center highlights Alibaba’s progress in the domestic semiconductor race. While the scale of adoption remains unclear, state media coverage signals government support and growing recognition of Alibaba’s T-Head unit as a credible player in China’s AI ecosystem. Competitive Landscape Huawei Technologies:  Dominates with Ascend series but faces intensify...

  Beijing’s Tech Push Gains Momentum China’s tech giants Alibaba Group and Baidu Inc have begun deploying their own internally designed chips to train artificial intelligence (AI) models, marking a strategic shift away from heavy reliance on Nvidia processors, according to a report by  The Information . Alibaba and Baidu’s In-House Efforts Alibaba : Has used its self-designed chips since early 2025 to train smaller-scale AI models. Baidu : Is testing its Kunlun P800 chip to train updated versions of its flagship Ernie AI model. Both companies are still deploying Nvidia’s hardware for their most advanced AI projects, but the gradual shift underscores the effort to reduce vulnerability to US export restrictions. Why It Matters The move comes amid escalating US curbs on the export of advanced AI semiconductors to China, which have disrupted access to Nvidia’s latest processors. In response, Beijing has stepped up pressure on domestic firms to accelerate adoption of homegrown tech...

  Key Takeaway: Broadcom’s surprise $10B AI chip order — likely from OpenAI — has reignited the “ASIC vs GPU” debate. While Nvidia faces pressure on market share, Taiwan Semiconductor looks set to win no matter which side dominates, and storage plays like Seagate are quietly becoming star performers. Broadcom’s Big Win Broadcom stock surged nearly 10% after earnings, powered by news that one of its custom AI chip projects has converted into a major qualified customer — widely believed to be OpenAI. This deal is expected to add $10B in revenue from FY26, with demand stretching into FY27. The ASIC vs GPU Equation Nvidia’s GPUs deliver ~75–80% gross margins, but they’re costly. According to TD Cowen, custom ASICs can yield higher returns: Nvidia GPU: ~24% IRR ASIC at 70% Nvidia performance: ~35% IRR ASIC at 55% performance: breakeven ~24% IRR This explains why hyperscalers like OpenAI, Meta, and others are exploring ASICs. However, the risk lies in scalability: custom chips must keep ...

Targeted Tracking to Enforce Export Controls US authorities have reportedly deployed location-tracking devices in select shipments of advanced AI chips to detect and prevent illegal diversions to China, according to Reuters. The covert measures apply only to shipments under investigation and focus on high-risk exports restricted under US trade rules. Industry sources told Reuters that trackers have been found hidden in server shipments from  Dell Technologies  and  Super Micro Computer , containing chips made by  Nvidia  and  AMD . These devices are reportedly placed both externally on packaging and internally within server components. High-Stakes Enforcement Amid Relaxed Sales Terms The program reflects Washington’s intensified enforcement of semiconductor export restrictions to China, even as the Trump administration has allowed limited AI chip sales under new revenue-sharing arrangements. In one 2024 case, Dell servers carrying Nvidia chips were equipped...

Covert Monitoring of High-Risk Shipments US authorities have been secretly placing location-tracking devices in certain shipments of advanced AI chips deemed at high risk of illegal diversion to China, Reuters reported, citing sources with direct knowledge. The trackers, concealed in packaging or even inside servers, aim to detect chips being rerouted to destinations under US export restrictions. These measures target select shipments already under investigation and are intended to build enforcement cases against individuals or companies violating export controls. Application Across the AI Supply Chain The practice has been observed in shipments from server makers such as Dell and Super Micro that include Nvidia and AMD chips. In some 2024 cases, both large smartphone-sized trackers and smaller hidden devices were reportedly used. Sources say the Bureau of Industry and Security (BIS) — which enforces export controls — often leads the effort, with Homeland Security Investigations (HSI) ...

Advanced Micro Devices (AMD)  shares climbed over  4% to $173.14  after UBS boosted its price target to  $210  from $150, citing rising confidence in AMD’s  AI semiconductor business . Key Points UBS Analyst Timothy Arcuri:  Reaffirmed Buy rating, forecasting  21% upside  and stronger-than-expected Q2 results driven by PC and server demand. AI Chip Momentum:  AMD’s new  Instinct MI350X and MI355X  GPUs could fetch around  $25,000 per unit , exceeding market expectations. Revenue Outlook:  Arcuri estimates AMD’s data center GPU business could hit an  annualized $10B  by year-end. Strategic Edge AMD’s acquisition of  ZT Systems  in March adds rack-scale server design capabilities, allowing AMD to deliver  full-stack AI systems  competing with NVIDIA’s end-to-end solutions. The AI market’s pivot to rack-scale infrastructure is a key driver behind the bullish outlook. Market Context AMD stock...

At least US$1 billion worth of banned Nvidia AI chips  have reportedly made their way into China through unauthorized channels in the months following tighter U.S. export restrictions, according to the  Financial Times . The smuggled chips include  high-end B200 processors , which are barred from sale in China under U.S. national security controls. Key Highlights: Banned but available : Despite restrictions,  Nvidia’s B200, H100, and H200 chips  are widely available in  China’s gray market , with multiple distributors from  Guangdong, Zhejiang, and Anhui  provinces involved. Black market surge : The chips reportedly surfaced in  May  and were sold to  data center suppliers serving Chinese AI firms , even as U.S. policy aimed to cut China off from advanced AI hardware. Southeast Asia's role : Countries in  Southeast Asia , including  Thailand , have become key  transit points  for these restricted chips, according...

At an AI-focused event in Washington,  AMD CEO Lisa Su  confirmed that  chips produced by TSMC in Arizona will cost 5% to 20% more  than those made in Taiwan. AMD expects to receive its  first batch from the Arizona fabs by end-2025 . Cost Implication: Su’s comments underline  ongoing concerns about the economics of US chip manufacturing , even with government incentives. The  cost premium  could impact pricing models for chip buyers—especially in AI and data center segments where AMD competes with Nvidia. AI Market Outlook: Su noted that  AI chip demand is booming , fueled by investments from  OpenAI’s Sam Altman  and  xAI’s Elon Musk . She forecasted the  AI chip market could surpass $500B  in the coming years, driven by rapid adoption of  accelerators  for model training and inference. Analyst Takeaway: While reshoring chip production boosts  supply chain resilience ,  cost pressures remain real...

Shares of  LG Electronics Inc  jumped in Seoul after reports surfaced that the company is diving into the  AI chip supply chain  — specifically in the development of  hybrid bonders  for  high-bandwidth memory (HBM)  chips. What’s the Buzz? Local media reports  LG is working on  tools essential to stack memory chips  used alongside AI processors like those from Nvidia. Targeted  mass production  of hybrid bonders is  set for 2028 , though LG has yet to confirm the timeline. Hybrid bonders are  vital for HBM manufacturing , enabling thinner, faster memory stacks for AI workloads. Market Ripple Effect LG Electronics stock  rose on the news. Meanwhile,  existing players  in the bonder space —  Hanmi Semiconductor (-6.5%) ,  Hanwha Vision (-4.7%) , and  Samsung Electronics (-1.3%)  — saw declines as investors reacted to potential new competition. Analyst Take “Entry barrier is high,...