From soaring fuel prices in Pakistan to a shortage of cooking gas in India, Asia is bearing the brunt of the energy crunch sparked by the Iran War. Hear how the crisis is unfolding on the Big Take Asia Podcast. (Source: Bloomberg)

Mistral AI on Monday launched Forge , an enterprise model training platform that allows organizations to build, customize, and continuously improve AI models using their own proprietary data — a move that positions the French AI lab squarely against the hyperscale cloud providers in one of the most consequential and least understood markets in enterprise technology. The announcement caps a remarkably aggressive week for Mistral, which also released its Mistral Small 4 model , unveiled Leanstral — an open-source code agent for formal verification — and joined the newly formed Nvidia Nemotron Coalition as a co-developer of the coalition's first open frontier base model. Together, these moves paint the picture of a company that is no longer content to compete on model benchmarks alone and is instead racing to become the infrastructure backbone for organizations that want to own their AI rather than rent it. Forge goes significantly beyond the fine-tuning APIs that Mistral and its competitors have offered for the past year. The platform supports the full model training lifecycle: pre-training on large internal datasets, post-training through supervised fine-tuning, DPO, and ODPO, and — critically — reinforcement learning pipelines designed to align models with internal policies, evaluation criteria, and operational objectives over time. "Forge is Mistral's model training platform," said Maliena Guy, head of product at Mistral AI, in an exclusive interview with VentureBeat ahead of the launch. "We've been building this out behind the scenes with our AI scientists. What Forge actually brings to the table is that it lets enterprises and governments customize AI models for their specific needs." Why Mistral says fine-tuning APIs are no longer enough for serious enterprise AI The distinction Mistral is drawing — between lightweight fine-tuning and full-cycle model training — is central to understanding why Forge exists and whom it serves. For the past two years, most enterpr...

Watch: Humanoid Robot Returns Tennis Shots With 96% Accuracy In Simulation Tests Authored by Atharva Gosavi via Interesting Engineering , Galbot Robotics has released a video on its official X handle on March 16 showing a humanoid robot rallying tennis shots with a human player in real time. Robot playing tennis The demonstration showcases the company’s LATENT system , developed in collaboration with researchers from Tsinghua University and Peking University. The system was tested on the Unitree G1 humanoid robot, which demonstrated the ability to respond to fast-moving balls, navigate across the court, and sustain rallies with a human opponent. “For the first time, a humanoid robot can sustain high-dynamic, long-horizon tennis rallies with millisecond-level reactions, precise ball striking, and natural whole-body motion, ” Galbot’s X post read. Teaching robots on limited movement data 🎾Your humanoid tennis player is here!🤖 Introducing LATENT (Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data) — the world’s first real-time whole-body planning and control algorithm for athletic humanoid tennis. For the first time, a humanoid robot can… pic.twitter.com/gCi38wxHVQ — Galbot (@GalbotRobotics) March 16, 2026 One of the key challenges in training robots for sports lies in the lack of accurate human movement data. This is especially true for tennis, where players cover large areas, balls can reach speeds of up to 30 m/s, and racket-ball contact lasts only a few milliseconds. To address this, the researchers avoided recording full matches. Instead, they focused on collecting short fragments of essential movements such as forehands, backhands, and side steps. The data were captured using a motion-tracking system within a compact 3×5-meter court, more than 17 times smaller than a standard tennis court. A total of five players contributed approximately five hours of recorded motion data. From basic motions to coordinated gameplay Using this dataset, the ...

GREENWICH, Conn., March 17, 2026 (GLOBE NEWSWIRE) -- Stardust Power Inc. (Nasdaq: SDST) (“Stardust Power” or the “Company”), an American developer of battery-grade lithium carbonate, today announced its preliminary results for the year ended December 31, 2025 and provided an update on the continued development of its lithium refinery project in Muskogee, Oklahoma. “2025 marked a year of meaningful progress for Stardust Power as we advanced the technical, commercial and regulatory foundations of our lithium refinery in Muskogee,” said Roshan Pujari, Founder and Chief Executive Officer of Stardust Power. “During the year, we completed key engineering milestones, including our FEL-3 study, entered into strategic feedstock supply agreements and received independent third-party validation of the refinery design. Phase I is expected to produce up to 25,000 metric tons per year of battery-grade lithium carbonate, forming the first stage of a planned 50,000 metric tons per-year refining platform, with estimated Phase I CapEx of approximately $500 million. The refinery is designed to support the development of a more secure domestic lithium supply chain. Following year-end, we also received our air quality construction permit from the Oklahoma Department of Environmental Quality, which helps position the refinery to move toward construction.” 2025 Refinery Project Milestones and Business Highlights: Completed the Front-End Loading Level 3 (“FEL-3”) engineering study by Primero for the planned lithium refinery in Muskogee, Oklahoma, providing engineering and project definition and establishing an estimated capital cost of approximately $500 million to support the next stage of development and construction planning. The Company continues to evaluate a range of project financing alternatives, including potential strategic partners, government-supported financing programs and capital markets transactions, to support the development and construction of the Muskogee lithium refine...

Most enterprise AI projects fail not because companies lack the technology, but because the models they’re using don’t understand their business. The models are often trained on the internet, rather than decades of internal documents, workflows, and institutional knowledge. That gap is where Mistral, the French AI startup, sees opportunity. On Tuesday, the company announced Mistral Forge, a platform that lets enterprises build custom models trained on their own data. Mistral announced the platform at Nvidia GTC, Nvidia’s annual technology conference, which this year is focused heavily on AI and agentic models for enterprise. It’s a pointed move for Mistral, a company that has built its business on corporate clients while rivals OpenAI and Anthropic have soared ahead in terms of consumer adoption. CEO Arthur Mensch says Mistral’s laser focus on the enterprise is working: the company is on track to surpass $1 billion in annual recurring revenue this year. A big part of doubling down on enterprise is giving companies more control over their data and their AI systems, Mistral says. “What Forge does is it lets enterprises and governments customize AI models for their specific needs,” Elisa Salamanca, Mistral’s head of product, told TechCrunch. Several companies in the enterprise AI space already claim to offer similar capabilities, but most focus on fine-tuning existing models or layering proprietary data on top through techniques like retrieval augmented generation (RAG). These approaches don’t fundamentally retrain models; instead, they adapt or query them at runtime using company data. Mistral, by contrast, says it is enabling companies to train models from scratch. In theory, this could address some of the limitations of more common approaches — for example, better handling of non-English or highly domain-specific data, and greater control over model behavior. It could also allow companies to train agentic systems using reinforcement learning and reduce reliance on t...

Australia’s market for bond sales is trailing US and European peers since the start of the Iran war, as the conflict stokes inflation and pushes local borrowing costs to some of the highest in the developed world. Sales in the Aussie-dollar syndicated bond market — a key yield source for Asia Pacific debt investors — have tumbled about 62% since the war began on Feb. 28, versus a year earlier, Bloomberg-compiled data show, denting a record start to the year. By contrast, US investment-grade issuance has jumped by a similar margin over the same period as borrowers seized on last week’s market reopening. Iran-related risks are intensifying price pressures in Australia, with policymakers responding by raising interest rates on Tuesday. Yields on the nation’s dollar corporate bonds have jumped about 40 basis points this month to their highest since late 2023, outstripping those in the US and Europe, a Bloomberg index shows, potentially further damping demand from offshore borrowers. “There was a very big pipeline, with high single digits in terms of the mandates that were out there at the start of last week,” said Craig Johnston , head of debt capital markets syndicate for Australia at Deutsche Bank. “Only a small number of these trades progressed.” Australia’s primary market has seen only a handful of corporate bond deals since the war began, missing out on the wave of multi-billion dollar offerings by US technology firms, in particular Amazon.com Inc.’s almost $54 billion-equivalent sale across dollars and euros earlier this month. Read: From Amazon to Airbnb, US Blue-Chip Bond Sales Flirt With Record In comparison, the largest Aussie-dollar deal during this period has been a A$1.3 billion ($918 million) issuance from US telecom giant Verizon Communications Inc. Government-owned broadband builder NBN Co. also priced an A$850 million note last week after issuance paused following the outbreak of the Iran war. “Activity hasn’t flipped back to business‑as‑usual yet,” sai...

格隆汇3月18日｜达飞海运：鉴于中东及近东地区的地缘政治紧张局势，紧急燃油附加费将于3月27日起调整并实施。

Joe Hendrickson/iStock Editorial via Getty Images The Schwab U.S. Large-Cap ETF ( SCHX ) launched on November 3, 2009. It is managed by Charles Schwab Investment Management, and it offers exposure to stocks within the large-cap segment of the U.S. equity market. It manages ~$62 billion and charges an expense ratio of 0.03%. When I last covered this ETF on September 14, 2025, I was more optimistic about its prospects amid a favorable monetary policy and single-name concentration. But these factors now represent meaningful headwinds due to the Iran conflict and loss of mega-cap momentum. The potential recovery of the software industry could offset them to a degree, but overall, the whole situation could result in a continuation of the flattish performance. Seeking Alpha Portfolio The weights of the top ten holdings indicate high single-name concentration. NVIDIA ( NVDA ), for instance, accounts for 6.9% of the portfolio, which makes its performance meaningfully relevant here. Apple ( AAPL ) and Alphabet ( GOOG ) ( GOOGL ) also have large weights at 6.3% and 5.2%, respectively, with Microsoft ( MSFT ) and Amazon ( AMZN ) close behind at 4.7% and 3.3%, respectively. Seeking Alpha The fund is also highly concentrated in tech stocks in general, which account for almost one-third of the portfolio, so it's worth keeping this in mind for the outlook section below. Together with Communication stocks at 10.2%, the long-duration-like growth exposure is significant here. Another thing that stands out is that a third of the fund is represented by Financials, Consumer Cyclical, and Industrials, which are cyclical sectors. Defensive sectors have a lower aggregated weight (17.5%, consisting of Health Care, Consumer Defensive, and Utilities). Seeking Alpha Performance First of all, over the years, the fund's tracking error has not been meaningful, making it a very efficient vehicle for tracking this particular index. This can be partly explained by the very competitive expense ratio ...

3月17日获悉，美国政府计划进一步放松对委内瑞拉石油行业的制裁，以在伊朗战争推高油价背景下提升原油产量。相关措施最早可能于本周公布，包括向更多外国公司发放个别许可，使其能够在不违反美国制裁的情况下参与委内瑞拉石油业务。知情人士表示，此举旨在推动委内瑞拉原油生产增长，从而缓解全球能源供应紧张局面。（央视） 风险提示及免责条款 市场有风险，投资需谨慎。本文不构成个人投资建议，也未考虑到个别用户特殊的投资目标、财务状况或需要。用户应考虑本文中的任何意见、观点或结论是否符合其特定状况。据此投资，责任自负。

当地时间3月17日，美国政府计划进一步放松对委内瑞拉石油行业制裁，以在伊朗战争推高油价背景下提升原油产量。相关措施最早可能于本周公布，包括向更多外国公司发放个别许可，使其能够在不违反美国制裁的情况下参与委内瑞拉石油业务。知情人士表示，此举旨在推动委内瑞拉原油生产增长，从而缓解全球能源（850101）供应紧张局面。(央视新闻)

Tesla Is Sued by Survivor of Deadly California Cybertruck Crash Bloomberg.com

Tesla is sued by survivor of deadly California Cybertruck crash MSN

格隆汇3月18日｜据路透：伊朗媒体证实，伊朗最高国家安全委员会秘书拉里贾尼遇害。

Pam Bondi, the US attorney general, has been formally subpoenaed to appear before a House panel to answer questions about the justice department’s handling of the investigation into Jeffrey Epstein and its release of the Epstein files. The move came amid growing criticism from lawmakers on both sides of the aisle over the justice department’s compliance with a law passed last year requiring the full release of Epstein-related files. Lawmakers have raised concerns that some disclosures exposed personal information about survivors, and have also accused the Department of Justice of improperly withholding and redacting material in violation of the law, including the names of people who may have committed crimes. James Comer, chair of the House oversight and government reform committee, wrote in the subpoena letter that there were “questions regarding the Department of Justice’s handling of the investigation into Jeffrey Epstein and his associates and its compliance with the Epstein Files Transparency Act”. Comer, a Republican congressman from Kentucky, added: “As Attorney General, you are directly responsible for overseeing the Department’s collection, review, and determinations regarding the release of files pursuant to the Epstein Files Transparency Act, and the Committee therefore believes that you possess valuable insight into these efforts.” The subpoena has requested that Bondi appear for a deposition on 14 April. It follows a 4 March vote in which five Republicans on the committee joined Democrats to approve a motion, put forward by Nancy Mace, a Republican representative of South Carolina, to authorize and issue a subpoena to Bondi. In February, Bondi clashed with Democratic lawmakers during a House judiciary committee hearing, where she defended the justice department’s handling of the Epstein files. In a statement to the Guardian on Tuesday, a spokesperson for the justice department called the subpoena “completely unnecessary”. “Lawmakers have been invited to...

Never miss an episode. Follow The Big Take Asia podcast today From soaring fuel prices in Pakistan to a shortage of cooking gas in India, Asia is bearing the brunt of the energy crunch sparked by the Iran War. On today’s Big Take Asia podcast we look at how war in the Middle East has pushed Asia to the forefront of a global crisis. Host K. Oanh Ha speaks with Bloomberg’s Faseeh Mangi about the impact of soaring fuel prices in Pakistan and Clara Ferreira Marques about how the least economically resilient countries are feeling the most pain.

Micron Technology Options Spot-On: On March 17th, 392.55K Contracts Were Traded, With 2.33 Million Open Interest 富途牛牛

Source: Bloomberg, 0:00

Lululemon athletica ( LULU ) beat top- and bottom-line expectations for the fourth quarter and exceeded estimates measuring comparable sales growth and profit margins. But shares are struggling for direction and are trading defensively on downbeat guidance, increased tariff costs, and continued weakness in its Americas segment. For the quarter ending February 1, lululemon ( LULU ) generated revenue of $3.6B, which was roughly flat from a year ago but beat expectations by $20B, as a 17% increase in revenue among its international markets was offset by 4% in the Americas. Comparable sales on a constant currency basis increased 2%, which was much better than expectations for a 1.75% contraction, while comparable sales in the Americas were a less-than-feared -2%. International comparable store sales rose 16% versus +11.4% estimates. Import tariffs and markdowns continued to weigh on the company’s profitability, resulting in a year-over-year net decrease of 550 basis points in gross margin to 54.9%, partially offset by higher pricing and lower product costs. Net income declined to $586.9M, or $5.01 per share, compared to $6.14 per share in the same quarter last year, but beat estimates of $4.78. To revive its U.S. business and regain its relevance in the highly competitive athleisure segment, lululemon ( LULU ) recently launched two new lines of workout gear called Unrestricted Power and ShowZero, the latter of which was designed to hide sweat. Interim co-CEO and CFO Meghan Frank expects these new collections as well as new product launches in 2026 will account for ~35% of the company’s product assortment, up from 23% in 2025. Accordingly, lululemon ( LULU ) expects net revenue to increase 1% to 3% in the first quarter to a range of $2.4B to $2.43B and earnings of $1.63 to $1.68 per share, both of which are below expectations of $2.48B and $2.07 per share, respectively. For 2026, the company expects to generate $11.35B to $11.5B in sales, less than $11.52B estimates. Ear...

格隆汇3月18日｜据CNBC：OpenAI正筹备上市，有望于年底前实现。

【有線新聞】國家發改革委公布新一批13個標誌性重大外資項目，總投資額134億美元，涵蓋多個製造業領域，同時首次將物流項目列入清單。 今年是國家發改委第九次更新標志性重大外資項目清單，共有13項，涵蓋電子製造、化工、汽車、機械等製造業領域，總投資134億美元。為加大對服務業的支持，當局還首次將物流項目納入清單，繼續支持生物醫藥等領域研發中心項目，推動現代服務業與先進製造業深度融合。 跟以往不同，今次投資來源地更多元，包括英國、德國、瑞士、瑞典等國家的跨國企業。發改委指將針對企業訴求開展專項行動，擴大吸引外資。國家發展改革委外資司副司長荊琴：「我們將以服務業為重點，擴大市場准入和開放領域，著力塑造吸引外資的新優勢。我們將實施好新版《鼓勵外商投資產業目錄》，引導更多外資投向先進製造、現代服務、高新技術、節能環保等領域。」 新一批項目分布在北京、上海、廣東等地，不少已開工建設，包括來自瑞士的羅氏診斷，在江蘇蘇州的擴建項目投資額30億元人民幣，目標是實現生產、實驗室和物流一體化，推動產品本地化。羅氏診斷蘇州總經理藍凱瑞：「中國市場以及整個亞太地區的需求在不斷變化，這促使我們必須繼續加大投資，擴大業務範圍。通過這次擴展，我們能更好地應對和響應市場需求。」 標誌性重大外資項目至今累計投資額達到1,080億美元。

幼稚園收生不足 議員促開放非本地幼童來港 蔡若蓮憂創造其他社會問題 To view this video please enable JavaScript, and consider upgrading to a web browser that supports HTML5 video 【有線新聞】幼稚園收生不足，有議員建議合併幼稚園和中小學校舍，以及開放學額給非本地學生。 教育界（教聯會）鄧飛：「特區政府有否研究至少開放幼兒階段、學前教育階段的學生，容許家長將小朋友帶到香港？」教育局局長蔡若蓮：「在幼稚園階段道理上沒有問題，18歲以下取得學生簽證就可以。但是當然學齡人口下降數以千計，同時幼童獨自留在香港其他方面的問題，解決了學校問題，創造了其他社會問題。」 選委會界別劉智鵬：「是否可以用一個資助形式，與上面的小學、同一個辦學團體中學，中小幼一條龍在一起？」蔡若蓮：「當幼稚園和小學在一起在校舍，管理、資助模式、配套硬件等等我們需要研究，我們不排除這個探討可能性。」