Actualités financières en direct

The economy, inflation and how those forces could impact the lives of Americans were front and center over the past week. Trips to the grocery store and gas station are more painful than they …

The economy, inflation and how those forces could impact the lives of Americans were front and center over the past week. Trips to the grocery store and gas station are more painful than they were last year, and rising costs are impacting the decisions of bot… The economy, inflation and how those forces could impact the lives of Americans were front and center over the past week. Trips to the grocery store and gas station are more painful than they were la…

Altman is holding out for a $1 trillion valuation, pushing OpenAI's public debut to 2027 at the earliest Sam Altman wants OpenAI to be worth $1 trillion before it touches the public markets. That’s …

Altman is holding out for a $1 trillion valuation, pushing OpenAI's public debut to 2027 at the earliest Sam Altman wants OpenAI to be worth $1 trillion before it touches the public markets. That’s the number. Anything below it is, in his view, a nonstarter. The OpenAI CEO has effectively killed any prospect of the company going public this year, calling a 2026 IPO ill-advised despite the company having already filed its S-1 registration statement with the SEC back in June. The new target is 2027, with a caveat that it could come sooner if the business environment cooperates. Gloria Bull case, bear case, or the catalyst that breaks it — Gloria Finance runs the whole research in one click. Run a playbook → A trillion-dollar threshold OpenAI’s most recent private valuation landed at roughly $852 billion as of March 2026. That’s an extraordinary number for a company that’s still posting losses. But it’s not $1 trillion, and for Altman, that gap matters. Advertisement Internal discussions shifted over the summer. Advisers and executives began recommending a delay, pointing to volatile market conditions and the performance of other high-profile tech IPOs as reasons to pump the brakes. SpaceX’s post-IPO stock volatility, in particular, served as a cautionary example that apparently resonated with OpenAI’s leadership team. On August 19, CFO Sarah Friar addressed employees directly, confirming that OpenAI would target a public offering in 2027. She left the door open by noting it could happen sooner if business conditions showed significant improvement, but the message was clear: 2026 is off the table. The math behind the patience The gap between $852 billion and $1 trillion is about $148 billion, roughly 17% growth from the last private round. For most companies, that would be a multi-year aspiration. For OpenAI, which has been roughly doubling its valuation with each fundraising cycle, closing that gap in the next several months isn’t unreasonable. What this means for the IPO landscape The conventional market approach OpenAI is taking, relying on traditional IPO mechanics rather than alternative listing structures, suggests the company wants a clean, straightforward public debut. No SPACs, no direct listings, no exotic financial engineering. Friar’s messaging to employees also serves an internal purpose. Stock-based compensation is a major retention tool for AI companies competing for elite engineering talent. Giving employees clarity about a 2027 timeline, rather than letting uncertainty linger, helps manage expectations and reduces the risk of key departures driven by frustration over liquidity timelines.

The Oracle co-founder's first major share sale this century raises questions about timing, restructuring costs, and where the company is headed. Larry Ellison has spent most of the 21st century not selling Oracle stock. …

The Oracle co-founder's first major share sale this century raises questions about timing, restructuring costs, and where the company is headed. Larry Ellison has spent most of the 21st century not selling Oracle stock. That changed on June 22, 2026, when he adopted a Rule 10b5-1 trading plan that would allow him to offload up to 50 million shares of Oracle through October 24, 2026. When the plan was filed, those shares were worth roughly $175 each, putting the potential total near $8.75B. Oracle’s stock has since pulled back to around $150, trimming the maximum expected proceeds to approximately $7.5B. Gloria You may own more of this than you think. Gloria Finance looks through your funds and identifies your hidden true exposure. Check your exposure → Why this is unusual To understand why this matters, some context on Ellison’s track record: the largest single share sale he has made this century was 25,000 shares. His preferred move has been to borrow against his holdings rather than convert them to cash, a common strategy among ultra-wealthy founders who want liquidity without triggering a taxable event or sending a signal to the market. Advertisement This plan involves up to 50 million shares. That is a very different kind of number. Even after any sales under the plan, Ellison would retain approximately 1.1 billion shares, keeping his ownership stake at over 40% of Oracle. A Rule 10b5-1 plan is essentially a pre-scheduled selling program, set up in advance so that executives can sell stock at predetermined prices or on a fixed schedule without running afoul of insider trading rules. Once the plan is in place, the trades can execute automatically, even if Ellison is deep inside a sensitive product decision or a board meeting about unreleased financials. The Oracle backdrop is complicated right now The timing of the disclosure lands alongside some notable turbulence inside Oracle itself. The company announced an additional $700M in restructuring and severance costs, driven primarily by its pivot toward AI infrastructure and the workforce reductions that come with reorganizing around new priorities. The share price tells part of that story. Oracle reached peaks in 2025, and the stock has retreated from those highs. Investor concerns about spending levels and balance sheet leverage have applied steady downward pressure, which means Ellison’s 50 million shares are now worth materially less than they were when he filed the plan. That gap matters because a 10b5-1 plan does not guarantee sales. It establishes conditions under which sales may occur. If Oracle’s stock continues sliding, some of those conditions may never trigger, and the full $7.5B to $8.75B in potential proceeds may never materialize.

A study of over 2 million product listings found that Google's AI-powered shopping results consistently surface pricier options with far fewer choices than traditional search. If you’ve been using Google’s AI Mode to shop, …

A study of over 2 million product listings found that Google's AI-powered shopping results consistently surface pricier options with far fewer choices than traditional search. If you’ve been using Google’s AI Mode to shop, you might want to check your receipts. A study by Productrise analyzing over 2 million product listings found that prices displayed in AI Mode averaged 21.6% higher than identical items surfaced through traditional Google Search. The analysis, which tracked 100,000 search and AI responses over 23 days from August 9 to August 31, 2026, paints a picture of two very different shopping experiences living under the same Google roof. Gloria Wall Street's edge, now anyone's — Gloria Finance is your agentic-AI investment research terminal. Get the edge → The numbers tell a striking story Traditional Google Search returned an average of 27.8 products per query. AI Mode? Just 3.9. That’s roughly seven times fewer options, which already limits a shopper’s ability to comparison-shop before clicking “buy.” Advertisement The overlap between the two experiences was almost nonexistent. Only 1.28% of products that ranked in traditional search also appeared in AI Mode for the same query on the same day. So it’s not just that AI Mode is showing you the same products at higher prices. It’s showing you almost entirely different products. When the study did find matching products across both surfaces, the pricing gaps were hard to ignore. About 38.1% of matched items showed price differences between the two modes. Of those disagreements, AI Mode was the more expensive option 68.4% of the time, carrying a median price premium of 22.2%. Zooming out to the full product universe each mode surfaced, the median product price in AI Mode clocked in at $149, compared to $100 in traditional search. That’s a 49% gap in the typical price point shoppers encounter. Even the merchants differed. The primary seller listed first for matched products changed nearly half the time, with a 49.6% variance rate between AI Mode and traditional search. Same data, different results Google’s response to the findings was straightforward: both AI Mode and traditional search pull from the same Shopping Graph, the company’s massive product database that aggregates listings across retailers. Shoppers can click through any listing to compare prices from multiple sellers.

The third investigation was approved in July and tied to suspected insider trading involving Google's 2025 Year in Search ranking. The Commodity Futures Trading Commission has launched at least three previously unreported investigations into …

The third investigation was approved in July and tied to suspected insider trading involving Google's 2025 Year in Search ranking. The Commodity Futures Trading Commission has launched at least three previously unreported investigations into suspected insider trading on Polymarket, according to a recent report from WIRED. The probes cover event contracts linked to pardons granted by former President Joe Biden, Iran and Google’s 2025 Year in Search rankings. Gloria Stocks in one app, crypto in another, wallets everywhere? Gloria Finance brings your whole portfolio into one view. See it all together → CFTC Chairman Michael Selig approved the first investigation in early May, which followed an NPR report about a trader who earned more than $300,000 from pardon-related markets after correctly predicting several preemptive pardons. Advertisement A second investigation, approved at the end of May, involved Iran-related contracts and followed a 60 Minutes report on accounts that reportedly earned $2.4 million with a 98% win rate. In July, the CFTC approved a third investigation into suspected insider trading involving Google’s 2025 Year in Search ranking. An agency official said the probe would examine additional individuals and noted that the Southern District of New York was conducting a parallel investigation. CFTC enforcement officials said the investigation was separate from an existing case against former Google engineer Michele Spagnuolo, who allegedly made more than $1.2 million trading Polymarket contracts based on confidential information about Google’s 2025 Year in Search rankings. Polymarket has been focusing on rebuilding its reputation in the US after four years in exile. In late 2025, the company relaunched in the US following the acquisition of QCEX, giving Americans limited access to its event contracts under CFTC oversight. The Justice Department, alongside the CFTC, previously examined whether the crypto prediction market circumvented restrictions on US traders imposed under a 2022 settlement with the CFTC. The authorities ended their probe last July.

Ukraine peace deal with Russia before 2027 Ukrainian President Volodymyr Zelenskyy has expressed his willingness to meet with Russian President Vladimir Putin at the upcoming G20 summit in Miami this December, according to reports …

Ukraine peace deal with Russia before 2027 Ukrainian President Volodymyr Zelenskyy has expressed his willingness to meet with Russian President Vladimir Putin at the upcoming G20 summit in Miami this December, according to reports from RPP Noticias. This development comes amid the ongoing high-intensity conflict between Russia and Ukraine, which began with Russia’s large-scale invasion in February 2022. The potential meeting at an international summit suggests renewed diplomatic efforts, despite the ongoing military confrontations and lack of formal negotiations between the two countries. This dynamic could influence the prediction markets focused on a potential peace deal or ceasefire agreement between Ukraine and Russia. Advertisement Key Takeaways Zelenskyy’s readiness to engage with Putin at the G20 summit appears to suggest a potential opening for diplomatic dialogue. The peace deal market pricing, currently at 9.5% YES, may indicate cautious optimism about possible negotiations before 2027. The ceasefire agreement market shows a slight increase to 22.5% YES, reflecting the potential impact of leadership discussions. What to Watch Market participants will closely monitor any official announcements or developments leading up to the G20 summit, particularly statements from both Ukrainian and Russian officials that could suggest shifts in diplomatic stances. Observers will also look for any international mediation efforts or changes in military activity that could indicate progress towards peace. The upcoming months will be critical in determining whether this diplomatic engagement will translate into tangible agreements or merely remain speculative. Get live prediction-market analysis, powered by Vera. Sign up for Vera.

The AI giant's rise from $100 billion to $852 billion tells a story of explosive growth, but secondary markets and rising competition hint at turbulence ahead Two years ago, OpenAI was worth roughly $100 …

The AI giant's rise from $100 billion to $852 billion tells a story of explosive growth, but secondary markets and rising competition hint at turbulence ahead Two years ago, OpenAI was worth roughly $100 billion. Today, its post-money valuation sits at $852 billion following a massive funding round that closed on March 31, 2026. That’s a 766% increase, the kind of number that makes even seasoned venture capitalists do a double-take. The funding milestones that got here OpenAI’s valuation trajectory reads like a hockey stick drawn by someone with a very steady hand. In early 2024, the company was valued in the $86 billion to $100 billion range. By October 2024, that figure had climbed to $157 billion. Gloria Stocks in one app, crypto in another, wallets everywhere? Gloria Finance brings your whole portfolio into one view. See it all together → Then came the real fireworks. A $122 billion funding round, one of the largest private capital raises in history, pushed the valuation to $852 billion by the end of March 2026. Advertisement The investors lining up weren’t exactly unknowns. Microsoft and Nvidia, both deeply embedded in the AI infrastructure stack, contributed to the momentum. On the revenue side, OpenAI’s annualized run rate has been estimated between $24 billion and $40 billion as of mid-2026, with monthly revenues reportedly hitting around $2 billion earlier in the year. The company remains unprofitable at scale. OpenAI reportedly achieved a 17x increase in computing power over the same two-year window. Secondary markets tell a different story While the primary valuation from the March 2026 round pegged OpenAI at $852 billion, secondary market signals have been less enthusiastic. By September 2026, composite marks from secondary trading suggested an implied valuation closer to $475 billion. That’s a 44% discount to the primary round price. OpenAI confidentially filed for an IPO in June 2026, with a potential public listing targeted for 2027. At the primary valuation, OpenAI trades at roughly 30x to 35x its revenue run rate. Anthropic enters the chat Anthropic’s valuation reached $965 billion, and its annual revenue run rate hit $47 billion, both exceeding OpenAI’s figures in 2026.

というSSRN論文をタイラー・コーエンが紹介している。原題は「Technological Unemployment in Victorian Britain: Young Workers and the Collapse of Entry」で、著者はH. G. Vipond（LSE）。 以下はその結論部。 Creative destruction has been ongoing for more than two centuries. In that time millions of jobs have been made …

というSSRN論文をタイラー・コーエンが紹介している。原題は「Technological Unemployment in Victorian Britain: Young Workers and the Collapse of Entry」で、著者はH. G. Vipond（LSE）。 以下はその結論部。 Creative destruction has been ongoing for more than two centuries. In that time millions of jobs have been made obsolete, and have disappeared. On the flip side, much of the work performed today is new: around 60 percent of employment in the United States is now in job titles that did not exist in 1940 (Autor et al., 2024). This underscores the degree to which maintaining employment has relied on the creation of new work. At the dawn of the Industrial Revolution, there was acute concern that the new machines would steal jobs. This concern abated considerably over time. Did we destroy so many jobs without causing substantial damage? If so- how? The usual answer is that new jobs were created at least as fast as old jobs were lost. What is less clear is how workers were reallocated out of waning and into emerging work. This paper follows a major technological transition from onset through to completion. As English bootmaking mechanized between 1851 and 1911, approximately 153,000 artisanal jobs disappeared and 140,000 new, more specialized jobs emerged. Incumbent workers did not simply move from the old jobs into the new ones. Nor, for the most part, were they displaced. Male incumbents experienced very little displacement and continued in the traditional bespoke work for which they had trained, although the experience of female incumbents was markedly different. Instead, adjustment operated primarily through young cohorts. On the destruction side, as mechanization drove a decline in demand for artisanal labour, there was a collapse of entry into the profession. Young men became substantially less likely to enter, and would have borne meaningful costs had they persisted. On the creation side, the new and more specialized jobs were also taken up predominantly by young workers, and overwhelmingly by those born in the counties in which the new work emerged. Adjustment on both sides of the transition- destruction and creation- therefore operated via young cohorts. However, while bootmaking became a less viable option for young people across most of England, the new opportunities were concentrated in only a few counties. If technological change generated inequalities in this transition, it did so primarily by tilting young people’s access to opportunity. I propose we consider two types of technological obsolescence. When technological change is abrupt, and demand for old skills disappears in concert, adjustment falls on incumbents. They are displaced. However, when change is gradual, and demand for older skills persists, incumbents can remain while entry erodes. The pace of diffusion and the persistence of demand for older skills determine which margin bears the adjustment. Finally, the impact of technological change on workers, their families, and their communities cannot be understood by reading from the destruction side of the ledger alone. It matters a great deal how many new jobs emerge, where they emerge, and who is placed to take them. The incidence of labour displacing technological change is jointly determined by the forces of both job loss and job creation. In the modern day, understanding the labour-market and political-economy consequences of generative AI requires us to ask not only which jobs will disappear, but what new work will emerge, where, and for whom. （拙訳） 創造的破壊は2世紀以上に亘って続いている。その間に何百万という職が時代遅れになり、消滅した。それと裏腹に、今日行われている仕事の多くは新しいものである。米国の雇用者の約6割は今、1940年代には存在しなかった職種に就いている（Autor et al., 2024*1）。このことは、雇用の維持が新たな仕事の創造にどの程度懸かっているかを明らかにしている。産業革命の黎明期には、新しい機械が職を奪うという深刻な懸念があった。この懸念は時間と共に大きく低下した。我々は、あれだけ多くの職を、相当な損害を引き起こすこと無しに破壊したのだろうか？ もしそうならば、どのようにして？ 通常の回答は、少なくとも古い職が失われるのと同じ速度で新たな職が作られた、というものである。 それよりも明らかでないのは、労働者が衰退する仕事から新しく生まれた仕事にどのように再配分されたか、である。本稿は、主要な技術的遷移を開始から終了まで追った。英国の靴製造は1851年から1911年に掛けて機械化され、およそ153,000の職工の職が消滅し、140,000の新たなより専門化された職が登場した。既存の労働者が単純に古い職から新しい職に移動したわけではない。彼らは大体において、置き換えられたわけでもない。男性の既存の労働者はほとんど置き換えを経験せず、自分が訓練を受けた昔ながらの注文品の仕事を続けた。ただし、既存の女性労働者の経験はかなり異なるものだった。 代わりに、主として若年層のコホートを通じて調整が行われた。破壊側では、機械化によって注文品についての労働需要が低下したことで、この職業への参入は崩壊した。若い男性が就職する可能性は大きく低下し、それが持続していたならば大きな対価が生じていたであろう。創造側では、新たなより専門的な職に就いたのはやはり主として若年労働者で、しかも新しい仕事が登場した国の出身者が大多数だった。従って、遷移の両面――破壊と創造――における調整は、若年層のコホートを通じて働いた。ただ、英国の大半において靴製造が維持可能な選択肢ではなくなっていった一方で、新たな機会は少数の州に集中していた。この遷移における技術変化が格差を生じたとすれば、それは主として、若い人の機会へのアクセスを偏らせることによってであった。 2種類の技術的な衰退を考慮することを提案したい。技術変化が急で、古い技能への需要の消滅がそれと軌を一にする場合は、調整は既存の労働者に降り掛かる。彼らは置き換えられるのだ。しかし、変化が徐々に進み、古い技能への需要が持続する場合は、参入が無くなっていく一方で既存の労働者は職に留まることができる。普及の速度と古い技能への需要の持続性が、どちらが調整を担うかを決める。 また、技術変化が労働者とその家族ならびに共同体に与える影響は、台帳の破壊側だけを見ていては理解できない。どれだけ新しい仕事が生じるか、それがどこに生じるか、それに就くのはだれか、が非常に重要になる。労働を置き換える技術変化の影響は、職の喪失と職の創出の両方の力によって同時決定される。今日では、生成AIの労働市場と政治経済への影響を理解するためには、どの職が消滅するか、だけでなく、どのような新しい職がどこで誰のために生じるか、を問う必要がある。

Anthropic CEO Dario Amodei has urged the AI industry to decelerate its capability advancements to ensure safety measures can keep pace. The San Francisco-based AI lab, known for its flagship model Claude, is emphasizing …

Anthropic CEO Dario Amodei has urged the AI industry to decelerate its capability advancements to ensure safety measures can keep pace. The San Francisco-based AI lab, known for its flagship model Claude, is emphasizing a safety-driven approach by involving third-party evaluators throughout AI training processes. This move aligns with Anthropic’s strategy to position itself as a safety-focused alternative in the competitive AI landscape, involving tech giants like OpenAI. The announcement comes amidst Anthropic’s substantial $65 billion funding round, which placed the company among the highest-valued AI startups globally. Advertisement Key Takeaways The call to slow down AI capability advancements suggests potential delays in product development, which could impact Anthropic’s valuation trajectory. Market pricing appears supportive of a scenario where Anthropic’s valuation may not reach $600 billion by December 31, 2026. Anthropic’s emphasis on safety and alignment processes may affect its competitive posture in the AI race, potentially influencing investor perception. What to Watch Observers should look for any announcements regarding new funding rounds or strategic partnerships from Anthropic, as these developments could shift market sentiment and valuation expectations. Further statements from Dario Amodei on this strategic direction will be closely monitored, as will any reactions from major investors like Amazon and Google. The impact of these safety initiatives on Anthropic’s competitive edge, especially against rivals like OpenAI, will be a key indicator in the coming months. Get live prediction-market analysis, powered by Vera. Sign up for Vera.

XRP (CRYPTO: XRP), a payments coin developed by Ripple, is one of the most successful cryptocurrencies. It's the fifth largest at the time of this writing (Sept. 8) and has a market cap of …

XRP (CRYPTO: XRP), a payments coin developed by Ripple, is one of the most successful cryptocurrencies. It's the fifth largest at the time of this writing (Sept. 8) and has a market cap of $89 billion. It has also gotten much cheaper recently, trading at abou… XRP (CRYPTO: XRP), a payments coin developed by Ripple, is one of the most successful cryptocurrencies. It's the fifth largest at the time of this writing (Sept. 8) and has a market cap of $89 billio…