Prenchie

J’ai vu un fun fact : Si vous aviez investi 1,00 $ sur le marché boursier américain lorsqu’il a ouvert pour la première fois en 1790, aujourd’hui vous seriez mort.

Les enfants recommencent a se chamailler

Meme si tu gagnes une fortune en crypto t’es pas sorti d’affaire Le vrai cauchemar c’est pour retirer les sous. Examen du compte Preuve de l'origine des fonds Explication de l’historique des transactions Problèmes fiscaux Compte bloqué Le vrai défi c’est pour retirer. Faut pas gagner en crypto, c’est plus facile

@profpicsou Tu nages pas dans les billets ? Mais qui es tu si tu n’es pas picsou ?

Vous voyez cette courbe de $NBIS C’est aussi celle de mon portefeuille. -18/20% depuis l’ATH de mon portefeuille boursier. Je l’ai dis et je le redis. Je peux vite retomber dans les abysses.

Ça fait 2 mois que « le deal est proche » mais il est toujours en guerre

Pendant que frenchie discutent avec xi jiping et des astronautes pour savoir où investir t’en as qui demande si c’est trop tot pour vendre ou trop tard pour acheter

Pas besoin de paniquer ce n’est qu’une journée… ou alors ce n’est que la première journée…

@damnang2 J’ai mis un marque page, je reviens plus tard

Everyone keeps debating whether the right bet is optics or copper. Honestly, if you are not sure, just skip the whole debate and look at these names.

@grok @TheValueist J’ai encore moins compris mais on va faire genre que c’est clair

Prenchie_ Résumé du post TheValueist : - Core : Valide hyperscalers dans prepay muni mais pas DC direct. Financing role. - Deal : 1.2B$ forte demande, Pioneer CCA scalable. - Mécanique : Arbitrage tax-exempt pour 8-10% discount ratepayers. Risques reset. - Légal : Loi 2005, benefit clients locaux. - AI : Aide affordabilité politique. - Valorisation : +25% valeur sur 200bps, limites IA. - Bénéfices : Hyperscalers, public power, banques. Challenge privés. Nuance : Spécifique, pas universel.

CORE CONCLUSION The Alphabet-Pioneer transaction is a genuine market signal, but the economically precise interpretation is narrower than the most aggressive narrative. The transaction validates that a hyperscaler-grade obligor can be inserted into the municipal prepaid-energy market at scale, with strong buy-side demand and attractive relative funding costs. It does not validate that Alphabet directly financed electricity for a Google data center through tax-exempt municipal debt. The publicly reported deal was an upsized nearly $1.2B prepaid energy financing issued through the California Community Choice Financing Authority, underwritten by Goldman Sachs, with Alphabet acting as funding recipient and with the bonds carrying Moody’s Aa2 largely tied to Alphabet’s corporate rating; reported demand exceeded $10B from roughly 100 accounts. The key correction in the public reporting is critical: the prepaid electricity was not tied to an Alphabet data center, and Alphabet’s role was a financing role. The investment conclusion is that muni prepay is potentially important as a cost-of-capital and political-economy mechanism for AI power, but not as a universal substitute for corporate debt, project finance, data-center ABS, private credit, or regulated utility rate base. The structure is most powerful where 3 conditions coexist: a tax-exempt public load-serving entity, retail load that qualifies under service-area and customer-use rules, and a creditworthy taxable counterparty willing to monetize the tax-exempt/taxable basis by sharing savings with ratepayers. In that configuration, the transaction can create a structurally lower delivered energy cost for a public utility or CCA, while giving the taxable counterparty access to a deep tax-exempt capital pool. Outside that configuration, the scalability claims become materially less certain. The $100B incremental issuance claim should be treated as an upside scenario rather than a fully validated public-market forecast. Public reporting supports the direction of the claim, with Bond Buyer citing Morgan Stanley expectations that prepaid-energy deals may 2x to 3x in coming years and JPMorgan strategists indicating prepay issuance was up 94% year over year to $15B through the end of May. However, public reporting reviewed here does not provide a detailed, transaction-level bridge to $100B of incremental issuance. The more defensible underwriting assumption is that the category is moving from niche muni-finance product to a potentially major cross-over infrastructure funding channel, with realized scale dependent on tax-risk tolerance, municipal market capacity, guarantor supply, load-service qualification, and political acceptance. TRANSACTION READ-THROUGH The deal’s immediate market read-through is strong. A nearly $1.2B Alphabet-linked municipal prepay that produced >$10B of orders and secondary spread tightening indicates that tax-exempt investors are willing to underwrite corporate-credit-like exposure when wrapped in a muni structure and anchored by a top-tier obligor. That matters because the municipal market has historically been dominated by government, nonprofit, and essential-service credits, while this deal introduces a recognizable mega-cap technology credit into a prepay structure previously more commonly associated with bank, insurance, and commodity-trading intermediaries. The result is a new price-discovery point for hyperscaler-adjacent municipal paper, not merely a 1-off procurement transaction. Pioneer is also an important proof point because it is not a mega-utility. Pioneer Community Energy is a community-owned, not-for-profit electricity provider serving 150,000+ residential and commercial customers in parts of Placer and El Dorado County and nearby incorporated communities. A CCA of this size participating in a transaction of nearly $1.2B demonstrates that the relevant gating item is not necessarily utility scale in the traditional IOU sense, but legal eligibility, contract architecture, counterparty credit, advisor capacity, and access to conduit issuance. The transaction also has reputational and regulatory value for hyperscalers. AI data-center growth is increasingly framed as a ratepayer-affordability problem, not just a technology capex problem. The White House Ratepayer Protection Pledge, dated March 4, 2026, calls on hyperscalers and AI companies to build, bring, or buy the energy needed for data centers, pay full costs for related energy and infrastructure, negotiate separate rate structures where needed, and avoid shifting incremental costs to households. A structure that explicitly routes savings to a public utility’s ratepayers fits the current political direction better than a purely private behind-the-meter procurement model. STRUCTURAL MECHANICS The basic prepay architecture is a financing arbitrage embedded in a commodity-supply structure. A municipal conduit issues tax-exempt bonds and uses proceeds to fund an upfront prepayment for long-term electricity or gas deliveries. The taxable prepay supplier or funding recipient receives the upfront capital and, in exchange, delivers energy to the public load-serving entity at a discount to an agreed market index, contract price, or spot reference. The public entity pays as energy is delivered, rather than bearing direct recourse for the conduit debt. The economic spread comes from the taxable counterparty’s ability to use low-cost tax-exempt proceeds more advantageously than a conventional taxable funding channel, with part of that value shared through lower commodity prices. A CCA prepay overview describes the process as existing energy supply contracts being assigned to a taxable prepay supplier, municipal bonds being issued by a conduit, and the load-serving entity paying the prepay supplier at a discounted rate to achieve procurement savings. The mechanism should not be described as a frictionless free lunch. It is a federal tax expenditure being monetized through a financing structure. The gross value is split among ratepayers, intermediaries, guarantors, underwriters, swap providers, bond investors, counsel, advisors, and potentially the taxable funding recipient. The public-policy defense is that the savings are directed to public utility customers and support affordability. The critical economic question is not whether value is created versus a taxable funding alternative; value is clearly created when the tax-exempt/taxable basis is wide enough. The critical question is who captures the value and whether the distribution remains politically defensible as mega-cap technology companies become funding recipients. The often-cited 8%-10% or 8%-12% savings range is plausible but should be interpreted as a procurement-cost discount on qualifying delivered energy, not as a blanket WACC reduction across an entire AI infrastructure project. Pioneer’s prior 2023 prepay release described a 30-year bond-funded clean-electricity prepay, nearly $2.8M of annual savings for the 1st 7 years, and a historic discount range of 8%-10%. Another CCA prepay overview estimated 8%-12% annual savings on delivered power quantities, subject to market conditions. Those savings do not automatically apply to land, buildings, GPUs, networking, cooling, transmission upgrades, backup generation, tenant improvements, or merchant power exposure. The duration claim also requires precision. Prepay transactions can reference 30 years of energy deliveries, but many structures include a mandatory tender at the end of an initial 5-10 year interest-rate period, followed by refunding or remarketing at new rates. Therefore, the contractual energy-delivery framework can be long-dated while the financing spread and customer discount remain exposed to reset, remarketing, tax-exempt demand, rate levels, guarantor spreads, and basis risk. This materially weakens any claim that a muni prepay mechanically locks in a 20-30 year all-in cost advantage with no reset risk. LEGAL AND TAX FOUNDATION The statutory and regulatory basis is real but nuanced. Federal tax law generally prevents tax-exempt bond proceeds from being used to acquire higher-yielding investments, and §148 defines arbitrage bonds by reference to proceeds expected to be used directly or indirectly to acquire higher-yielding investments. The 2005 Energy Policy Act added a safe harbor for prepaid natural gas by excluding qualified natural-gas supply prepayments from investment-type property. The statutory natural-gas safe harbor ties allowable volumes to historical customer purchases within the utility’s service area and includes service-area concepts based on the most recent 5 calendar years. Electricity is more nuanced than natural gas. The 2005 Act includes an exception for qualified electric and natural-gas supply contracts in the nongovernmental output-property context, but the explicit §148 safe harbor is framed around natural gas. Project-finance tax counsel have historically noted that IRS regulations permitted prepaid gas and electricity transactions, subject to qualifying use, while Congress codified a gas-specific safe harbor and did not expressly codify electricity in the same way. The practical market conclusion is that electricity prepay transactions are financeable, but they carry a higher need for precise tax structuring, qualified-use monitoring, and legal defensibility than a simplistic reading of the natural-gas safe harbor would imply. The strongest version of the structure is therefore not “a city can issue tax-exempt bonds for any AI data center.” The stronger formulation is that a tax-exempt municipal utility or CCA can prepay for qualifying electricity or gas supply used to serve retail customers in its traditional service area, with savings passed to those customers. Structures that look primarily like subsidized financing for a private hyperscaler’s proprietary load would attract greater tax, political, and disclosure scrutiny. This distinction is central to sizing the true addressable market. AI POWER CONTEXT The timing is important because power is becoming a binding constraint on AI infrastructure. Morgan Stanley Research forecasts U.S. data-center demand could reach 74 GW by 2028, with a projected shortfall of roughly 49 GW in available power access, and expects large technology companies to commit >$1T of spending in 2025-2026, with debt markets playing a major role in financing the buildout. IEA estimates data centers consumed 415 TWh in 2024, or around 1.5% of global electricity, with consumption expected to more than 2x to around 945 TWh by 2030; it also estimates that U.S. data centers accounted for 45% of global data-center electricity consumption in 2024 and will represent nearly 50% of U.S. electricity demand growth through 2030. The U.S. power-demand backdrop is no longer theoretical. EIA projections reported on June 9, 2026 indicate U.S. power demand reached a record 4,195B kWh in 2025 and is expected to rise to 4,271B kWh in 2026 and 4,397B kWh in 2027, driven by AI data centers and electrification. Commercial electricity demand is expected to exceed residential demand in 2026 for the 1st time on record. This creates a structural opening for financing models that simultaneously fund incremental energy supply, insulate residential bills, and create politically acceptable pathways for large-load interconnections. The relevance to AI infrastructure is therefore indirect but material. AI infrastructure economics are increasingly determined by power availability, power price, interconnection timing, and the capital structure used to secure energy and capacity. Muni prepay does not solve transformer shortages, gas-turbine lead times, transmission queues, local water restrictions, or zoning opposition. It can, however, improve the affordability and political acceptability of power procurement where a public load-serving entity is the counterparty and where the structure demonstrably returns value to ratepayers. VALUATION AND COST-OF-CAPITAL IMPLICATIONS The valuation sensitivity in the original thesis is directionally correct. $800M of stabilized NOI capitalized at 10.0% implies $8.0B of asset value; the same NOI capitalized at 8.0% implies $10.0B. A 200 bps reduction in the discount or capitalization rate creates a 25% uplift before considering leverage, tax, development risk, residual value, working capital, or reinvestment needs. This is why any durable 150-400 bps capital-cost advantage would be economically material for long-duration contracted infrastructure. The key caveat is asset-duration matching. The 200 bps math is most applicable to contracted, stabilized, essential-service-like cash flows with long tenors, high-credit tenants, low churn, and predictable maintenance capex. It is less directly applicable to neocloud/HPC platforms where GPU obsolescence, customer concentration, utilization volatility, software stack differentiation, collateral depreciation, and refinancing cadence can dominate the spread benefit. Current AI infrastructure financing examples show wide dispersion: CoreWeave priced $1.75B of 9.750% senior unsecured notes due 2031 in April 2026, while separately closing an $8.5B investment-grade GPU-backed financing with an A3/A(low) rating, including a fixed-rate tranche at approximately 5.9%. That spread segmentation supports the central thesis that financing structure and collateral quality materially affect equity value. Muni prepay could pressure private-credit and data-center ABS spreads at the margin, but only for the subset of energy-linked financing that can migrate into public-power structures. It will not displace construction loans for campuses, GPU-backed debt, lease-backed securitizations, or hyperscaler corporate bonds. Hyperscale data-center construction financings are still underwritten on tenant credit, lease cash-flow quality, lease risk allocation, developer track record, and construction counterparty financeability; project-level construction debt typically covers the construction-to-operational phase and then transitions to permanent financing once cash flows stabilize. WHO BENEFITS For Alphabet and other mega-cap hyperscalers, the structure offers 3 benefits: incremental funding diversity, political signaling around ratepayer savings, and potential access to a tax-exempt investor base not fully captured by corporate bond markets. Near-term EPS impact is likely modest for Alphabet given the company’s scale, but the strategic value is not trivial. Hyperscalers face a multi-year capex cycle in which every credible funding sleeve matters: unsecured bonds, project finance, vendor financing, JV capital, sale-leasebacks, private credit, utility tariffs, and now potentially prepay-related muni structures. The most valuable feature may be narrative control: participation in a structure that lowers public power costs is politically superior to being perceived as imposing grid costs on households. For public power and CCAs, the structure can be a competitive weapon. Public power utilities serve 55M+ Americans in 2,000 communities across 49 states, representing roughly 1 in 7 U.S. electricity customers. These entities already use tax-exempt municipal bonds for infrastructure and are structurally designed to pass benefits back to communities rather than maximize shareholder earnings. A financing mechanism that lowers procurement costs can improve rate competitiveness versus IOUs and retail energy alternatives, especially in regions facing data-center-driven load growth. For banks, commodity desks, and muni underwriters, the transaction is an attractive fee-and-spread opportunity. Goldman’s position as underwriter, J. Aron’s role in assigned electricity contracts, and the broader historical role of bank commodity units in prepay structures suggest that large financial intermediaries can monetize structuring expertise, swap capacity, remarketing relationships, and balance-sheet credibility. The largest incremental opportunity accrues to firms that can coordinate tax counsel, commodity trading, municipal distribution, hyperscaler relationships, and project-finance execution inside 1 integrated platform. For muni investors, the structure introduces desirable spread and recognizable obligor quality, but also changes the risk composition of the municipal market. Alphabet-linked Aa2/AA+ exposure is likely attractive relative to traditional California municipal alternatives, especially at wider spreads. However, widespread adoption would increase corporate-credit correlation inside tax-exempt portfolios. The sector could become less about traditional public essential-service risk and more about corporate obligor selection, guarantor ratings, commodity-supply mechanics, and tax-structure durability. Bond Buyer’s coverage explicitly noted investor questions about whether prepaid deals should be differentiated by underlying obligor as the sector evolves. For private data-center developers, neoclouds, and HPC operators, the implication is more threatening than immediately disruptive. For-profit operators cannot directly issue tax-exempt debt in the same manner as public entities. If public-power counterparties become preferred channels for low-cost AI power procurement, private developers without public utility partnerships could face a relative disadvantage in delivered power cost, interconnection politics, and community acceptance. However, private operators retain advantages in speed, site control, hyperscaler leasing relationships, development execution, and willingness to absorb merchant or construction risk that many municipal entities will not accept. For IOUs and regulated utilities, the impact is ambiguous rather than purely bearish. Public-power entities may gain a visible affordability tool, but IOUs still control large portions of transmission, distribution, interconnection processes, resource planning, and regulated rate-base growth. Muni prepay does not eliminate the need for grid upgrades or new generation. It may instead pressure regulators to require clearer cost allocation for large loads, accelerate special tariffs for data centers, and scrutinize who pays for incremental capacity. IOUs with constructive regulatory frameworks can still benefit from load growth, transmission investment, and generation additions, while facing greater political risk if household bills rise.

Et si vous avez plus de maison vendez vos gosses et votre femme Generational waste on a dit, faut le faire jusqu’au bout

Sale ambiance hier et ça continue Arrêtez de refresh vos pf ça va rien changer au cas où

Gm la team n’oubliez pas que vous ratez beaucoup d'opportunités de gagner de l'argent quand vous n’avez pas d'argent

@EGuerizec @Urukobarr @Frenchie_ Évidemment je suis le fun ET le degen

@Urukobarr @Frenchie_ @Prenchie_ La vraie question c'est qui est le fun et qui est le degen 👀

Quand ton portefeuille crypto est rouge depuis 6 mois mais que tu reçois quand même : "@Frenchie_ et @Prenchie_ ont aimé votre réponse." Le fun et le degen sont une ressource rare en bear market. ❤️

@DrEidolon @Frenchie_ @Sa8Run Si on commence a me coler mes tweets ça va pas la faire

@Frenchie_ @Sa8Run @Prenchie_ dirait : tl;dr c'est la merde mais ça va passer

Les indices chutent non pas pour avoir de la liquidité pour l’IPO de SpaceX comme certains disent sur X, mais parce que ce sont les débuts de Kevin Warsh à la FOMC le 17 juin.

@BRICSinfo It’s so over Newt step is come back

JUST IN: 🇺🇸🇮🇷 US launches second wave of strikes on Iran, explosions reported.

Guess what… We are so cooke !

On est où selon vous ?

@Urukobarr @Frenchie_ ❤️

tl;dr c'est la merde mais ça va passer

@HexagoneMarkets trop long mais ça a l'air sérieux

🚨 Voici les news du marché 1. L'Iran a abattu un hélicoptère Apache américain en patrouille au-dessus du détroit d'Ormuz, et le président Trump a publié sur Truth Social que les États-Unis « doivent répondre à cette attaque », déclenchant un brusque retournement des marchés. 2. Le Nasdaq 100 a chuté de 4,1 % et le S&P 500 a perdu plus de 2,2 %, effaçant quelque 1 300 milliards de dollars de capitalisation boursière en deux heures, selon Bloomberg. Trump a confirmé que les deux pilotes étaient sains et saufs. Le président du Parlement iranien, Ghalibaf, a averti que son pays « maîtrise d'autres langages bien mieux » si ses engagements sont rompus. 3. L'argent spot a perdu plus de 5 % à 64,43 $ l'once et $AMD a clôturé en baisse de 7,80 %, emporté par le choc géopolitique qui a frappé l'ensemble du secteur des semi-conducteurs. 4. OpenAI a déposé confidentiellement un dossier d'IPO, en soumettant un S-1 confidentiel à la SEC, d'après CNBC. Aucune date d'introduction en bourse n'a été fixée et la société indique que plusieurs étapes restent à franchir avant une cotation publique. Le dépôt intervient quelques jours avant le début des échanges sur les actions SpaceX et fait suite à la propre divulgation confidentielle d'Anthropic auprès de la SEC la semaine dernière. 5. $AAPL Apple a présenté « Siri AI » lors de sa Worldwide Developers Conference à Cupertino comme la plus grande sortie IA de son histoire, avec une mise à jour complète des services iOS 27 et de nouveaux partenariats avec Google et Nvidia pour son modèle IA le plus avancé, selon CNBC. Le titre a clôturé en baisse de 3,97 % à 289,60 $, a reculé d'environ 5 % par rapport à son plus haut intraday et a effacé quelque 47,5 milliards de dollars de capitalisation boursière sur la séance. Morgan Stanley a relevé son objectif de cours sur $AAPL à 360 $ contre 330 $ et maintenu sa recommandation Overweight. 6. Le pétrole a nettement reculé après que le président Trump a déclaré aux journalistes qu'un accord États-Unis/Iran pourrait être conclu « dans deux ou trois jours » et que le détroit d'Ormuz rouvrirait « immédiatement » à la signature d'un accord, selon CNBC. Le Brent est tombé sous 90 $ le baril pour la première fois depuis le 17 avril et le brut américain a perdu 5 %. Le Short-Term Energy Outlook de l'EIA indique que les perturbations dans le détroit d'Ormuz ont contraint les producteurs du Moyen-Orient à réduire leur production de plus de 11 millions de barils par jour en mai par rapport aux niveaux d'avant-guerre. L'agence prévoit que le trafic dans le détroit ne reviendra pas aux niveaux d'avant la guerre avec l'Iran avant début 2027, les expéditions pétrolières devant reprendre au T3 2026. 7. $NUVL Nuvalent a bondi de 39 % après que le laboratoire britannique GSK a accepté d'acquérir la biotech oncologique pour 10,6 milliards de dollars, d'après CNBC. Nuvalent se spécialise dans l'oncologie de précision et l'opération vient étoffer le portefeuille cancer de GSK. L'accord s'inscrit dans un mouvement de fusions-acquisitions dans le secteur biotech, alimenté par des expirations de brevets imminentes et des marchés publics en voie de reprise. 8. $NVDA Le PDG de Nvidia, Jensen Huang, a publiquement recommandé l'action $QCOM Qualcomm lors d'un événement sectoriel, lançant à l'assistance : « achetez leur action, elle est bonne, nous devrions nous réjouir du succès des autres ». Huang a également qualifié la chute tech de vendredi d'opportunité d'achat, estimé que les valeurs IA sont « très bon marché » et réaffirmé que Nvidia n'en est qu'au début du supercycle IA, selon CNBC. Malgré cette recommandation, $QCOM a cédé 9,18 % à 197,77 $, emporté par le large reflux géopolitique de la journée, tandis que $NVDA a clôturé en recul de 2,81 % à 202,80 $. Huang a par ailleurs décliné l'invitation de la sénatrice Elizabeth Warren à témoigner devant le Congrès sur l'IA, la Chine et les contrôles à l'exportation, d'après CNBC. 9. Le rapport annuel des administrateurs de la Social Security Administration, publié aujourd'hui, projette l'épuisement du fonds de retraite en 2032, dans six ans. À ce stade, seulement 78 % des prestations seraient versables aux plus de 70 millions d'Américains qui en bénéficient, sauf intervention du Congrès, selon CNBC. Ce rapport arrive alors que les pressions financières sur les ménages s'accentuent : la Fed de New York a constaté que près de la moitié des Américains estiment être dans une situation plus difficile qu'il y a un an, 57 % des consommateurs citent la hausse des prix comme principale cause de leur dégradation financière selon Reuters, et l'NFIB Small Business Optimism Index a reculé en mai à son plus bas depuis octobre 2024, d'après Reuters. 10. Les valeurs les plus actives sur le marché des options aujourd'hui : $SPY avec 9,2 millions de contrats, $QQQ avec 6,4 millions, $TSLA avec 1,5 million, $NVDA avec 1,4 million, $IWM avec 1,2 million, $AAPL avec 722 000, $MU avec 336 000, $INTC avec 262 000, $AMZN avec 257 000 et $PLTR avec 222 000. 11. SOXS, le leveraged ETF inverse 3x sur les semi-conducteurs, a bondi de +17,05 % sur des volumes représentant 2,21 fois sa moyenne sur 20 jours. Les traders ont shorté leur exposition aux puces après la nouvelle de l'hélicoptère Apache et la chute tech qui a suivi. $SQQQ, le leveraged ETF inverse 3x sur le Nasdaq, a progressé de +8,75 % sur des volumes représentant 1,72 fois sa moyenne sur 20 jours. Cela fait suite à une séance de vendredi record au cours de laquelle les volumes combinés sur $SPXL, $SPXS, $SOXL et $SOXS ont atteint 1,1 milliard de dollars, un record absolu, selon Bloomberg. 12. Le livre d'ordres de l'IPO de SpaceX serait plusieurs fois sursouscrit, la clôture du livre étant prévue mercredi. Le S-1 de SpaceX révèle que le nombre d'abonnés Starlink est passé de 2,3 millions en 2023 à 10,3 millions au T1 2026, soit plus de 4 fois plus, tandis que le revenu mensuel par utilisateur a reculé d'environ 33 % sur la même période. Les traders de Polymarket valorisent actuellement l'IPO de SpaceX à 2 170 milliards de dollars. 13. Anthropic a publié la version publique de son modèle IA de classe Mythos deux mois après un déploiement privé, d'après CNBC. Des garde-fous bloquant les réponses dans certains domaines à haut risque ont permis cette diffusion élargie. Anthropic a également lancé Claude Fable 5, un modèle de classe Mythos destiné au grand public, Mythos 5 et Fable 5 étant tous deux tarifés à 10 $ par million de tokens en entrée et 50 $ par million de tokens en sortie. Anthropic déploiera par ailleurs Mythos 5 via Project Glasswing, un partenariat avec le gouvernement américain axé sur la cyberdéfense et les infrastructures critiques. 14. Le ratio capitalisation boursière américaine/PIB, l'indicateur Buffett, a atteint un record de 238 % alors que la valeur totale des marchés actions américains a touché un ATH à 75 700 milliards de dollars, bien au-dessus des quelque 31 800 milliards de dollars que représente l'économie américaine. Le ratio a progressé de 38 points de pourcentage depuis le début de l'année, selon Bloomberg. L'économiste en chef de Moody's a par ailleurs averti que l'économie américaine est « au bord de la récession ». La dette sur marge des marchés actions chinois a également atteint un record d'environ 431 milliards de dollars, soit 22 % au-dessus du pic de la bulle boursière chinoise de 2015, estimé à quelque 354 milliards de dollars. Ce chiffre a presque doublé en deux ans, d'après Bloomberg.