Large Language Models

Today's Stories

1. 1

   Apple to push smart home with AI-powered Siri at center

   Apple is preparing to make a major push into smart home technology, with an AI-enhanced version of Siri as the core of its strategy. Smart home is a growing market segment, and Apple's move signals the company sees AI-powered voice assistants as central to how people will control connected devices in their homes. For Apple, this represents a way to deepen its ecosystem integration and compete in a category where Amazon and Google already have established footholds.

   Details on when the new smart home devices and AI Siri capabilities will launch, and how they will integrate with Apple's existing ecosystem of products and services.

2. 2

   Gen Z turns to AI for 69% of trip planning, far outpacing older travelers

   A Booking.com survey of 32,800 travelers across 34 countries found that 69% of Gen Z respondents use generative AI to plan trips and navigate while traveling—the highest share of any age group. Millennials follow at 64%, Gen X at 41%, and Boomers at just 11%. Gen Z travelers are more than six times as likely as Boomers to consult a chatbot for travel advice. The sharp generational gap suggests that AI-assisted travel planning is becoming a standard habit for younger travelers, reshaping how the travel industry will need to serve its audience.

   The survey, conducted in July 2026, tracked the same generational divide that shapes how people travel—indicating AI adoption in travel planning will likely rise as younger cohorts become a larger share of the travel market.

3. 3

   AI tokens may become the standard unit for measuring AI, like kilowatt-hours for electricity

   Industry observers are drawing parallels between AI tokens—the basic units that AI models process—and kilowatt-hours, the standard measure of electricity consumption. The analogy suggests that tokens could evolve into a common metric for quantifying and pricing AI work. Just as kilowatt-hours allow us to compare energy use across different sources and devices, tokens could provide a universal standard for comparing AI model performance and costs. This standardization would make it easier for businesses to evaluate and price AI services across different providers.

   Whether tokens become an industry standard for pricing and measuring AI value depends on whether major AI providers and customers adopt a common token definition. Currently, different AI companies may use tokens differently, so convergence on a unified standard would be a significant shift in how the AI industry operates.

4. 4

   BlackRock to own 80% of Meta's $14 billion（約2.2兆円） El Paso AI data center

   Meta and BlackRock announced a partnership Tuesday on Meta's 1,000-acre, one gigawatt AI data center campus in El Paso, Texas. Meta will contribute land and in-progress construction assets worth about $2.3 billion（約3700億円）, BlackRock will invest about $4.9 billion（約7800億円） in cash, and Meta will receive a $1 billion（約1600億円） distribution. The result: BlackRock-managed funds will own 80% of the project while Meta retains about 20%. This deal signals a new financing model for AI infrastructure—Wall Street is becoming a major financier of the AI race as capital costs have alarmed investors. Meta's prior earnings miss and higher 2026 capital expenditure forecast (attributed partly to AI data center costs) spooked the market, wiping about $175 billion（約28兆円） from the company's market value in a single day. This structure lets Big Tech offload ownership and funding burden to financial firms while maintaining operational control.

   Meta will report second quarter earnings after the bell Wednesday. The company has already used this financing model before—in October it announced a similar 80% ownership stake to Blue Owl Capital for the Hyperion data center campus in Louisiana, which involved about $27 billion（約4.3兆円） in project bonds and about $3 billion（約4800億円） in direct Blue Owl investment.

5. 5

   CrowdStrike positioned as cybersecurity winner in AI agentic era: Loop Capital

   Loop Capital has identified CrowdStrike as the cybersecurity company best positioned to win in the era of AI agents (autonomous AI systems that can act independently). As organizations deploy AI agents, the attack surface and complexity of cybersecurity threats are likely to expand, making advanced threat detection and response capabilities increasingly valuable for enterprise customers.

   The competitive dynamics in enterprise cybersecurity as AI adoption accelerates, and whether CrowdStrike's platform can capture market share from competitors in this emerging threat landscape.

6. 6

   Anthropic revenue run rate hits $71 billion（約11兆円）, surpassing Starbucks and McDonald's combined

   Anthropic has reached an estimated annual revenue run rate of $71 billion（約11兆円） according to data compiled by AI investment research platform Funda, making it now account for roughly 60% of the combined annual revenue run rate generated by Anthropic and OpenAI (estimated at about $120 billion（約19兆円） combined). At that level, Anthropic's annualized revenue exceeds Starbucks' reported annual revenue of approximately $37.2 billion（約6兆円） and McDonald's $26.9 billion（約4.3兆円） combined. The growth reflects rapid enterprise adoption of generative AI across software development, customer service, data analysis and workplace productivity, with Anthropic gaining traction through its Claude family of AI models and major commercial customers in finance, healthcare and technology sectors.

   Anthropic has achieved this revenue level within five years of its founding in 2021, and continues to expand through partnerships with cloud providers including Amazon Web Services and Google Cloud. Combined, OpenAI and Anthropic are operating at an annual revenue run rate of about $120 billion（約19兆円）, placing both companies among the largest businesses by revenue if sustained.