The evening briefing.
Today across AI and tech: Elon Musk's xAI faces scrutiny over its financials while securing a massive compute deal with Anthropic, Nvidia posts record earnings, and OpenAI claims a math breakthrough.
AI Economics. Elon Musk's xAI is making significant financial moves, with SpaceX's IPO filing revealing a substantial $6.4 billion loss for xAI in 2025. Despite this, xAI has secured a major deal, with Anthropic agreeing to pay $1.25 billion per month for compute capacity. This highlights the intense capital requirements for frontier AI development and the complex interdependencies forming within the industry.
Market Dynamics. The broader AI market continues to show mixed signals, as Nvidia reported another record quarter, while also disclosing a massive $43 billion in startup holdings. Meanwhile, Anthropic is projected to achieve its first profitable quarter, signaling maturation for some players. However, traditional software companies like Intuit are facing workforce reductions, with Intuit planning to cut 17% of its staff, as generative AI models pose a threat to their established business models.
AI Capabilities & Ethics. OpenAI has made a notable claim, stating its reasoning model successfully disproved an 80-year-old geometry conjecture, a significant step in AI's problem-solving abilities. Concurrently, discussions around AI's impact on employment are intensifying, with the Cloudflare CEO detailing his approach to replacing employees with AI. Apple Music is also proactively addressing AI's role in the creative industry, outlining efforts to ensure fair compensation for artists.
Tech Shifts & Security. Google is expanding its AI-powered ad formats within search, raising questions about the future user experience and content discovery. In other developments, Tesla's Full Self-Driving software is expanding into Europe, indicating a wider rollout of advanced driver-assistance systems. On the security front, Google published exploit code for a Chromium vulnerability before a patch was widely available, posing a risk to millions of users.
Anthropic will pay xAI $1.25B per month for compute
Elon Musk's xAI has entered into Cloud Services Agreements with Anthropic PBC, with Anthropic agreeing to pay $1.25 billion per month for access to compute capacity across xAI's COLOSSUS and COLOSSUS II data centers. This deal, revealed in SpaceX's IPO filing, highlights the massive demand for AI infrastructure.
xAI burned $6.4B last year, according to SpaceX’s IPO filing
SpaceX's IPO filing provides the first public look at Elon Musk's AI financials, revealing that xAI lost $6.4 billion in 2025 while planning a massive expansion for its Grok AI model. This significant expenditure underscores the high costs associated with developing advanced AI.
Nvidia posts another record quarter, reveals $43 billion of holdings in startups
Nvidia announced another record revenue figure after market close, but forecasted that revenue growth would slow in the following quarter. The company also disclosed $43 billion in holdings across various startups, indicating its broad investment strategy in the tech ecosystem.
Mind-blowing growth is about to propel Anthropic into its first profitable quarter
Anthropic is on the cusp of achieving its first profitable quarter, driven by what is described as "mind-blowing growth." This milestone reflects the increasing demand for its AI models and services.
OpenAI claims it solved an 80-year-old math problem
OpenAI claims its reasoning model has disproved a geometry conjecture that remained unsolved since 1946. This achievement is reportedly backed by mathematicians who previously exposed an earlier incorrect claim by the company.
Zuckerberg's Meta layoffs memo: 'Success isn't a given' in the AI era
Meta commenced its latest round of layoffs, affecting about 10% of the company's workforce or roughly 8,000 jobs. Mark Zuckerberg's memo emphasized that "success isn't a given" in the rapidly evolving AI era.
Cloudflare CEO on how he chooses which employees to replace with AI
The CEO of Cloudflare discussed his methodology for identifying roles and employees that could be potentially replaced by artificial intelligence. This offers insight into how companies are approaching workforce integration with AI.
Google is bringing new AI-powered ad formats to search
Google is introducing new AI-powered ad formats to its search platform, aiming to enhance the in-search advertising experience. This move follows Google I/O announcements and raises questions about the evolving nature of search results.
Apple Music shares what it is doing to ‘keep music fair’ in an AI world
Apple Music released an open letter to the music industry, outlining its commitment to ensuring fair practices for artists in the age of AI. The company emphasized that AI music currently represents less than 1% of plays on its service.
Tesla’s Full Self-Driving software is creeping into Europe
Tesla's Full Self-Driving (FSD) software is expanding its presence in Europe, with initial rollouts in the Netherlands and Lithuania. More European countries are expected to follow, marking a significant step for the driver-assistance system.
Kansas City Public Schools to replace 30,000 Windows PCs and Chromebooks with Apple devices
Kansas City Public Schools announced plans to transition to an "all-Apple district," replacing 30,000 Windows PCs and Chromebooks with Apple devices. This move follows a comment made during Apple’s Q2 2026 earnings call.
We may only have a year of the RAM crisis left if this ex-Samsung boss is right
An ex-Samsung executive suggests that the global RAM crisis could begin to ease starting from the second half of next year. This potential shift is attributed to a correction in supply and demand dynamics.
NanoCo bets the future of enterprise AI is one sandboxed agent per employee
NanoCo, the Tel Aviv startup behind the NanoClaw agent framework, launched a managed enterprise service that proposes a future with one sandboxed AI agent per employee. This aims to customize AI for specific enterprise tasks.
Google publishes exploit code threatening millions of Chromium users
Google has published exploit code for a Chromium vulnerability, despite a patch for the issue, reported 29 months earlier, not yet being widely fixed. This action could potentially expose millions of Chromium users to threats.