SpaceX Eyes $1T IPO as DeepSeek Chip Scandal Grows

The past week delivered major moves across space, AI, and enterprise software as SpaceX charts a path to a trillion-dollar IPO, DeepSeek faces allegations of evading US chip export controls, and leadership shuffles ripple through Slack, OpenAI, and Hinge. Meanwhile, regulators on three continents are tightening scrutiny of AI training practices, and the industry’s biggest players are joining forces to standardize the emerging agent economy.

SpaceX Targets $25B+ Raise at Trillion-Dollar Valuation

SpaceX is preparing an IPO for mid-2026 that could raise over $25 billion at a valuation exceeding $1 trillion, according to Reuters. The company projects revenue growth from $15 billion in 2025 to $24 billion in 2026, driven largely by Starlink’s subscriber base. Proceeds would fund space-based data centers and significant chip purchases as Elon Musk positions SpaceX to dominate both orbital infrastructure and AI compute.

DeepSeek Accused of Using Banned Nvidia Chips

Chinese AI startup DeepSeek allegedly acquired thousands of banned Nvidia Blackwell chips to train its reasoning models, reigniting concerns over export control enforcement. Federal prosecutors charged two men in Operation Gatekeeper with smuggling $160 million worth of H100 and H200 GPUs, which were repackaged as generic computer parts and routed through third-country data centers. The hardware reportedly helped DeepSeek bypass yield issues with Huawei Ascend 910C processors affecting its R2 model development.

Nvidia Tests Location-Tracking Software for AI Chips

In response to the smuggling allegations, Nvidia is testing optional location-verification software for its AI chips, starting with the Blackwell line. The system uses performance and communication delay signals to estimate chip locations. The move follows US approval for Nvidia to export older H200 chips to approved China customers, as the company navigates an increasingly complex regulatory environment.

Slack CEO Denise Dresser Joins OpenAI

Denise Dresser, who spent 14 years at Salesforce and Slack, is leaving to become chief revenue officer at OpenAI, bringing enterprise sales expertise to the AI giant’s revenue strategy. Rob Seaman, Slack’s chief product officer, will serve as interim CEO. The hire signals OpenAI’s continued push to capture enterprise market share as competition with Anthropic and Google intensifies.

Meta AI Team Creates Internal Tensions

Mark Zuckerberg hired entrepreneur Alexandr Wang to lead a siloed AI team called TBD Lab, which operates in glass-paneled offices next to the CEO and is isolated from corporate bureaucracy, The New York Times reports. The arrangement has sparked disagreements with executives like Chris Cox over strategic priorities, with Wang arguing the group must catch up to OpenAI and Google before fulfilling internal requests for advertising and feed improvements.

State Attorneys General Demand AI Safety Fixes

Dozens of US state attorneys general sent a letter to major AI companies demanding action on harmful “delusional” outputs linked to mental health incidents, including suicides. The letter calls for mandatory third-party audits, pre-release safety testing, and incident-reporting systems. The action comes as the federal government pushes pro-AI policies and President Trump plans an executive order limiting state-level AI regulation.

OpenAI Warns New Models Pose ‘High’ Cybersecurity Risk

OpenAI disclosed that its next-generation AI models could create or assist with advanced cyberattacks, including zero-day exploits. The company is investing in defensive tools, stronger access controls, and infrastructure hardening while launching a Frontier Risk Council with security experts. Tiered access to enhanced capabilities will be offered to qualified cyber defenders.

Linux Foundation Launches Agentic AI Standards Initiative

Anthropic, Block, and OpenAI have joined forces as founding contributors to the Agentic AI Foundation under the Linux Foundation, with AWS, Google, Bloomberg, and Cloudflare as supporting members. The initiative aims to establish open standards for AI agent interoperability, safety, and tool integration—addressing vendor lock-in concerns as the agent economy rapidly expands. Anthropic is contributing its MCP protocol, while Block brings Goose and OpenAI adds AGENTS.md.

Google Elevates Data Center Architect Amid AI Arms Race

Google created a new chief technologist for AI infrastructure role and elevated Amin Vahdat, the 15-year architect behind the company’s TPU chips, Jupiter network, and Borg systems, according to PCMAG. Reporting directly to Sundar Pichai, Vahdat will oversee Google’s projected $93 billion capex spend in 2025 as it races to compete with OpenAI, Anthropic, and Nvidia. The company’s new Ironwood TPUs deliver 42.5 exaflops per pod.

ChatGPT Tops Apple’s 2025 Most-Downloaded Apps

ChatGPT claimed the top spot on Apple’s most-downloaded free iPhone apps in the US for 2025, surpassing Threads, Google, TikTok, WhatsApp, and Instagram. The app rose from fourth place last year, reflecting surging mainstream AI adoption. ChatGPT also became the most-downloaded app globally earlier in the year.

EU Opens Antitrust Probe Into Google’s AI Practices

European regulators launched an antitrust investigation into Google’s scraping of publisher content for AI products without compensation. The probe focuses on “all-or-nothing” data harvesting choices for AI Overviews and YouTube blocking rival developers from video data access. Publishers argue that generative summaries cannibalize traffic by eliminating the need for click-throughs.

Australia Bans Social Media for Users Under 16

Australia enacted a social media ban for users under 16, with platform fines reaching up to AU$49.5 million for non-compliance. Verification will require government ID or video selfies. Exemptions cover YouTube without login, messaging apps like WhatsApp and Discord, and gaming platforms including Roblox and Steam.

Hinge CEO Launches AI Dating Startup Overtone

Justin McLeod, founder and CEO of Hinge, is stepping down to launch Overtone, an AI- and voice-powered dating app incubated within Hinge. Jackie Jantos takes over as CEO. Match Group holds a substantial ownership stake in the new venture, which aims to address declining engagement across traditional dating apps with AI-driven experiences.

India Proposes Charging AI Companies for Training Data

India’s government proposed a mandatory royalty framework for AI companies using copyrighted content in training. The structure would include blanket licenses with automatic access and a single collecting body for payment distribution. Industry groups Nasscom and the Business Software Alliance are pushing back, advocating for text-and-data-mining exceptions instead.

This week underscores the accelerating collision between AI ambition and regulatory scrutiny. SpaceX’s IPO plans signal confidence in space infrastructure economics, while the DeepSeek chip scandal reveals the geopolitical stakes of AI hardware supply chains. As industry leaders standardize agent protocols and governments from Brussels to Canberra tighten oversight, expect the AI landscape to reshape dramatically heading into 2026.