Anthropic's annualized revenue run rate reached $65B at the end of July — a sevenfold year-over-year increase. Series H was led by Altimeter, Dragoneer, Greenoaks, and Sequoia. Investors are now pricing a potential Anthropic IPO at $2T valuation with 53% probability by 2027.
Why it matters: At this scale Anthropic is no longer a startup — it's an infrastructure company competing with the biggest tech platforms on earth. The IPO timeline will dictate the next 18 months of AI investment patterns.
AI automation startup Relay is shutting down. CEO Jacob Bank and the team are joining Google Chrome as VP of Product. Free users lost access on August 15; paying customers have until September 14 to migrate.
Why it matters: A cautionary tale for teams relying on AI automation startups for critical workflows. Always have an exit plan before integrating a tool that could be acquired or shut down.
Reports confirm Amazon pulped physical rare books from its low-velocity inventory to train AI models. The incident reveals that AI training data has a physical supply chain — and that supply chain can destroy cultural heritage.
Why it matters: Raises serious ethical questions about data provenance for AI models trained on physical media. Expect regulatory scrutiny on AI training data sourcing to intensify.
Nvidia is guaranteeing up to $105B for OpenAI's Ohio data center through SB Energy/SoftBank. The facility will have 8 GW capacity with the first 800 MW online in 2028. Nvidia is also making a $1.5B direct investment.
Why it matters: This is one of the largest single data center financing commitments in history. Signals that AI compute infrastructure is becoming a sovereign capital project, not just a tech company expense.
The Department of Justice has opened an antitrust investigation into Andreessen Horowitz's board seats at competing AI companies — specifically Databricks and Fivetran. The probe examines whether a16z's cross-portfolio governance creates anti-competitive coordination.
Why it matters: Venture capital governance is now under federal antitrust scrutiny. This could reshape how VCs invest in competing AI companies and who gets board representation.
Google has reversed its mandatory AI watermark policy. Users can now remove visible watermarks from AI-generated images, after pushback from creators who argued watermarks degraded image quality and were easily bypassed anyway.
Why it matters: Watermarking AI content remains a contentious policy debate. Google's reversal signals that mandatory watermarking faces significant pushback from the creative community.
Writer has launched a new proprietary AI model with an upgraded harness for automated model routing by task complexity. The company claims 30–60% token cost reduction by matching tasks to the cheapest capable model.
Why it matters: Model routing is becoming the new battleground for enterprise AI cost control. Writer's harness joins the growing category of AI infrastructure tools that optimize spend across multiple LLM providers.
The Pentagon's Chief Digital and Artificial Intelligence Office (CDAO) AI-first procurement mandate is driving a new funding round for Smack, an AI-powered defense contracting platform. The move signals continued military investment in AI procurement infrastructure.
Why it matters: Defense AI procurement is accelerating. Tools that streamline defense contracting with AI assistance are seeing rapid adoption as the Pentagon pivots to AI-first operations.
This week showed AI at full maturity as an economic force — $65B revenue run rates, $105B infrastructure deals, antitrust probes into VC governance, and physical-world consequences of training data sourcing. The builders are now the story, and the stakes are genuinely infrastructure-level.