The Public Market Phase

SpaceX began trading on Nasdaq this morning. Anthropic, which this week disclosed it pays $1.25 billion per month to rent xAI compute, is simultaneously worth $965 billion, has crossed $47 billion in annualized revenue, and has confidentially filed for an IPO. OpenAI is behind it in the same pipeline. The AI companies enterprise leaders have built their strategies around are entering the public market. The information environment is about to become significantly more legible — and the compliance clock is ticking.

In this edition: Anthropic at $965B — $47B ARR, IPO pipeline, and what the S-1 will tell enterprise buyers  ·  SpaceX SPCX opens trading  ·  NVIDIA and SK Hynix lock in memory for AI factories  ·  EU AI Act: 52 days to enforcement

Anthropic at $965 Billion: The Vendor Disclosure Event Enterprise Buyers Must Prepare For

In late May, Anthropic raised $65 billion in a Series H financing at a $965 billion post-money valuation — the highest-valued AI startup in history, overtaking OpenAI. Run-rate revenue crossed $47 billion. The company has since confidentially filed for an IPO, targeting a fall 2026 listing. When Anthropic's S-1 becomes public, enterprise AI buyers will have access to financial disclosure about their primary AI vendor at a level that has never previously existed. The organizations that have built the analytical frameworks to evaluate it will be at an advantage.

The valuation trajectory tells a specific story. In February 2026, Anthropic was valued at $380 billion. Three months later, in May, the Series H values the company at $965 billion — a near-tripling. That trajectory reflects the Fable 5 launch and the continued acceleration of enterprise AI adoption, but also something more structural: the compute contracts disclosed in SpaceX's IPO filing this week confirm that Anthropic's $47 billion in annualized revenue is sufficient to support a $1.25 billion monthly compute bill while still generating a valuation that has surpassed every prior AI benchmark.

Anthropic's enterprise distribution strategy is also visible in this round. The $5 billion Amazon commitment, combined with Google and Microsoft relationships, means Claude is embedded in all three major cloud platforms simultaneously — a procurement posture that no single-platform AI vendor can match. Enterprise teams evaluating Claude through AWS Bedrock, Google Vertex AI, or Azure AI Studio are accessing a vendor with strategic commitments from all three hyperscalers. That multi-cloud distribution is now backed by $65 billion in capital and infrastructure partnerships with the memory suppliers building the next generation of AI chips.

The IPO filing is the more significant event for enterprise procurement. When Anthropic's S-1 becomes public, the financial information that enterprise buyers have been using in vendor evaluations — extrapolated from press releases, analyst estimates, and pricing sheets — will be replaced by a regulatory filing. Revenue by segment. Enterprise churn rates. The margin structure of Claude API and Enterprise plans. The full operating cost picture including compute contracts. Customer concentration. The organizations that have built analytical frameworks to evaluate vendor S-1s like vendor financial audits will have a significant informational advantage in the procurement conversations that follow.

Moving Pieces

Three more developments that matter to enterprise leaders this week

NVIDIA and SK Hynix Lock In Memory for the AI Factory Era

NVIDIA and SK hynix announced on June 7 a multiyear technology partnership to co-develop memory for AI factories at scale. SK hynix will codevelop memory for NVIDIA's Vera Rubin AI supercomputers, Vera CPUs, RTX Spark-powered PCs, and Jetson Thor robotics platforms — covering the full hardware stack from data center to edge AI. The partnership also applies AI to semiconductor manufacturing itself, using NVIDIA CUDA-X and PhysicsNeMo to accelerate chip simulations, and building factory digital twins via NVIDIA Omniverse. The strategic connection worth noting: SK hynix is simultaneously a strategic infrastructure investor in Anthropic's $65 billion Series H. The memory supplier building next-generation AI hardware for NVIDIA's roadmap has also committed capital to the frontier AI lab running on NVIDIA's GPUs. The AI infrastructure ecosystem is consolidating around a small number of deeply interconnected relationships — and the boundaries between customer, supplier, and investor are dissolving.

EU AI Act: 52 Days to High-Risk Enforcement — €15M Fines on August 2

August 2, 2026 — 52 days from today — is the binding enforcement date for high-risk AI system obligations under the EU AI Act. The deadline covers Articles 9-17 (provider requirements) and Article 26 (deployer requirements), applying to AI systems used in employment, credit decisions, education, and law enforcement contexts. Penalties: up to €15 million or 3% of global annual turnover, whichever is higher. U.S. companies operating high-risk AI systems that touch EU data subjects or EU markets are within scope. There has been legislative movement toward a 16-month postponement for new or substantially modified systems, but compliance experts uniformly advise treating August 2 as fixed. Enterprise organizations that have not completed their Annex III system inventory, risk classification, technical documentation, and human oversight procedures are in a compressed window with no margin for extended planning cycles. Colorado's comprehensive state-level AI law for high-risk systems takes effect June 30 — in 18 days.

SpaceX SPCX Opens Trading — AI Compute Infrastructure Enters Public Markets

SpaceX shares began trading on Nasdaq today under the ticker SPCX, opening at its IPO price of $135 per share — a $1.75 trillion valuation and the first public market session for the company that supplies compute to Google ($920M/month) and Anthropic via xAI ($1.25B/month). Retail allocation was approximately 30% of the $75 billion offering, triple the industry norm, drawing retail orders exceeding $100 billion. The transition from private to public creates a new information channel for enterprise AI procurement: quarterly earnings reports will now track the compute contracts that set the floor for enterprise AI inference pricing. Enterprise teams with cloud AI workloads now have a publicly traded proxy for the infrastructure tier beneath their vendors — one that will report on its largest customers, the Google and Anthropic compute contracts, in its first public filing cycle.

until August 2, 2026 — the EU AI Act's binding enforcement date for high-risk AI system obligations. Non-compliance: €15 million or 3% of global annual turnover, whichever is higher.

High-risk AI systems under Annex III include AI used in employment (hiring, performance evaluation, work allocation), credit and insurance decisions, education and vocational training, and access to essential services. Many enterprise organizations have AI pilots or production deployments that fall into these categories without having conducted formal risk classification. The compliance window is now measured in weeks. Enterprises that have not completed their Annex III system inventory, technical documentation, conformity assessments, and human oversight procedures need to begin immediately. The $15 million penalty floor applies to smaller organizations — the 3% of global annual turnover provision is the operative risk for large enterprises. Colorado's comprehensive state AI law takes effect in 18 days.

Source: Holland & Knight  ·  ComplianceHub.Wiki  ·  EU AI Act

When Your Vendor Goes Public

This week, the AI infrastructure that enterprise leaders have been building on became publicly legible. SpaceX trades live. Anthropic is in the IPO pipeline at $965 billion. OpenAI filed its S-1 in June.

When a primary AI vendor files an S-1, the enterprise procurement relationship changes in a specific way. The financial information that has been used to evaluate vendor stability, pricing trajectory, and strategic direction becomes public record. Churn rates, revenue by customer segment, compute costs as a percentage of revenue, operating margins — all of it becomes readable. The organizations that have built the analytical frameworks to treat a vendor S-1 like a financial audit will have a significant informational advantage in the procurement conversations that follow. Most enterprise vendor review processes are not built for this. They should be.

The EU AI Act countdown running in parallel — 52 days to August 2 — converges on the same question: how well do you actually understand the AI infrastructure you depend on? The companies entering the public market are the same companies whose systems your legal team needs to document, classify, and assess for EU compliance. The S-1 will disclose things about those systems that the vendor's sales team never has. And the Colorado state law activates in 18 days, before any of the S-1s land.

The week the enterprise AI market became readable. Start reading.

AK  /  Spearhead  /  Building AI systems, not tools