Version 34

What

Why it matters

Open questions

• How will Anthropic's S-1 characterize the competing compute relationships — a 10-year $100B+ AWS commitment [2], $45B SpaceX arrangement [14], and more than a dozen self-managed data center LOIs backed in part by Google [17][16] — without creating apparent contradictions?

• Will the SEC require Anthropic to standardize its non-standard run-rate methodology (28-day consumption × 13 plus monthly subscriptions × 12) [28], and how does the $45B/$47B discrepancy between the S-1 announcement [8] and Series H materials [6] resolve under formal review?

• Is Anthropic's 41% share of US business paid AI subscriptions [23] a durable commercial lead over OpenAI, or a point-in-time reading from a single index that will shift before the IPO roadshow?

• Will Claude Sonnet 5 confirm the leaked 1M context window and promotional pricing [33], and how does a major model release during the SEC quiet period interact with IPO disclosure obligations?

Narrative

Timeline

• 2026-02-12: Anthropic closes $30B Series G at $380B post-money valuation. [1]
• 2026-04-07: Anthropic previews Mythos cybersecurity AI model with Nvidia and Cisco as launch partners. [46][47]
• 2026-04-20: Anthropic and Amazon announce a 10-year $100B+ AWS commitment; Anthropic confirms $30B annualized revenue run rate. [2][3]
• 2026-05-06: Anthropic officially announces SpaceX compute deal: Colossus 1 access, 300+ MW and 220,000+ NVIDIA GPUs. [12]
• 2026-05-23: Multiple reports confirm Anthropic achieved Q2 operating profitability. [4][5]
• 2026-05-27: SemiAnalysis reports AWS gross margins jumped ~10 percentage points while Azure and GCP remained flat, driven primarily by Anthropic's Bedrock growth. [37][24]
• 2026-05-28: Anthropic closes $65B Series H at $965B post-money valuation; reports $47B run-rate revenue. [6][7][48]
• 2026-05-29: Bessemer projects $100B annual run rate by year-end; Anthropic shifts enterprise customers to usage-based billing after $500M accidental-spend incident. [39][29][30]
• 2026-05-31: Reuters discloses Anthropic's non-standard run-rate methodology: last 28 days of consumption sales × 13, plus monthly subscriptions × 12. [28]
• 2026-06-01: Anthropic officially confirms confidential S-1 submission to the SEC for a proposed IPO of common stock. [8][49][50]
• 2026-06-02: Musk states SpaceX holds no long-term Colossus lease with Anthropic; SpaceX's own S-1 filing discloses Anthropic pays $1.25B/month in a $45B total arrangement. [13][14][15]
• 2026-06-09: OpenAI files confidentially for an IPO at ~$852B valuation, one week after Anthropic's $965B filing. [10][35][36]
• 2026-06-11: DXC Technology joins Anthropic's Claude Partner Network; reports 95%+ Claude-generated code in OASIS platform at 10x development speedup. [25]
• 2026-06-12: Multiple outlets confirm Anthropic has signed more than a dozen data center lease LOIs, with Google backing a specific $5B facility and 1GW+ of planned U.S. capacity. [18][17][16][51][38]
• 2026-06-17: Anthropic opens Seoul office; NAVER deploys Claude Code org-wide, Samsung SDS, LG CNS, and Hanwha Solutions deploy Claude at employee scale. [26]
• 2026-06: Anthropic CEO projects $10B in 2026 data center spend and $100B annually from 2027, while separately stating mistimed investment could be "ruinous"; Anthropic recruits senior exec for European data center acquisition. [21][22][19][20][52]
• 2026-06-30: Ramp's AI Index shows Anthropic at 41% of US businesses with paid AI subscriptions, overtaking OpenAI (39.5%) for the first time. [23]
• 2026-06-30: Leaked specifications describe Claude Sonnet 5 as launching with a 1M token context window at promotional pricing of $2/$10 per million input/output tokens. [33]
• 2026-07-01: Jelani Nelson, UC Berkeley EECS department chair, joins Anthropic on leave from the university. [27]

Perspectives

Tensions

• Musk personally stated SpaceX holds no long-term Colossus lease with Anthropic [13], but SpaceX's own S-1 filing discloses Anthropic pays SpaceX $1.25B/month in a $45B total arrangement [14][15] — a direct contradiction across two companies' simultaneous SEC filings with no public reconciliation. [13][14][15][34]
• Anthropic's shift to self-managed data centers — more than a dozen LOIs, Google backing a specific $5B U.S. facility [17][16], and a European expansion hire [19] — sits alongside a 10-year $100B+ AWS commitment [2]; both cannot simultaneously be the primary compute strategy. [17][16][18][2][19]
• Anthropic's run-rate methodology — 28-day consumption × 13 plus monthly subscriptions × 12 [28] — is non-standard, and the $45B/$47B discrepancy between the S-1 announcement [8] and Series H materials [6] is unresolved disclosure the SEC will formally evaluate. [28][6][8]
• Anthropic ($965B post-money) and OpenAI (~$852B post-money) are simultaneously in confidential SEC registration [8][10], forcing parallel public disclosure of overlapping competitive intelligence neither controls independently — now extended by Anthropic's first-time lead over OpenAI in enterprise paid subscriptions [23]. [8][10][43][44][23]
• MCP as foundational enterprise infrastructure vs. a CVE-tracked vulnerability surface: registered CVEs for Claude Code [31] and OWASP's Tool Poisoning catalogue [32] challenge the enterprise-ready narrative at peak institutional momentum. [31][32][40][45]
• Anthropic's CEO publicly projects $100B in annual data center spending from 2027 [21] while separately stating that mistimed investment by a couple of years could be "ruinous" [22] — an internal tension between expansionary public commitment and an acknowledged downside scenario. [21][22]