Amazon Anthropic stake of $190B linked to FTX funding raises compliance questions
The $500 million transaction was executed by a special-purpose vehicle that sourced capital from FTX’s pooled customer accounts.
Amazon Anthropic stake emerges from pre-bankruptcy FTX funding
Amazon’s balance sheet now reflects a $190 billion equity position in AI startup Anthropic. The holding consists of $98 billion of convertible notes and $92 billion of non-voting preferred shares. This exposure originated from a $500 million purchase of Anthropic equity in April 2022 that was financed with customer funds held by the cryptocurrency exchange FTX, which was then led by Sam Bankman-Fried (SBF). The direct link between Amazon’s investment and pre-bankruptcy FTX capital answers the core question of how a major U.S. corporation became entangled with the failed crypto exchange.
Origin of the Amazon Anthropic stake
The $500 million transaction was executed by a special-purpose vehicle that sourced capital from FTX’s pooled customer accounts. When FTX filed for Chapter 11 in November 2022, the vehicle transferred the Anthropic shares to Amazon in a secondary-market deal. Amazon did not purchase the shares directly from FTX, but the provenance of the equity ties the Amazon Anthropic stake to the collapsed exchange and, by extension, to SBF’s management of those funds.
METR: The safety evaluator with EA roots
Anthropic’s AI safety arm, Model Evaluation and Threat Research (METR), operates as an independent evaluator of frontier-AI models. METR’s predecessor, the Alignment Research Center, received a $1.25 million grant from the FTX Foundation in July 2022. After an internal ethics review, METR returned the grant in 2024, stating that the money morally belongs to FTX customers and creditors. Today METR is funded primarily by Good Ventures Foundation and Coefficient Giving, both of which are linked to the effective-altruism (EA) movement. Beth Barnes, METR’s CEO, has a long-standing affiliation with EA, and several researchers maintain ties to the University of Oxford, a known EA hub. Holden Karnofsky, co-founder of GiveWell and Coefficient Giving, is married to Daniela Amodei, Anthropic’s president and sister of CEO Dario Amodei. These relationships illustrate how the Amazon Anthropic stake intersects with EA-driven philanthropy and AI safety research.
Operational implications for Amazon and the AI market
Amazon began working directly with METR in 2025 when METR performed a safety review of Amazon’s in-house frontier model. For operators building large language models, this partnership signals a shift toward formal third-party safety audits, which could become a de-facto compliance requirement as regulators tighten AI oversight. From a capital-allocation perspective, Amazon’s financial performance now moves in tandem with Anthropic’s valuation trajectory. If Anthropic’s next funding round or a potential IPO in 2027 raises its market cap, Amazon’s balance sheet will reflect a proportional increase in goodwill and equity value. Conversely, regulatory clampdowns on AI safety standards—especially those involving METR’s risk assessments—could depress Anthropic’s valuation and directly affect Amazon’s earnings.
Regulatory and legal context
The link between a major U.S. corporation and funds derived from a failed crypto exchange raises compliance questions. The SEC continues to scrutinize crypto-related financial flows, and any future investigations into FTX-derived capital could extend to downstream investors like Amazon. In addition, recent public statements from the Department of Defense labeling effective altruism “misaligned with Americanism” suggest a broader political pushback against EA-linked funding structures, potentially influencing future policy on AI safety financing.
Second-order risks for institutional stakeholders
- Reputational exposure – Institutions that partner with Anthropic or rely on its models may inherit the reputational baggage of the FTX connection. Asset managers could face ESG-related inquiries if they hold exposure to Amazon’s AI portfolio.
- Compliance overhead – The return of the $1.25 million grant demonstrates a proactive stance on moral liability. Companies may need to audit historical funding sources for AI projects to avoid similar retroactive liabilities.
- Strategic realignment – As METR’s risk assessments gain prominence, AI developers might prioritize safety-first architectures, potentially slowing time-to-market for certain features but enhancing long-term stability.
What to watch next
- Anthropic’s valuation path – Upcoming funding rounds or a potential IPO will clarify how much of the Amazon Anthropic stake translates into market-based returns.
- SEC enforcement trends – Any enforcement action targeting crypto-derived capital in non-crypto enterprises would set a precedent for Amazon and similar firms.
- METR’s governance evolution – If METR restructures to distance itself from EA funding, its credibility as an independent safety auditor could rise, affecting industry adoption of third-party safety reviews.
Institutional perspective on effective-altruism funding
The entanglement of EA-aligned foundations with high-stakes AI projects illustrates a broader financing trend where philanthropically motivated capital fuels commercial AI development. While this can accelerate safety research, it also creates opaque funding pathways that regulators may later scrutinize. Institutional investors should therefore map the provenance of AI-related capital to assess hidden exposure to legacy crypto-funded assets.
Impact on Amazon Web Services (AWS)
AWS stands to benefit from preferential access to Anthropic’s models, allowing the cloud division to bundle advanced generative-AI capabilities into its AI-as-a-Service portfolio. However, any regulatory restriction on Anthropic’s models could limit AWS’s ability to market AI-enhanced services, forcing the division to diversify its model portfolio.
Is METR truly independent from Anthropic?
METR originated from the Alignment Research Center, which received direct funding from the FTX Foundation and continues to draw support from EA-linked donors. Although METR claims operational independence, its staffing and funding sources suggest a close alignment with Anthropic’s strategic interests, a point highlighted by former White House AI adviser David Sacks.
How might the FTX connection affect future AI safety standards?
Regulators may view the FTX-derived capital as a conflict of interest, prompting stricter disclosure requirements for safety auditors. If METR’s funding sources become a focal point, the industry could see a push for more transparent, diversified safety-audit funding models.
For a deeper dive into the institutional risks surrounding METR’s role, see the analysis of the Anthropic safety evaluator conflict Anthropic safety evaluator conflict: Institutional risks and market impact.
Trusted source: SEC filing
Related coverage
- Manifund hires Caroline Ellison while fighting $1.5M FTX lawsuit
- Sam Bankman-Fried Supreme Court appeal raises new legal stakes for crypto
- President blocks superintelligence: How policy shapes AI safety