What technology stack does BNY run, and how is AI built on it?
BNY runs a hybrid stack with its Eliza AI platform on top. Its Q1 2026 investor deck dates the layers: an expanded cloud and AI collaboration with Microsoft to scale generative AI in 2023, then on-premise GPU infrastructure 'powered by NVIDIA SuperPODs to supplement our cloud solutions' and the launch of Eliza, now described as a model-agnostic platform that integrates models from OpenAI, Google and Anthropic. Microsoft says Eliza runs partly on Azure and draws context from Microsoft 365 through the Graph API; Google Cloud's Gemini Enterprise was added on December 8, 2025. CIO Leigh-Ann Russell describes a hybrid of cloud and BNY's own facilities, credited with a 90% fall in serious incidents, and the 2026 proxy says more than 70% of employees worked in the platforms operating model at the end of 2025. All 7,000 developers have the Windsurf AI coding tool, and Russell told Bloomberg Intelligence in August 2026 that '60% of the bank's code is AI generated'. Software and equipment expense was $2,147 million in 2025, against $1,962 million in 2024; CNBC put total technology spending at about $3.8 billion, roughly 19% of revenue, and the CFO cites about $4 billion a year on engineering. BNY does not disclose its cloud share, GPU count or AI spending as a separate line.[1][2][3][4][5][6][7][8][9]
Data as of Oct 31, 2026 · BNY profile updated Oct 19, 2026 · 31 sources cited on this page · 141 banks profiled on this site · Markdown
Which clouds, chips, platforms and models make up BNY's stack?
What BNY, its vendors or a named executive have said about each layer. 'Since' is the first dated mention.
| Layer | What BNY uses | Since | Who said it |
|---|---|---|---|
| Public cloud | An expanded cloud and AI collaboration with Microsoft to scale generative AI; Eliza runs partly on Microsoft Azure; in 2024 Fortune described Eliza's cloud infrastructure as powered by Azure and Google Cloud | 2023 | Q1 2026 investor deck; Microsoft; Fortune[1][2][10] |
| Compute on premises | An NVIDIA DGX SuperPOD with DGX H100 systems and InfiniBand networking, the first at a major bank, built to supplement BNY's cloud solutions; early uses were deposit forecasting, payment automation and predictive trade analytics | March 18, 2024 | NVIDIA; Q1 2026 investor deck[11][1] |
| AI platform | Eliza: about 14,000 users, a quarter of staff, in August 2024; about 35% of employees enabled at year-end 2024; Eliza 2.0 launched in 2025 and adopted by 99% of employees | 2024 | Fortune; 4Q24 deck; 2026 proxy[10][12][5] |
| Foundation models | OpenAI's GPT-4, Google's Gemini and Meta's Llama in 2024; models from OpenAI, Google and Anthropic in the 2026 deck; Gemini Enterprise integrated into Eliza on December 8, 2025 | 2024-2026 | Fortune; Q1 2026 investor deck; BNY and Google Cloud[10][1][3] |
| Microsoft 365 link | Eliza draws context from Azure, Outlook, Teams, SharePoint and Dynamics CRM through the Graph API; Copilot offered broadly to employees from November 2023, now licensed to roughly half the workforce | November 2023 | Microsoft WorkLab[2] |
| Coding tools | GitHub Copilot and Microsoft Copilot 365 broadly; Windsurf for all 7,000 developers; 'Every single one of our coders, nearabout, is using AI assisted coding' | 2025 | Leigh-Ann Russell, CIO[4] |
| Governance layer | Eliza built 'as a model-agnostic AI platform and governance layer' to keep control over data, spending and risk; 2025 additions included 'a single pane of glass for enterprise AI initiatives' | 2025-2026 | Bloomberg Intelligence; Q1 2026 investor deck[6][1] |
| Data centers and resiliency | A hybrid of cloud and BNY's own facilities with data center upgrades; serious incidents (P1s and P2s) down 90%, credited to automation, self-healing and machine learning for anomaly detection | January 9, 2026 | Leigh-Ann Russell, CIO[4] |
How much does BNY spend on technology, and how much of its code does AI write?
Expense lines are from the 10-K income statement; the spending totals are as CNBC and the CFO described them.
| Measure | Value | As of |
|---|---|---|
| Software and equipment expense | $2,147 million, against $1,962 million in 2024 and $1,817 million in 2023 | Full year 2025[7] |
| Technology spending | About $3.8 billion in 2025, roughly 19% of revenue, the highest share among large-bank peers in CNBC's data | February 9, 2026[8] |
| Engineering spending | About $4 billion a year, which funds the Eliza platform, per the CFO | September 19, 2026[9] |
| AI-written code | '60% of the bank's code is AI generated' (Bloomberg Intelligence); 'over 60% of the bank's software code is built with AI's help' (American Banker) | August 11, 2026; October 1, 2026[6][13] |
| Platforms operating model | More than 70% of employees working in it at year-end 2025, after about 50% transitioned during 2025 | December 31, 2025[5] |
| Workforce | Approximately 48,100 full-time employees; about 60% of all employees based outside the U.S. | December 31, 2025[14] |
Does BNY build or buy its AI technology?
It buys compute and models and builds the platform. BNY rents cloud capacity from Microsoft and Google, owns NVIDIA GPUs on premises and licenses models from OpenAI, Google and Anthropic; Eliza, its governance and the agents built on it are BNY's own. Fortune put it this way in 2024: the chips and clouds are bought, but Eliza's mind 'is a BNY original'.[10][1][2]
Engineering is organized as platforms. Russell said BNY moved from 'three big businesses that tended to operate in a bit of a silo' to a platform model in which client platforms such as treasury services or payments sit alongside engineering platforms, including the API layer, architecture layer and software frameworks, and horizontal practices for 'cyber, production services, infrastructure, data, AI, architecture and controls'.[4]
Not disclosed: the share of applications in the cloud, the number of GPUs, payments to model providers or AI costs as a line item; American Banker reported after the October 15, 2026 call that BNY would not disclose its AI costs.[15][7]
What does the BNY profile record on this topic?
The rows below are the entries of the BNY AI strategy profile that bear on this page, as of Oct 19, 2026; the profile holds the full record of 39 dated moves, 14 use cases and 58 sources.
| Use case | What the record shows | Status | Source |
|---|---|---|---|
| Eliza and employee-built agents | Enterprise platform; more than half of employees building agents; ~140 digital employees on repetitive tasks. | In production | [1][8] |
| AI-generated code | The CFO says engineers write a growing share of code with AI; the bank routes tasks to the most appropriate models. | In production | [16] |
| OpenAI and Google Cloud models on Eliza | Multi-model platform: OpenAI (Feb 2025) and Gemini Enterprise (Dec 2025). | In production | [17][18] |
| Azure document intelligence in lockbox | More than 500,000 check images a day read with Azure document intelligence, alongside Copilot tools for employees. | In production | [2] |
| Contract Review Assistant | Agent built on Eliza that cut legal review of vendor agreements from four hours to one (75%) across more than 3,000 agreements a year. | In production | [19] |
| Frontier cyber-model testing | Early access to OpenAI GPT-5.4-Cyber and Anthropic Mythos for defensive testing; digital employees also find and fix code vulnerabilities. | Pilot | [20][13] |
| Date | Move | Detail | Source |
|---|---|---|---|
| Oct 1, 2026 | Russell profile: usage up from about 20% to 99% trained in a year | American Banker says about 20% of staff used Eliza at the start of 2025 against a 65% year-end target; 99% were trained within six months, more than 50% build agents and over 60% of code is built with AI help. It also credits Russell with digital employees that detect and fix code vulnerabilities, now more than 150. | [13] |
| Sep 19, 2026 | CFO: ~$4 billion annual engineering spend funds Eliza; expense outlook raised partly on AI | At an investor conference McDonogh frames BNY's next growth phase around cross-selling and AI investment, says AI is meant to create capacity to reinvest in client delivery and product development rather than just cut headcount, flags AI's potential to modernise spreadsheet-heavy loan-market processes, and raises the 2026 expense-growth outlook to 6%-7%, partly on added data and AI investment. | [9] |
| Sep 17, 2026 | AI named as a factor amplifying payments fraud | Carl Slabicki, Head of Commercial, Global Payments & Trade, tells PYMNTS that AI is changing the fraud landscape by amplifying social engineering and making deepfakes and synthetic-identity fraud more convincing. | [21] |
| Aug 12, 2026 | Vince rates his own AI use nine out of 10 | In an Axios interview two weeks after joining OpenAI's board, Vince describes an Eliza daily planner that reads his calendar, email and documents, a link between Eliza and the 360 feedback process, and model routing chosen by the app; he cites 46,000 employees. | [22] |
| Jul 31, 2026 | Q2 10-Q: AI enhances cyber threats | The quarterly report says recent advancements in artificial intelligence have enhanced threat actors' capability to identify and potentially exploit previously unidentified cybersecurity vulnerabilities, and lists AI among the forward-looking topics alongside the transition to a platforms operating model. | [23] |
| Jul 28, 2026 | CFO: about half of code now AI-authored | McDonogh tells Fortune BNY tracks outcomes across innovating, prospecting, onboarding, transacting and streamlining rather than tokens; token costs are 'modest within modest'. | [24][25] |
| Jul 21, 2026 | Vince joins OpenAI's boards | Named to the OpenAI Foundation and OpenAI Group PBC boards ahead of a prospective IPO. | [26][27] |
| Jul 15, 2026 | Q2 call and deck: AI value in three dimensions; operating model activated | Vince frames AI as running the company better, building better products and expanding BNY's perimeter, and says it is 'a mistake' to see it only as efficiency. The deck says the platform operating model's activation phase finished in Q2 and BNY is tracking toward about $500 million of incremental investment and about $450 million of efficiency savings in 2026; the CFO puts AI spend as 'de minimis' within a roughly $4 billion engineering budget. | [25][28] |
| May 5, 2026 | Named among Anthropic's finance-agent customers | Anthropic's launch of ten finance agent templates carries a BNY statement: 'With Eliza and Claude, we're giving processes new digital employees who work the case end to end.' | [29] |
| Apr 27, 2026 | Federated-learning fraud detection proof of concept | Kinexys by J.P. Morgan, BNY, RBC, DeepTempo and NVIDIA trained local fraud models without pooling transaction data, using NVIDIA FLARE on DGX Cloud; the result outperformed single-institution models and nearly matched a centralised one. | [30] |
| Apr 16, 2026 | Early access to OpenAI and Anthropic cyber models | Axios reports BNY is among the firms testing GPT-5.4-Cyber and Anthropic's Mythos after Treasury Secretary Bessent and Fed Chair Powell convened bank CEOs on AI-enabled attack risk; Vince says the labs want real-world feedback. | [20] |
| Apr 16, 2026 | Q1 call: 218 AI solutions in production, up fourfold | Vince recounts the 2023 AI Hub and BNY's DGX SuperPOD with NVIDIA, the first at a global bank, and says 2026 moves from point solutions to end-to-end processes. | [31] |
Which cloud does BNY use?
Microsoft Azure and Google Cloud. BNY's Q1 2026 deck dates an expanded cloud and AI collaboration with Microsoft to 2023, Microsoft says the Eliza platform runs partly on Azure, and Google Cloud's Gemini Enterprise was integrated into Eliza in December 2025. The CIO describes a hybrid that also uses BNY's own data centers.
Does BNY have its own AI supercomputer?
Yes. NVIDIA said on March 18, 2024 that BNY was the first major bank to deploy an NVIDIA DGX SuperPOD with DGX H100 systems and InfiniBand networking, first used for deposit forecasting, payment automation and predictive trade analytics. BNY describes the on-premise GPUs as a supplement to its cloud solutions.
Is BNY's Eliza tied to one AI vendor?
No. BNY describes Eliza as model agnostic: in 2024 it used OpenAI's GPT-4, Google's Gemini and Meta's Llama, and the Q1 2026 deck lists models from OpenAI, Google and Anthropic. Bloomberg Intelligence described it in August 2026 as a model-agnostic platform and governance layer that keeps BNY in control of data, spending and risk.
How much does BNY spend on technology?
Its 10-K reports software and equipment expense of $2,147 million in 2025, up from $1,962 million in 2024. CNBC put BNY's total technology spending at about $3.8 billion in 2025, roughly 19% of revenue, and the CFO said in September 2026 that BNY spends about $4 billion a year on engineering. AI spending is not broken out.
How much of BNY's code is written by AI?
About 60%, by CIO Leigh-Ann Russell's account: she told Bloomberg Intelligence in August 2026 that 60% of the bank's code is AI generated, and American Banker reported in October 2026 that over 60% of its software code is built with AI's help. All 7,000 developers have the Windsurf AI coding tool.
- Quarterly Update Presentation, 1Q 2026
- The making of a Frontier Firm: How AI is redesigning work at BNY
- BNY Collaborates with Google Cloud to Advance its Eliza AI Platform with Gemini Enterprise
- Engineering The Backbone Of Global Markets At BNY
- BNY 2026 Proxy Statement (Schedule 14A)
- Tech Disruptors: BNY CIO on AI and Bank Modernization (Bloomberg Intelligence podcast)
- Annual Report on Form 10-K for 2025, Exhibit 13.1 (financial section and risk factors)
- Digital employees, AI bootcamps: America's oldest bank is spending billions on tech
- BNY Sees Cross-Selling and AI Investments Fueling Its Next Growth Phase
- Exclusive: How BNY's new AI tool Eliza is minting an army of disposable assistants
- BNY First Global Bank to Deploy AI Supercomputer Powered by NVIDIA DGX SuperPOD With DGX H100
- Exhibit 99.3: Quarterly Update Presentation, 4Q 2024 (Form 8-K)
- The Most Powerful Women in Banking Women to Watch, #18, Leigh-Ann Russell, BNY
- The Bank of New York Mellon Corporation Annual Report on Form 10-K for 2025 (main document)
- While embedding AI 'everywhere,' BNY won't disclose costs
- BNY: AI bet is paying off, client impact is 'measurable'
- BNY, America's Oldest Bank, Signs Multiyear Deal With OpenAI
- BNY looks to Google Cloud for agentic AI push
- BNY builds 'AI for everyone, everywhere' with OpenAI
- BNY gets access to OpenAI, Anthropic advanced cyber models
- Digital Fraud Forces Issuers to Rethink Trust in the AI Era
- AI in the C-suite: BNY CEO and OpenAI board member Robin Vince
- Quarterly Report on Form 10-Q for the period ended June 30, 2026
- BNY skipped the 'tokenmaxxing' craze. Here's what AI metrics it tracks instead
- BNY (BNY) Q2 2026 Earnings Call Transcript
- OpenAI Names BNY, Nubank CEOs to Board Ahead of IPO
- OpenAI appoints two new members to board of directors
- Exhibit 99.3: Quarterly Update Presentation, 2Q 2026 (Form 8-K)
- Agents for financial services
- Kinexys Brings Federated Learning for Fraud Detection
- BNY Mellon (BK) Q1 2026 Earnings Call Transcript
When BNY moves on AI, you'll read it here first.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning