LSEG buys Octus, Tradeweb buys Trumid, Standard Chartered acquires custodian Zodia, and more

It’s suddenly fall here in New York (that’s autumn in the rest of the world). Leaves are beginning to change color and fall off the trees, and temperatures are dropping. But they’re rising in the fintech world amid acquisitions that see LSEG strengthening its position in fixed income and narrowly averted possible data breaches. For more, read on …

Announced this week

LSEG to buy private fixed-income data provider Octus

The London Stock Exchange Group has struck an agreement to buy Octus, a provider of credit data and workflows established in 2013, for $3.535 billion in a bid to create a leader in the high-growth areas of credit and private markets. By combining Octus with LSEG’s fixed-income data and analytics, clients will be able to source deals, manage risk, and run portfolios through LSEG’s platform.

Kent Collier, founder and CEO of Octus, will continue to run the business, reporting to Gianluca Biagini and Ron Lefferts, the co-heads of LSEG’s Data & Analytics division.

The deal is subject to regulatory approval but is expected to close in the first half of 2027.

Tradeweb announces plan to buy Trumid

Tradeweb proposed an acquisition of Trumid, a credit trading platform, for almost $2 billion.

Tradeweb officials say the combination will expand Tradeweb’s multi-asset platform across the bond lifecycle, from trade inception to the secondary market, on one platform, creating a more efficient way to access liquidity.

Combining the two will enable the sharing of Tradeweb’s AllTrade with Trumid’s Swarms and Attributed Trading, allowing access to broader liquidity pools; extend Tradeweb earlier into the credit lifecycle, with grey markets and new issue trading; and expand Tradeweb’s penetration into other market participants with the addition of Trumid’s hedge funds and systematic traders.

Standard Chartered to acquire Zodia Custody

Standard Chartered has agreed to acquire Zodia Custody, the parent of regulated operating entities in Singapore, Hong Kong, Australia and the United Arab Emirates and Luxembourg. The bank will integrate the business into its existing digital asset custody services, supporting plans to integrate traditional and digital asset securities services for institutional clients. Zodia will separate its digital assets services and professional services business, while its Zodia Solutions business will continue to offer infrastructure to financial institutions.

S&P expands AI for private markets

S&P has expanded its AI capabilities across its private markets data, including iLevel additions that embed AI into private markets workflows and enable clients to connect to tools such as Copilot and Claude via a Model Context Protocol server. The additions give clients greater flexibility with AI tools while retaining existing permissions and access to iLevel.

Solve previews New Issue Monitor tool

Solve, a provider of pre-trade data for fixed-income markets, has showcased a New Issue Monitor tool to portfolio managers, traders, analysts, and sales and trading desks to give users a view of primary and secondary markets together. Putting them together brings new deals into Solve Quotes to monitor deal flow, showing each deal with its tranches beneath it, and allowing firms to monitor deal flow in a single place, following each stage, and screening deals by size, weighted-average life, credit enhancement, and structure, and allowing clients to trade bonds through their lifecycle.

Beeks, OmniPhi launch on-site cloud AI

Cloud connectivity provider Beeks Financial Cloud group has announced an investment and partnership with OmniPhi to launch a new business line, Beeks Sovereign AI, an AI solution that runs on infrastructure owned by client firms. Combined with Beeks’ GPU-enabled Proximity Cloud infrastructure in a co-location site or a client’s own site, and OmniPhi’s agentic harness that manages AI agents across research, strategy, risk and execution, the solution allows firms to use agentic AI platforms without routing any data through public infrastructure, bypassing sovereignty and compliance hurdles that prevent firms from deploying AI at scale.

Versana, Allvue ally for loans, private markets trading, data

Digital infrastructure provider Versana has partnered with private markets technology vendor Allvue Systems to connect the vendors’ platforms in the $9 trillion broadly syndicated loan (BSL) and private credit markets, enabling straight-through processing on Versana’s loan data and Allvue’s investment accounting system, which officials say will result in more accurate reporting, streamlined workflow and lower costs, and will enable market-wide scalability. Allvue’s Trade Order Management system will also deliver trader data to Versana, enhancing the vendor’s digital loan offerings.

What you might have missed from us

Examining Google’s jump into financial research with enterprise Gemini

Nyela Graham records how Deutsche Bank and Google co-designed multi-purpose agents which led to the development of a financial research agent sitting within Google’s Gemini for Financial Services framework. Launched in August for corporate and capital market clients, it was trained on 50 “foundational skills” including know-your-customer screening, sales, and risk management, among others.

SIX is preparing its data for clients that aren’t people

Swiss exchange operator SIX Group is making its market data more consumable by non-human clients such as AI agents, reports Eliot Raman Jones. This includes every aspect relating to a financial instrument, including those covered by its Valordata Feed of more than 32 million assets, and ensuring these are described similarly and mapped to the same sources.

Fraud, breach of contract suit against LSEG can proceed, judge rules

MayStreet founder Patrick Flannery’s lawsuit against the London Stock Exchange Group, can proceed, rejecting an argument by the LSEG to dismiss the case. The lawsuit alleges fraud and breach of contract by LSEG after its 2022 purchase of MayStreet. An LSEG spokesperson says the exchange continues to dispute Flannery’s allegations.

AI assistants for human overseers? Or human checks and balances for AI actors?

At a recent Bloomberg conference, executives outlined their advances in the AI space and highlighted the challenges they and others have faced so far in dealing with the new technology; I report in my latest Waters Wrap. Topics covered include what the endgame of heavily regulated AI use is, and whether users believe AI will replace us all.

Barclays builds a system to spot AI agents that might go rogue

Rob Mannix reports how Barclays quants have built a monitoring system that can read the minds of AI agents and catch them before they do anything too risky. For banks exploring AI agents for coding, research and customer compliance, after-the-fact analysis of decision quality does little to mitigate risk or save costs. But in tests, the setup predicted when large language models would call a tool and could identify it as low, medium, or high risk about 90% of the time.

Waters Wavelength Ep. 360: Innovation, regulation, and exposure

In this episode of the Wavelength podcast, Tony and Wei-Shen discuss how regulators are approaching AI governance. Simply put, it’s worrisome, especially in the US.

In other news

AI just found a bug that puts up to 70% of the world’s websites at risk, The Wall Street Journal

We hear so much about the dangers of AI. Well, here’s an example of how it can be helpful—and if your organization has a website, you may have already heard about this. Robert McMillan of the Wall Street Journal reports that the makers of PHP, a coding language powering up to 70% of websites, warned users Wednesday that cyberattackers could harness its software to steal information or launch cyberattacks.

It’s “the most severe AI vulnerability that was ever found,” and probably wouldn’t have been discovered without AI, said one executive involved. The problem is that hackers now also use AI that could find that vulnerability; long-overlooked bugs hiding in PHP code, dubbed PHP2Shell and considered too difficult to exploit, could now be exploited—and a hack could take seconds on smaller systems and hours on more complex ones.

As a result, a group fixed 200 files and wrote more than 10,000 lines of code. A corrective patch was scheduled for October 22, but it was rushed out early after news about the patch and the reasons for its release leaked.