Agentic AI, Quantum Qubits and Tokenised Everything: Finance's New Frontier Just Got Very Real
This week's launches from Google Cloud, JPMorgan, BlackRock and the DTCC show that the technology frontier is no longer where finance experiments. It is where finance operates.
Every quarter for the last decade, someone has predicted that AI, quantum computing or blockchain would remake finance. Every quarter, most people rolled their eyes and went back to Excel. As of this week, the eye-rolling looks harder to justify. Google Cloud is shipping agentic AI purpose-built for banks. JPMorgan and BlackRock are minting tokenised money market funds. Google, IBM and JPMorgan are quietly moving quantum algorithms out of the lab and into pricing models.
None of these stories is brand new on its own. Put them side by side, though, and September 2026 starts to look like a quiet inflection point.
Agentic AI Comes for the Middle Office
The most concrete institutional move came on 25 August 2026, when Google Cloud unveiled Gemini Enterprise for Financial Services, a purpose-built agentic AI stack for banks, insurers and asset managers. The product includes a Google-managed Financial Research agent, more than 50 new skills with specialised agentic instructions for financial roles, enterprise data connectors and an expanding third-party agent ecosystem, all sitting on top of the wider Gemini Enterprise platform.
Coverage from HPCwire's AIwire this month framed it as the first mainstream attempt to package agentic AI specifically for regulated financial workflows, rather than as a generic productivity add-on. That framing matters. Wolters Kluwer, cited in Neurons Lab's 2026 research roundup, expects 44 per cent of finance teams to use agentic AI this year, an increase of more than 600 per cent in twelve months.
From assistants to operatives
Pennant Technologies captured the shift in a September note titled Agentic AI in Lending 2026, arguing that agents have evolved from helpful assistants to "an operative core within lending operations." The Next Web made a similar observation in enterprise finance more broadly, describing a move from generative AI, which drafts and summarises, to agentic AI, which decides and acts.
The most public real-world example remains AT&T's finance organisation, which is using LangGraph to automate manual journal entry preparation under SOX controls. It is not the sexiest use case in the world, but it is exactly the sort of process that a treasury or controllership team would love to hand over. Lloyds Banking Group, in its own September commentary, called 2026 "the year of Agentic AI, and a new era for finance." Coming from a British high street bank not known for hype, that is a strong signal.
Where the risk sits
None of this is without complication. Insentra's 2026 deep dive on agentic AI warned that enterprises are moving faster than their governance frameworks, particularly around auditability of agent decisions. Financial services regulators, from the ECB to the Bank of England to the SEC, are already asking whether existing model risk management frameworks can handle agents that reason across tools and take actions autonomously. Expect a wave of supervisory statements over the coming months.
Quantum Quietly Crosses a Threshold
Quantum computing tends to arrive with either too much hype or not enough. This week's read is more measured. According to a widely shared analysis published on 2 September 2026, IBM has now deployed 433-qubit Condor processors while Google Quantum AI is operating 1000-qubit Willow systems. More importantly, error rates for two-qubit gates have dropped below the crucial one per cent threshold across leading platforms, making quantum error correction viable rather than aspirational.
Google's own surface code error correction breakthrough, first published earlier in 2026, showed that adding more physical qubits to a logical qubit reduces the logical error rate, crossing the break-even point that has haunted the field for two decades.
Finance is already at the table
Financial applications are moving in parallel. IBM's Think Insights team highlighted this month that quantum computing is showing "promising potential" in areas from portfolio optimisation to derivatives pricing. JPMorgan Chase has published research on quantum algorithms for Monte Carlo option pricing that show a roughly 100 times speed-up for certain path-dependent options versus classical methods. Goldman Sachs continues to work on quantum Monte Carlo for risk analysis.
Both IBM Quantum Network and Google Quantum AI are preparing enterprise cloud quantum services with 99.9 per cent uptime commitments. In practical terms, that means large banks will soon be able to procure quantum compute the way they procure GPUs today, on demand and under contract.
What this means for CIOs
The honest answer is: not much this quarter, but a great deal by 2028. Serious quantum advantage in production risk and pricing systems is still a few years away. Yet the difference between banks that have been quietly building quantum-ready algorithms and skills, and those that have not, will start to matter. The story is no longer whether quantum will affect finance. It is which desks and models get retooled first.
Tokenisation Stops Being a Pilot
If quantum is the slow-burn story, tokenised securities are the one moving in real time. On 15 July 2026, the Depository Trust & Clearing Corporation (DTCC) completed its first production trades using tokenised versions of traditional securities, according to CoinDesk. Nearly 40 institutions took part, including BlackRock, Vanguard, JPMorgan, Goldman Sachs and the New York Stock Exchange. Assets in the pilot included tokenised shares of Microsoft, the Invesco QQQ ETF, the SPDR S&P 500 ETF (SPY), the iShares 0 to 3 Month Treasury Bond ETF (SHV) and US Treasuries.
JPMorgan pushed further in May with its OnChain Liquidity-Token Money Market Fund (JLTXX), which maintains blockchain-based token balances tied to investors' ownership records and lets approved users submit purchase, redemption and transfer requests through Ethereum. The underlying blockchain infrastructure is operated by Kinexys Digital Assets, JPMorgan's blockchain unit formerly known as Onyx. TheNextWeb reported that JPMorgan has since filed a second tokenised money market fund on Ethereum.
BlackRock, not to be outdone, filed for a new tokenised Treasury reserve fund with Securitize and proposed to create onchain shares for a seven billion dollar money-market fund earlier this year, according to CoinDesk. In July, the UK Government named BlackRock, Goldman Sachs, JPMorgan and Morgan Stanley to its new tokenisation taskforce.
The numbers behind the noise
The tokenised real-world asset market has grown more than 200 per cent over the past year and now exceeds 30 billion dollars, according to rwa.xyz. Boston Consulting Group has estimated that the RWA market could reach 88 trillion dollars by 2035, comfortably dwarfing today's combined three trillion dollar crypto and stablecoin market.
Those numbers should be read with a healthy dose of caution. Long-range TAM projections rarely survive contact with reality. But the direction of travel is difficult to argue with when the DTCC, the world's largest post-trade infrastructure, is running production tokenised trades with the largest asset managers on the planet.
Three Frontiers, One Story
Look at these three technology waves together and a common thread emerges. Agentic AI is compressing the middle office. Quantum computing is preparing to compress modelling times for risk and derivatives. Tokenisation is compressing settlement, custody and distribution.
In every case, the technology is not just faster. It changes where the decision, the record and the risk actually sit. That is why the winners will not simply be those with the biggest AI budget or the flashiest blockchain proof of concept. They will be the institutions that can rewire operating models, controls and talent around infrastructure they used to outsource entirely.
What to Watch
Three signals to track over the next month. First, adoption metrics for Gemini Enterprise for Financial Services and its competitors from Microsoft and OpenAI, particularly in Tier 1 banking and insurance. Second, further DTCC tokenisation phases and any move by European CSDs such as Euroclear or Clearstream to match them. Third, quantum procurement contracts in financial services, which will be an early indicator of who is genuinely investing rather than just publishing whitepapers.
The frontier keeps moving. This week, it moved a little closer to the trading floor.
