JPMorgan Chase has picked SambaNova Systems to run on-premises artificial intelligence inference — announced on July 8, 2026 alongside SambaNova’s $1 billion Series F first close at an $11 billion valuation — and the pairing lands in the same week one of Europe’s biggest banks moved its core ledger in exactly the opposite direction. Having tracked bank infrastructure deals through 2026, the split is the story: Intesa Sanpaolo just committed its core banking to Google Cloud, while America’s largest bank is pulling its most sensitive AI workloads back inside its own walls. Bank AI strategy is bifurcating, not converging.
The deal makes JPMorgan Chase the flagship enterprise customer for SambaNova’s SN40L and SN50 Reconfigurable Dataflow Units — purpose-built accelerators designed to run large models without shipping data to a third-party cloud (Finextra). SambaNova says its racks are built to run multi-trillion-parameter frontier models on a single rack for sovereign-cloud, neocloud and enterprise customers including Saudi Aramco (The AI Insider). The Series F was led by General Atlantic with T. Rowe Price, BlackRock and Qatar Investment Authority participating, and follows a $350 million Series E as recently as February 2026 — a near-tripling of ambition inside five months (CNBC).
The competitive responses are already visible. Google Cloud’s counter-argument is sitting in the same news cycle: Intesa Sanpaolo’s decision to run core banking on public cloud in Italy — covered in our report on the Intesa Sanpaolo Google Cloud migration — shows the hyperscalers still winning the systems-of-record layer even as inference walks out the door. Nvidia, whose GPUs dominate cloud inference, has not commented on the JPMorgan deployment. And the banks themselves are hedging: JPMorgan remains one of the largest public-cloud spenders in finance, so this is workload-level repatriation, not a divorce.
“Having JPMorgan Chase decide they’re going to use SambaNova for their inference solution is a big deal. It sends a message to the banking industry that it’s time not to completely depend on cloud services,” said Rodrigo Liang, chief executive and co-founder of SambaNova Systems (IndexBox). Liang also said the company continues to field acquisition interest but expects growth to carry it toward a public listing.
Why it matters: inference — running trained models against live data — is where a bank’s most sensitive material actually touches AI: transaction flows, client positions, fraud signals, credit files. Keeping that on-premises answers the two questions every bank chief information security officer now asks: where does the prompt go, and who else can see the weights? Regulators are asking the same questions from the other side — operational-resilience rules in both the US and EU now treat concentration on a single cloud provider as a reportable dependency, which quietly strengthens the internal case for owned inference hardware. It is the same sovereignty logic that pushed Klarna toward its own US bank charter — owning the critical dependency instead of renting it — and it gives every other systemically important bank a reference architecture to point at in budget meetings.
The prediction from here follows the money. SambaNova’s $1 billion raise is earmarked for supply-chain expansion, which signals order books, not research; expect at least one more top-ten US or European bank to announce an on-prem inference deployment before year-end, because the JPMorgan reference removes the career risk of going first. The losers in that scenario are not the hyperscalers — training and general workloads stay in the cloud — but mid-tier core-banking vendors, who now face banks assembling their own AI stacks chip-up. The agentic layer is racing ahead regardless, as our coverage of Visa taking agentic payments live with 30 European issuers shows: the question for 2027 is whose infrastructure those agents run on, and JPMorgan just gave its answer.
Rick Steves
Rick Steves has seen business and economics through many lenses. He joined the financial services industry in 2009, and has been a financial journalist since 2011. He holds a degree in Business Administration and has experience producing real-time news, from both buy-side and sell-side, as well as for retail traders, brokers and service providers. Steves' work has appeared in a variety of online publications including FX Street, NewsBTC, FinanceFeeds, and The Industry Spread. Rick has great interest in the dynamics of the trading industry. The never-ending clash between technology, economics, regulation, and more importantly, the people.
