Tabby's $6.5bn Round, Goldman's Tokenised Bricks and the Quiet Rewiring of Credit
As of this week, the lending stack looks less like a product line and more like a software platform. BNPL is growing up, property is going on-chain, and AI is rewriting the loan file.
The credit business has spent the last decade insisting it was becoming a technology business. In the first days of October 2026, that claim finally has receipts. A Saudi buy now, pay later giant has closed one of the biggest fintech rounds of the year. A Wall Street bank has quietly plumbed real estate funds into a blockchain. And a cohort of AI-native lenders is reporting approval lifts that would have been laughed out of a credit committee three years ago. Taken together, this is not a set of one-off announcements. It is a stack rewiring itself in real time.
Tabby's Series F: BNPL, but grown up
Start with the headline that will not stop echoing across the Gulf. On 22 September 2026, Riyadh-based Tabby closed a $233m Series F led by Blue Pool Capital, with participation from Arbor Ventures, Wellington Management and HSG, valuing the company at $6.5bn, according to Wamda and Yahoo Finance. Tabby now reports over 15 million registered users and partnerships with more than 40,000 sellers across Saudi Arabia and the UAE, on trailing revenue of $378m.
What makes the round interesting is not the multiple. It is the pivot. Tabby is not raising to sell more instalments on trainers. As Fintech Futures reported in its September round-up, the capital is earmarked for consumer and SME lending, digital wallets, accounts, cards and money transfers. In other words, BNPL has become the front door to a full financial services suite. That is the same story Grab told when it bought Atome, and the same story the embedded finance market is telling as it races from $94.42bn in 2025 to $115.03bn in 2026, per Coherent Market Insights.
The subtext: BNPL is no longer a category. It is a customer-acquisition wedge for a broader credit and money-management business. The regulators sceptical of shadow credit now have a very different animal to supervise: licensed, capital-heavy, and increasingly indistinguishable from a bank.
AI loan origination stops being a pitch deck
While Tabby was hoovering up mind-share, a quieter revolution was happening inside credit unions and community lenders. According to hesfintech, Casca (Cascading AI) closed a $29m Series A led by Canapi Ventures earlier this cycle, and Zest AI expanded its CU Lending Collective with Commonwealth Credit Union in February 2026 to bring AI credit scoring to smaller institutions. Scienaptic AI's Temenos integration, delivered in April, was engineered so that credit unions could add AI decisioning without a core migration.
The performance numbers, if they hold, are eye-watering. Zest AI reports a 25 per cent lift in approvals with no added risk, and models that reduce defaults by 20 per cent at constant approval. Some underwriting AI vendors are showing approval-rate gains of 18 to 32 per cent alongside bad-debt reductions above 50 per cent. Origence, per hesfintech, funded $62bn in 2025, up 17 per cent year on year, across 89 new or expanded credit-union contracts.
The tokenised bricks arrive on institutional balance sheets
On 4 June 2026, Goldman Sachs partnered with Apex Group and Archax to launch a tokenised real estate fund, using Goldman's GS DAP blockchain platform, per Bisnow. Archax serves as custodian; Apex provides AIFM and fund administration through Fundrock LIS and Apex Fund Services Luxembourg; LRC Group is fund manager; Ownera is the liaison between participants and distribution channels.
September 2026 pushed that model further. Apex Group advanced additional work on tokenised real estate funds. Tribe Tokenisation announced a partnership with TheBlock to bring real estate tokenisation to Dubai brokers. A widely circulated case study from Wroclaw described a 42-unit multifamily building whose tokenisation cut closing costs by 28 per cent and shortened settlement to a matter of hours.
What this week actually means for lenders
Three separate stories, one rewiring. The lending stack of 2026 combines a BNPL-shaped acquisition wedge (Tabby's Series F), an AI-native underwriting brain (Zest AI, Scienaptic, Casca) and a tokenised asset layer (Goldman, Apex, Archax). Each piece can now be procured from a serious institutional partner. That was not true 18 months ago. The SEC has signalled it is willing to give tokenised assets room to breathe, the AI lenders have the approval and default numbers to back their pitch, and BNPL players are quietly turning into digital banks in everything but the branding.
For consumers, credit will feel less like an application and more like a decision baked into checkout, a mortgage journey or a property purchase. For SME borrowers, especially in emerging markets, the combination of embedded lending and AI origination is likely to compress underwriting times from weeks to minutes. For investors, the pool of yield-bearing, on-chain real-world assets is on track to include serious institutional real estate for the first time.
As of this week, that is not a forecast. That is the running system.
