7 min read · 15 sources

The US Open is no longer just a tennis tournament. As of the 2026 main draw, it is also, formally, a prediction market: Kalshi has become the first-ever Official Prediction Market Partner of the US Open under a multi-year exclusive deal with the USTA. Branding goes up on screens and signage, live odds feed into match coverage, and Kalshi will shift prices continuously as points are played. Tennis trading volume on the venue is reportedly up 25x year-over-year.

That legitimacy is the headline. It is also the prompt for everything else in fintech this week, from bank stablecoins to AI-native bookkeeping to a one-click checkout startup asking its old investors for survival money.

Bolt’s valuation collapsed 97% from $11 billion to $300 million, and Breslow is now extracting a pay-to-play lifeline from the same investors who rode it down.

The US Open becomes a betting floor

Source: news.kalshi.com ↗

Kalshi’s deal is more than signage. The company and the USTA are building a joint integrity framework that covers data sharing, education, and market surveillance, with explicit restrictions on markets tied to umpire decisions, injuries, and code violations. The International Tennis Integrity Agency is also in the loop. Tennis is the only major sport where players face strict gambling bans; the new setup pushes commercial betting closer to the athletes than it is in, say, the NBA.

Ben Rothenberg’s take makes the obvious point bluntly: spectators at qualifying matches were visibly betting on phones, sometimes loudly enough for players to hear. Engineers building or competing in prediction-market platforms should read this as a real legitimacy signal - brand-safety acceptance is now an Olympic-grade tennis tournament - and as a real integrity problem, because live in-venue betting on your opponent is qualitatively different from betting from your couch.

Hyperliquid tries to walk through a CFTC door

Source: bloomberg.com ↗

Hyperliquid Labs is in advanced talks with Kraken parent Payward to offer Hyperliquid-linked perpetual futures to US traders through Bitnomial, the CFTC-regulated derivatives venue Payward agreed to acquire earlier this year. If the structure lands, Hyperliquid gets a regulated on-ramp into the US without building one itself. Talks are unconfirmed and subject to regulatory approval, and chatter about a HIP-3 implementation has not been verified.

For trading infrastructure teams, the relevant question is what a regulated Hyperliquid wrapper would look like at the matching-engine and risk-engine boundary. Bitnomial is a futures venue, not a spot exchange; perps routed through it will be a different product from Hyperliquid’s current order book.

Visa bets on transfer learning for A2A fraud

Source: finextra.com ↗

Account-to-account payments are projected to surpass 5.8 trillion by 2028, a 160% increase from 2024, and fraudsters follow the volume. Visa’s A2A Protect uses AI plus transfer learning so a bank gets global risk insight on a transaction immediately, without waiting months for its own models to collect local data and learn what “normal” looks like.

The practical claim is that a new bank customer - or a small bank with thin data - gets the benefit of Visa’s view of fraud across its entire network on day one. For teams running real-time payments rails, the operational consequence is shorter warm-up periods for risk models and a smaller cold-start problem when onboarding new corridors.

Stablecoins pivot from trading to payments

Source: finextra.com ↗

Stablecoins moved $33 trillion in 2025. Only about $390 billion of that - roughly 1% - was real-economy payments for goods and services. The plumbing to mint and move digital dollars is largely solved. The plumbing to spend them at a supplier, send them to family, or buy something without round-tripping through fiat is not.

This is the gap the new wave of stablecoin-payments startups is building against. It is also why the bank consortium below matters: incumbent finance wants to own the spend side of that gap.

Goldman, BofA and 19 friends are forming a stablecoin company

Source: reuters.com ↗

Twenty-one institutions - Goldman Sachs, Bank of America, Citi, Deutsche Bank and 17 others - plan to form a company this year and launch a dollar-backed stablecoin in the first half of 2027. Euro and other G7 currency tokens are expected to follow. The consortium will sit alongside rival banking group Qivalis and in direct competition with Tether.

The history here is unfriendly: bank-issued stablecoins to date have seen limited adoption. The interesting engineering question is whether a consortium token can credibly be both “bank-grade” (regulated, redeemable, audited) and “crypto-native” (cheap, fast, programmable) - those are usually traded against each other, not optimised jointly.

Airwallex launches T:0, an AI-native bookkeeper

Source: airwallex.com ↗

Airwallex has spun T:0 out as a subsidiary led by CEO and co-founder Lance Co Ting Keh. The pitch is autonomous, real-time bookkeeping: the books update as the business earns and spends, revenue recognition is automated, and the platform keeps cash, runway, and burn visible continuously as startups add entities, revenue models, and geographies. It exited private beta and is available immediately to any US startup. Direct competitors are Pilot, Bench, and QuickBooks; the strategic claim is that Airwallex’s payments rails give T:0 first-party transaction data that standalone bookkeepers do not have.

Block's Cash App Score goes external

Source: businesswire.com ↗

Block is making its proprietary Cash App Score available to outside lenders for the first time via Nova Credit’s Cash Flow Intelligence Platform. The score uses real-time Cash App activity - spending, saving, repayments, deposits, peer-to-peer - to supplement traditional credit files. Block claims the score approves 38% more Cash App Borrow customers at the same loss rate as traditional scores. Use cases range from credit cards and auto loans to personal lending, device financing, and tenant screening.

For risk teams, the interesting bit is the data input: the score uses activity that thin-file consumers already generate, which is why it produces approvals rather than rejections. The pricing question is whether lenders will pay per pull, per funded loan, or per approval.

Wall Street banks are renegotiating with Big Law

Source: ca.finance.yahoo.com ↗

Morgan Stanley, Citigroup, and Goldman Sachs are pressing major law firms to share AI-driven efficiency gains through lower rates, competitive tendering, and fixed-fee arrangements. Citi’s Adam Meshel said the bank now asks firms during bidding how much they are saving with AI and expects cost per transaction to drop significantly, with a “different working model” likely within a year. Morgan Stanley’s Eric Grossman said most external legal work would be competitively tendered and paid via alternative arrangements like fixed fees by end of year, calling the associate-hourly leverage model “extraordinarily unstable.”

For legal-AI vendors, this is the procurement signal: outcome-priced contracts, not seat licences, and explicit demands that savings be passed through.

Bolt asks its old investors for survival money

Source: techcrunch.com ↗

Bolt CEO Ryan Breslow is raising up to $27M in a bridge round structured as a convertible note with a pay-to-play clause that dilutes non-participating backers. Breslow himself is committing $5M and expects roughly 100 existing investors to contribute at least $15M combined. Bolt peaked at an $11B valuation in early 2022 before falling 97% to $300M. Breslow returned as CEO in March 2025 and claims the company is nearing profitability, betting on a 60-person, AI-powered team and a new financial super app. For fintech builders, the structural one is the pay-to-play term: it converts a bridge round into a governance test of investor confidence.

Singapore bets S$220M on fintech infrastructure

Source: finextra.com ↗

The Monetary Authority of Singapore will invest US$173M over three years to accelerate AI, blockchain, quantum computing, and digital assets. The program also funds shared infrastructure, attracts global financial institutions, and supports at least 1,000 fintech internships on top of an existing ecosystem of more than 1,800 firms. For teams choosing where to build APAC infrastructure, this is the headline number.

Quick links

Source: finextra.com ↗

• TabaPay closes $155M led by FTV Capital and agrees to buy a Denver-based bank.
• Orange becomes the first merchant in France to offer subscription bill payment with Wero.
• Félix raises $200M in equity and debt (including an $87M Series C) to expand its WhatsApp-based remittance platform for Latino immigrants into a broader set of financial products.
• US unsecured personal loan balances hit a record $281B, up 9.6% YoY. Total US consumer debt hit a record $18.6T in Q2. Fintechs captured 45% of personal loan originations, up 5.2 percentage points YoY, while banks, credit unions, and traditional finance companies all lost share. Subprime originations jumped 29%, but average loan size to subprime shrank 6.8%.