6 min read · 11 sources

S&P Dow Jones Indices is partnering with Kalshi to clear binary event contracts against official S&P index values. Under the agreement, Kalshi gains access to benchmark cash index settlements, while S&P will ingest Kalshi’s order-book probabilities to create commercial data feeds covering macroeconomic metrics, political outcomes, and sports.

The deal transforms event markets from retail speculation into institutional settlement infrastructure. Prediction platforms have spent years battling state gaming regulators and the CFTC over product legality. Tying contract expiries directly to Wall Street’s primary benchmark administrator gives the asset class a deterministic pricing anchor and integrates event contracts straight into institutional data terminals.

The engineering lift shifts prediction markets onto traditional market data architecture. Instead of bespoke oracle resolution or manual committee sign-offs, Kalshi contracts will settle against published index ticks via deterministic data pipelines. For trading desks, it bridges raw event risk into risk management systems that already ingest standard S&P Dow Jones feeds.

Anthropic generated $16.3 billion in revenue in the first half of 2026, up 1,300% year over year, with more than $6.6 million in trailing revenue per employee.

Anthropic's S-1 Shows $16.3B in First-Half Revenue

Source: mostlymetrics.com ↗

Anthropic has filed its S-1 registration statement to list on the Nasdaq under the ticker ANT. The filing reveals the company is seeking an IPO raise of up to $100 billion at a valuation approaching $2 trillion, up from its previous private mark of $965 billion. FY2026 total revenue reached $19.7 billion, driven by an explosive second quarter that generated $11.6 billion alone - a 1,368% jump year over year.

The underlying metrics demonstrate unprecedented operating leverage for software. Anthropic posted roughly 900% net revenue retention, generated over $6.6 million in trailing revenue per employee, and logged an operating profit in Q2 2026. Roughly 80% of top-line revenue is consumption-based pay-for-intelligence API usage rather than fixed seat licensing.

The structural risks sit entirely in the infrastructure layer. The S-1 details tens of billions in forward compute commitments to cloud providers alongside hyper-concentrated infrastructure dependencies. Anthropic faces minimal enterprise switching barriers at the API layer, meaning retention relies on continuous frontier model dominance while running massive capital expenditure cycles.

Ripple Secures National Bank Charter and Inks Robinhood Settlement Deal

Source: bankingdive.com ↗

The Office of the Comptroller of the Currency (OCC) has granted Ripple unconditional charter approval to form a national trust bank. The charter restricts Ripple from retail deposit-taking or commercial lending, limiting its operational scope to digital asset custody and reserve management for its RLUSD stablecoin. Simultaneously, Ripple expanded its integration with BNY Mellon to clear tokenized trade finance and let institutional clients mint and redeem RLUSD directly against custody reserves.

At the same time, Ripple partnered with Robinhood to deploy RLUSD as a settlement asset. Robinhood will list the digital dollar while integrating it into back-office treasury routing, trade settlements, and financing rails across Ripple Prime and Bitstamp.

This dual-track strategy bypasses correspondent banking plumbing. By using a federal trust charter to back an enterprise stablecoin, Ripple gives retail brokers like Robinhood a 24/7 alternative to Fedwire and clearinghouses for intra-entity float. For fintech infrastructure teams, it converts stablecoins from speculative trading instruments into a balance-sheet settlement layer that clears transactions outside standard banking windows.

Agents Get Native Wallets via Ghost Cards and Stablecoin Rails

Source: linkedin.com ↗

Autonomous software agents are receiving their own commercial payment rails. A recent implementation demo from personal assistant Instinct showed an agent spending $700 autonomously from a restricted balance. The pipeline holds stablecoins inside an embedded Privy wallet on the Tempo network, off-ramps liquidity to USD via Stripe, routes the fiat balance to a prepaid custodial account at Lead Bank, and issues a single-use Visa virtual card to execute merchant checkout.

The model highlights a broader shift toward what Fintech Brainfood describes as ghost cards. While Stripe relies on ephemeral, just-in-time virtual card issuance, networks like Visa and Mastercard are shipping competing frameworks - Visa Intelligent Commerce and Mastercard Agent Pay - that tokenize an existing consumer payment credential directly into agent runtime environments.

This introduces new friction at the fraud layer. With consumer apps like Meta’s Muse capturing 10.2 million US downloads in roughly 40 days at a $200 average order size, fraud engines are retooling for machine-speed transactions. Platforms like Sardine are deploying autonomous policy agents directly at the gateway to evaluate prompt lineage, programmatic velocity, and synthetic identity markers without stalling API authorization rates.

Ramp Deploys Collaborative Coding Agent Across Internal Engineering

Source: x.com ↗

Ramp has opened access to Inspect, an internal coding agent that automates software deployment across its engineering, design, and product organizations. The system monitors input channels across Slack, Linear, and GitHub, parses natural language requirement specs, builds the feature branch in a sandbox environment, and serves back a tested, visually verified pull request.

Ramp reports that 99% of its product and engineering staff now use Inspect, with the system generating 74% of all merged production PRs across the organization. The deployment has correlated with a 3.4x year-over-year increase in merged pull requests per engineer.

The platform relies on Ramp Router, an internal routing layer that balances task complexity against LLM inference pricing and latency budgets. By abstracting away local development setups and manual branch rebasing, the workflow changes software development bottlenecks: capacity is no longer determined by lines of code written, but by how fast engineers can audit diffs and triage execution boundaries.

Guidewire Buys Hyperexponential for Up to $2.1B

Source: guidewire.com ↗

Insurance core platform Guidewire has signed a definitive agreement to acquire hyperexponential (hx) for $1.7 billion in cash, alongside up to $400 million in performance earnouts. London-based hx builds pricing, modeling, and actuarial software tailored for property and casualty (P&C) carriers, currently calculating risk pricing across more than $75 billion in annual gross written premium.

The transaction signals a structural change in specialty insurance infrastructure: the shift to “agentic underwriting.” Enterprise actuaries typically model risk across fragmented Python models, Excel sheets, and disconnected legacy pricing databases. Integrating hx directly into Guidewire’s transaction engines lets AI systems ingest submission documents, cross-reference historical loss distributions, and generate live actuarial pricing quotes without manual ingestion delays.

Hong Kong Card Issuer RedotPay Eyes $5B US Listing

Source: bloomberg.com ↗

Hong Kong-based crypto card provider RedotPay is preparing a confidential US IPO filing targeting a valuation between $4 billion and $5 billion before year-end, with JPMorgan and Goldman Sachs advising. The company claims more than 9 million users and roughly $180 million in annualized run-rate revenue by providing physical and virtual cards that liquidate crypto balances at payment terminals.

The planned public offering arrives despite material friction. RedotPay is navigating an ongoing $473 million legal dispute with Binance, which complicates its cross-border custody audits. If the filing proceeds, SEC disclosures will provide one of the first deep-dive views into interchange economics, AML/sanctions compliance costs, and banking off-ramp relationships for cross-border stablecoin card operations.

Venture Rounds and Market Velocity

Source: bloomberg.com ↗

Enterprise finance agent developer Hyphenate (formerly Maximor) closed a $45 million Series A led by Nexus Venture Partners, with participation from B Capital. The round pushes the startup’s total funding to $54 million. The company builds autonomous finance agents designed to plug into enterprise ERPs, reconciliation tables, and ledger databases to run enterprise accounting without manual spreadsheet hygiene.

The growth funding sits against an otherwise depressed market. Global fintech funding dropped 29% year over year to $11 billion in Q3 2026. Deal counts dropped to 650 transactions globally - the lowest deployment volume recorded since Q3 2024. Capital remains strictly bifurcated: horizontal payment commoditization is being starved of capital, while capital-intensive AI infrastructure, agentic execution layers, and bank-chartered digital dollar rails capture remaining institutional allocations.