OakNorth reports £223M pre-tax profit, US drives 40% of lending

OakNorth reported pre-tax profit of £223m for 2025, with gross originations up 33% to £2.8bn and total facilities rising 18% to £7.2bn. Efficiency ratio improved to 26% from 29% (a ~10% operating leverage improvement), and gross revenue was £605.9m; the bank cites AI implementation in credit decisions and continued US expansion (40% of originations) and has a pending acquisition of Community Unity Bank subject to regulatory approval.

Analysis

AI-driven automation in credit underwriting is a multi-step demand driver for the tech stack, not just a spike in software spend. Lenders that prefer in‑house model governance and low-latency inference will push spend toward on‑prem and co‑located server vendors, expanding the addressable market for high-density OEMs by a low-double-digit percentage over 12–24 months. Adtech/monetization platforms that can fold first‑party behavioral signals into real‑time bidding will extract higher ARPU per user, creating a visible revenue kicker in the next 2–4 quarters if CPI/CPM conditions hold.

Key reversal risks are model performance and oversight: a single high‑profile misclassification, data breach, or a regulatory enforcement action focused on algorithmic credit decisions can force conservatism across the sector and dump hardware orders within weeks. Macroeconomic tightening or an ad spend pullback are shorter‑horizon catalysts (0–3 months) that would compress multiples quickly; structural validation of credit models plays out over 12–24 months and is the main slope for rerating. Operational providers face inventory and back‑log cyclicality — strong quarterlies can be followed by order softness if hyperscalers reallocate capex.

The consensus is placing too much weight on near‑term efficiency gains and not enough on integration, data quality and regulatory lag. That makes hardware and ad‑tech names attractive on disciplined pullbacks but vulnerable to binary news events. Watch three concrete signals as decision triggers: OEM server order flow (monthly/quarterly reported), measured uplift in lender loss rates attributable to model changes (12–24 months), and any regulatory guidance or investigations — these will be the clearing events that decide winners and losers.

Key Decisions for Investors

• Buy SMCI (long equity or 12-month calls) to capture secular on‑prem AI server demand from financial services and mid‑market lenders; target 40–80% upside over 6–12 months if order flow and OEM announcements accelerate; set a tactical stop at 25% to limit inventory/cycle risk.
• Buy APP (AppLovin) equity or 9–12 month calls to play improved ARPU from AI-driven personalization in mobile monetization; target +30–50% in 6–12 months if CPMs stabilize and ARPU beats; hedge with a 20–25% trailing stop given ad‑spend cyclicality and privacy/regulatory tails.
• Pair trade — Long SMCI / Short GS (equal dollar, 6–12 month horizon) to express tech‑enabled disintermediation of mid‑market lending while hedging broad financials exposure; target 1.5x return on the spread if hardware adoption outpaces bank fee growth, stop the pair if spread moves 15% against position.
• Risk hedge: Buy out‑of‑the‑money puts on SMCI (3–6 month) as cheap insurance against a fast commodity‑like drawdown in server demand triggered by hyperscaler reallocation or regulatory shock; treat this as portfolio tail protection rather than a P&L driver.