2 Under-the-Radar Growth Stocks That Just Got Cheaper Thanks to the Market Sell-Off

Equity indices are weaker amid Middle East tensions and higher oil prices (S&P 500 ~9% off its high; Nasdaq-100 down ~11%), creating a buy-the-dip backdrop. Lemonade reported ~3.0M customers at end-2025 (+23% Y/Y), Q4 in-force premium $1.24B (+31% Q4), full-year revenue $738M (+40% Y/Y), gross loss ratio improved to 64% (target 75%), and management guides revenue +61% to $1.19B in 2026 (forward P/S ~3.9). Datadog generated record $3.43B in 2025 (+28% Y/Y), ~5,500 customers using AI products (+57% Y/Y), MCP usage up 11x QoQ, and the stock trades at a P/S of ~12.1 near post-IPO lows—both names are presented as long-term buy opportunities during the sell-off.

Analysis

Datadog’s LLM/observability extension creates a structural data-gravity moat: once telemetry and model context are piped into a single billing/alerting plane, customer switching costs rise materially because observability becomes embedded in CI/CD and incident-response playbooks. That gives Datadog an outsized opportunity to expand ARPU via metered AI observability while retaining high gross margins, but the window to capture this is 12–24 months before cloud hyperscalers or large APM incumbents push integrated alternatives.

Lemonade’s AI-driven cost curve is real and scalable at the unit level, yet the more interesting second-order is balance-sheet sensitivity as policies scale. Rapid IFP growth magnifies exposure to reinsurance cycles and catastrophic aggregation; a step change in reinsurance pricing or a single concentrated catastrophe would compress returns on capital far faster than top-line deceleration would indicate, so valuation should be modeled as a function of loss-ratio volatility and capital access, not only premium growth.

The current geopolitical-driven volatility is shortening multiple expansion tailwinds for growth names over the next few quarters, creating attractive entry points but also elevating macro tail risks (oil/insurance/reinsurance feedback loops). In this regime, asymmetric optionality structures (long convexity on secular winners, capped exposure to binary insurer shocks) dominate straight long equity exposure.

Net-net: treat Datadog as a 12–36 month “platform monetization” play with durable downside protection, and Lemonade as a high-beta optionality bet that needs insurance-specific hedges. NVDA remains the end-market anchor for AI compute but is not the highest idiosyncratic upside per dollar in this setup; prefer software leverage to pure-play hardware here.

Key Decisions for Investors

• Buy DDOG (size 2–3% portfolio) long-term: accumulate shares or buy Jan 2028 LEAP calls (delta ~0.6) with a 12–36 month horizon. Risk/reward: if Datadog converts a meaningful portion of LLM observability into recurring ARR, expect 1.5x–3x upside; set a tactical stop at -25% and trim into strength. This expresses sticky, meterable AI telemetry expansion.
• Tactical small-weighted long in LMND (size 0.5–1%): prefer Jan 2027 LEAP calls or stock purchased with a 6–9 month protective put (cost financed by selling short-dated calls). Rationale: captures asymmetric upside if unit economics scale while capping downside from a reinsurance or catastrophe shock. Target 2–4x upside vs potential 50% downside in a severe underwriting cycle.
• Pair trade: long DDOG / short INTC (equal notional, net zero cash) over 12–24 months. Thesis: duration of software-based observability monetization > visibility of legacy silicon cycle. Expect relative outperformance of 1.5–2x vs the market if AI shift accelerates; cut if DDOG customer metrics plateau for two consecutive quarters.
• Risk-off hedge: during sustained Mideast volatility, buy downside protection on growth-heavy exposure via SPY 3–6 month puts sized to cover 5–10% portfolio risk or buy NVDA 3–6 month puts as an expensive-but-concentrated hedge if GPU funding/compute budgets roll back. These options act as insurance against macro-driven multiple compression that would disproportionately hit LMND and DDOG in the near term.