What Are 2 Great Tech Stocks Flying Under the Radar Right Now?

SoundHound's revenue nearly doubled in 2025 and the company is nearing adjusted EBITDA profitability after acquiring Amelia to create a voice-first agentic AI customer-service platform; the stock is down ~40% YTD. AppLovin reported revenue up 66% last quarter, gross margins +420bps to 88.9% and S&M down 21%, and is expanding its Axon-2 adtech with a self-serve ad manager, international rollout and e-commerce initiatives despite its stock being down >40% YTD.

Analysis

The pivot toward natural-language, voice-driven automation creates a bifurcated opportunity: firms that own verticalized conversation stacks and customer data (CRM, contact-center incumbents and their integrators) can capture most early-dollar value, while pure-play ASR/voice-layer vendors will be paid as inputs unless they bundle task-completion flows. That implies a winner-takes-most dynamic on enterprise deal economics — once an integrator proves a 20–40% FTE productivity uplift in a single business unit, expansion tends to accelerate across the enterprise within 2–4 quarters, but initial pilots typically take 6–12 months to convert to meaningful ARR.

Upstream, the compute demand impulse is nuanced. Large-scale inference can raise demand for datacenter accelerators, but rapid moves to lightweight on-device inference and specialized ASICs will shift mix away from general-purpose datacenter GPUs over 12–36 months. That creates an asymmetric payoff for companies positioned to sell inference silicon, edge MCUs, or voice-optimized stacks to OEMs and telcos — and a downside risk for vendors that assume perpetual high-margin datacenter GPU demand.

Two practical risk vectors can reverse the narrative: privacy/regulatory constraints that force on-prem or anonymized deployments (slowing SaaS monetization), and model-behavior liabilities (hallucinations) that increase enterprise legal/insurance costs and stall procurement. Near-term catalysts to watch are: 1) first 6–12 enterprise production rollouts converting from pilot to multi-site contracts, and 2) announcements from hyperscalers to bundle comparable voice-agent products — either could materially re-rate multiples within 3–12 months.

Key Decisions for Investors

• Small asymmetry long: establish a 1–2% NAV position in SOUN (or SOUNW if liquidity preferable) via long-dated calls (12–18 month LEAPS). Hedge with 25–30% notional in 12-month puts to cap downside. R/R: asymmetric upside (target 2.5x) vs capped downside to ~-25% net if pilots fail; time entry on any >10% pullback.
• Growth + risk-managed exposure to APP: buy APP stock sizing 2–3% NAV and fund by buying modestly OTM 9–12 month protective puts (cost ~2–4% of position). Thesis: continued margin expansion vs customer-concentration risk; target +40–60% upside in 12 months, stop-loss at -25% post-hedge.
• Hedge compute-consolidation tail: purchase a 9–15 month NVDA put spread (buy 1x nearer-term put / sell 1x lower strike put) sized to cover ~30–50% of our long AI exposure. Cost-efficient insurance if hyperscalers internalize voice stacks and compress GPU demand; protects against >20% NVDA drawdown at limited cost (~3–6% of notional).
• Catalyst-watch trading rule: if an enterprise announces multi-site production rollouts (3+ regions) from a vendor, add to that vendor up to +1% NAV within 30 trading days. Conversely, if a major hyperscaler launches a free/bundled voice-agent product with deep CRM integrations, reduce exposure to independent voice-stack vendors by 50% within one week.