The AI Stocks That Insiders Are Loading Up on for 2026

Salesforce and SentinelOne saw director purchases in mid-December — Salesforce Director David Blair Kirk bought >1,900 shares for more than $500,000 while activist ValueAct added roughly $25 million, and SentinelOne Director Mark Peek purchased nearly $600,000 of stock. Salesforce trades at a forward P/S of ~4.7x and a forward P/E around 17.5x on 2026 estimates and is positioning for agentic AI growth via the Informatica acquisition and its Data 360 initiative; SentinelOne, which grew revenue 23% last quarter, trades below a 4.5x forward P/S and has near-term catalysts including a Lenovo distribution ramp, Singularity Data Lake, and the Prompt Security acquisition. Both names are characterized as beaten-down but potentially undervalued, with insider activity and product/M&A catalysts that could influence investor positioning.

Analysis

Market structure: Salesforce (CRM), Informatica (INFA via acquisition), and platform integrators (e.g., NVDA partners) are the primary beneficiaries as enterprise spend reallocates toward clean master data and agentic AI — this increases SaaS sticky revenue and upsells (expect incremental ARR acceleration of +200–400 bps over 12–24 months if adoption ramps). SentinelOne (S) and Lenovo partnership beneficiaries gain share vs legacy SIEM providers (Splunk/CSCO) through lower-cost data-query offerings; incumbents with hardware-centric licensing (PANW) are most at risk of pricing compression. Cross-asset: a successful AI-agent adoption narrative should compress credit spreads for high-growth software names and increase call demand (higher IV) while modestly lifting risk appetite in equities; FX and commodities impact is negligible.

Risk assessment: Near-term (days–months) risks are execution (integration of INFA/Data 360, product-market fit) and macro-driven valuation multiple compression; medium-term (3–12 months) risks include slower-than-expected customer agent adoption and poor Lenovo rollout metrics. Tail risks: regulatory action on data residency/agentic AI or a material AI-driven breach could re-rate multiples by 30–50%; dependence on hyperscalers (AMZN/MSFT) for compute and on third-party LLM providers is a key hidden dependency. Catalysts to watch: Salesforce product launches and ValueAct activity (next 90 days), SentinelOne quarterly metrics and Lenovo installs over next 2–4 quarters.

Trade implications: Tactical core positions: selective longs in CRM (2–3% portfolio) to capture agentic AI upside over 6–18 months and in S (2–3%) to play valuation gap and specific catalysts; pair long S / short CRWD (equal notional) to express security-platform mean reversion while hedging sector beta. Options: use 9–12 month call spreads on CRM (10–25% OTM) to cap cost and buy Jan-2027 LEAP calls or calendar spreads on S to ride a multi-quarter Lenovo ramp while limiting theta. Time entries in the next 2–6 weeks, scale on product/partner confirmations, trim if CRM P/S >6 or S revenue guidance misses by >200 bps.

Contrarian angles: The market underprices the network effect of master-record data — if Salesforce converts 10–15% of large accounts to paid Data 360 agent seats, incremental margins could be +10–15 pts, which the market hasn’t modeled. Conversely, SentinelOne’s cheap multiple could reflect real moat concerns: incumbents can replicate Data Lake queries and bundle into existing suites, making S vulnerable to price competition. Historical parallels: platform winners often win through sticky data (MSFT, CRM analogs), but losers are those with technology parity and weaker sales motion; unexpected regulatory or large-customer concentration issues remain primary downside triggers.

Key Decisions for Investors

• Establish a 2–3% portfolio long in CRM within 4 weeks, financed by trimming 0.5–1% from broad mega-cap exposure; hedge with a 9–12 month 10–25% OTM call spread (buy lower strike, sell higher) to capture upside if Data 360/agent news arrives, take profits if CRM forward P/S >6 or stock up >30% in 6 months.
• Initiate a 2–3% staged long in S (build over 6–12 weeks) to play valuation catch-up and Lenovo/Prompt Security catalysts; set a protective stop at -25% and reduce to a 1% position if next quarter’s revenue guidance misses by >200 bps, add second tranche on verified Lenovo installations or >20% QoQ uptake in Singularity Data Lake queries.
• Execute a pair trade: long S (2%) / short CRWD (1.5%) sized to be roughly beta-neutral across the cyber software cohort; target convergence or outperformance over 6–12 months, unwind if CRWD cuts guidance or S misses two consecutive quarters.
• Use options to express asymmetric risk: buy 12-month LEAP calls on S (delta ~0.30) sized to 1% portfolio risk or employ calendar spreads around SentinelOne earnings in next 90 days to exploit potential IV skew; for CRM prefer 9–12 month debit call spreads to limit capital at risk while keeping upside exposure.