2 Popular AI Stocks to Sell Before They Fall 53% and 58%, According to Wall Street Analysts

Jefferies and Morningstar see substantial valuation risk in two popular AI stocks: Jefferies' $80 Palantir target implies 53% downside from $172, while Morningstar's $62 SpaceX target implies 58% downside from $150. Palantir posted Q2 revenue growth of 93% to $1.9 billion and non-GAAP EPS growth of 156% to $0.41, but trades at 68x sales and faces concerns over insufficient reinvestment. SpaceX grew Q2 revenue 92% to $7.8 billion but generated negative free cash flow of $25 billion in H1 2026; its 94x sales valuation assumes successful execution across Starship, AI compute and orbital data centers.

Analysis

PLTR’s primary vulnerability is not a near-term revenue miss but duration risk: at an extreme revenue multiple, even sustained high growth can produce a flat-to-down stock if forward growth decelerates or investors demand a lower terminal multiple. The market is likely underwriting both unusually durable government/enterprise AI adoption and continued operating leverage; incremental hiring, sales investment, or customer-concentration disclosure would challenge that combination. A relative de-rating versus profitable software peers can occur over 1-3 months around earnings even if reported results remain strong.

The more consequential competitive question is whether enterprise AI orchestration becomes a differentiated control plane or a feature bundled into hyperscaler, model-provider, and data-platform contracts. If buyers increasingly procure AI through existing cloud commitments, GOOG, MSFT and AMZN can capture more of the economics while standalone application-layer vendors face lower pricing power. Conversely, security, governance and deployment friction could preserve PLTR’s premium for longer than valuation-focused shorts expect; the thesis is falsified by continued net-new commercial customer acceleration alongside stable or expanding contribution margins.

SPCX should be treated as a liquidity and verification problem before it is treated as a fundamental short. The stated compute, launch and cash-flow assumptions require independent confirmation, and any private/limited-float trading vehicle can remain disconnected from valuation for extended periods. If the investment cycle is real, second-order beneficiaries are terrestrial AI infrastructure suppliers—NVDA, power equipment and data-center operators—because capacity spending occurs before speculative orbital economics are proven; failure of the orbital thesis would redirect rather than eliminate much of that capex.

Key Decisions for Investors

• Initiate a 1-3 month PLTR relative-value short versus GOOG, sized beta-neutral: short PLTR / long GOOG. Target a 15-25% relative move on multiple compression; cover if PLTR raises forward revenue growth expectations while maintaining operating-margin expansion, or if the pair underperforms by 10%.
• For long-only exposure, avoid adding PLTR ahead of the next earnings print unless implied revenue-growth durability is supported by bookings, remaining deal value and commercial-customer data. Reassess after guidance: a revenue-growth deceleration without a corresponding upward investment plan is the key downside catalyst.
• Do not recommend a directional SPCX short until venue, float, borrow availability, financial statements and cash-flow claims are independently verified. Create an event-driven alert for financing, launch-reusability milestones, or audited segment disclosures; these determine whether dilution/liquidity risk becomes tradable.
• Maintain NVDA as the cleaner AI-capex expression rather than underwriting speculative space-compute economics. Trim only if hyperscaler capex guidance weakens materially; the key 6-18 month risk is that AI capacity utilization fails to justify continued infrastructure spending.