Billionaire David Tepper Just Bought These 2 Artificial Intelligence (AI) Stocks That Wall Street Thinks Could Soar by 50% or More

David Tepper's Appaloosa Management initiated Q2 positions in SpaceX and AI-compute provider CoreWeave, reinforcing its technology and AI exposure. SpaceX posted 92% year-over-year Q2 revenue growth to $7.8 billion and carries a $222.32 average price target, implying 50% upside, but its 69.1x price-to-sales valuation leaves little margin for execution setbacks. CoreWeave grew Q2 revenue 112.5% to $2.6 billion and reported a $104 billion backlog, up 245.5% year over year, plus more than $25 billion contracted in Q3; its 5.8x sales multiple and $144.46 target imply a more favorable risk/reward profile despite widening losses.

Analysis

The relevant distinction is not AI exposure but cash-conversion quality. CRWV's contracted-demand profile can support a multi-quarter revenue ramp, yet its economics remain highly sensitive to GPU financing costs, power availability, and customer concentration; incremental growth that requires continuous debt-funded capacity additions can leave equity value subordinate to creditors despite strong top-line execution. The near-term read-through is constructive for NVDA and TSM demand, while power-constrained data-center operators and electrical-equipment suppliers should retain pricing leverage over the next 6-18 months.

SPCX appears priced as a bundled call option on several capital-intensive businesses, creating a material execution-duration problem: a delay in launch cadence, spectrum/telecom approvals, or AI monetization can compress the multiple well before underlying operations deteriorate. Analyst targets are a weak anchor following a volatile IPO because they generally extrapolate growth rather than assign a cost of capital to regulatory and engineering risk. Appaloosa's disclosed ownership is also not a timely catalyst; it is backward-looking positioning, not evidence of ongoing buying.

Consensus is likely underestimating that AI infrastructure demand can remain strong while AI-infrastructure equity returns diverge sharply. If hyperscalers shift procurement toward owned capacity, or if GPU supply loosens faster than cloud pricing resets, CRWV's utilization may stay high but returns on deployed capital could fall. Conversely, a sustained shortage of powered data-center capacity would make CRWV's backlog more valuable than its current earnings profile suggests; the key evidence is quarterly backlog conversion, contracted pricing, and net debt per incremental MW rather than headline backlog growth.

Key Decisions for Investors

• Initiate a 3-6 month long CRWV / short SPCX pair at equal dollar exposure only after verifying SPCX listing, float, and borrow availability. The pair isolates contracted AI-capacity monetization from long-duration, multi-business execution risk; target 20-30% relative performance, with a stop if CRWV reports backlog conversion below expectations or SPCX demonstrates accelerating profitable recurring revenue.
• For outright CRWV exposure, use a staged entry around quarterly results rather than chase sentiment. Add only if revenue growth is accompanied by improving operating leverage, stable utilization, and declining net debt-to-forward EBITDA; reduce if capacity capex or interest expense rises faster than backlog conversion. A 15-20% downside allowance is appropriate given financing and customer-concentration risk.
• Maintain a 6-12 month long NVDA or TSM basket as the higher-quality second-order expression of sustained compute deployment, paired against a modest short in a broad high-multiple software/AI ETF if rates rise. This captures infrastructure demand while reducing single-name balance-sheet risk; reassess on hyperscaler capex guidance and GPU lead-time normalization.
• Set an alert for CRWV disclosures on customer concentration, take-or-pay cancellation rights, power interconnection delays, and debt maturities. Any material weakening in these items falsifies the backlog-as-valuation-floor thesis and converts CRWV from a buy-on-execution setup to an avoid/short candidate.