Nebius Group N.V. vs. SoundHound AI: Which AI Stock Will Bring You Profits in 2026 and Beyond?

Nebius (NBIS) reported FY2025 revenue of ~$529.8M (+~350% YoY) and net income of ~$101.7M (net margin ~19.2%), but free cash flow was sharply negative at ~-$3.7B (with stock-based compensation ~21.6% of operating cash flow). SoundHound (SOUN) posted FY2025 revenue of ~$168.9M (+~100% YoY) with a narrowed net loss (~-$14.0M) and a lower P/S valuation (14.5x vs 43.8x for Nebius), while also showing negative free cash flow (~-$103.1M) and legal/competitive risks. Overall, the article frames Nebius as the more “all-in” AI infrastructure bet but highlights significant cash-flow and competition risks, making the outlook cautious despite rapid growth.

Analysis

NBIS is trading like an infrastructure compounder, but the core economic question is whether it can convert capex into durable free cash flow before pricing power normalizes. Accounting profitability is less relevant than utilization and incremental margin because a compute provider with negative FCF can still look self-funding right up until capacity expansion slows; that makes the equity much more sensitive to borrowing conditions and GPU supply than the headline growth rate implies. The cleaner beneficiary of that spend is NVDA, which monetizes every incremental build-out regardless of whether NBIS equity rerates.

SOUN is a different trade: it is a distribution story with real commercial traction, but the moat is fragile if large platform owners bundle voice features into broader ecosystems. The legal/privacy overhang matters less for the court outcome than for enterprise sales cycles; procurement delays and indemnity demands can quietly slow bookings even if revenue growth stays strong. CASY is a potential second-order winner only if voice ordering reduces labor hours or increases basket size enough to justify rollout economics; without that proof, the partnership is more narrative than earnings.

The consensus is probably over-weighting revenue growth and under-weighting balance-sheet self-funding. NBIS looks impressive on growth but still behaves like a capital allocator test, while SOUN looks cheaper on sales but may deserve a lower multiple if platform substitution is accelerating. Falsifiers: sustained positive FCF at NBIS over the next 2-3 quarters, or evidence that SOUN’s deployments convert into multi-year renewals with expanding gross profit dollars rather than just logo count.

Key Decisions for Investors

• Long NVDA / short NBIS for the next 1-3 months: thesis is that AI capex accrues more reliably to the chip supplier than the compute operator. Enter on any NBIS strength that keeps it above the current high-sales multiple; risk/reward is roughly 2:1 if NBIS remains FCF-negative and market rerates the business on cash burn instead of growth.
• Short SOUN on rallies, or use put spreads with a 1-2 month horizon: the setup depends on voice AI remaining a feature rather than a standalone platform. Cover if GOOGL/AAPL-related enterprise announcements show meaningful partner lock-in or if lawsuit risk is resolved in a way that improves procurement confidence.
• Do not chase NBIS long here; wait for evidence of sustained FCF conversion and lower SBC intensity over the next 2 quarters. If reported operating cash flow remains propped up by non-cash add-backs, the stock is vulnerable to multiple compression even if revenue growth stays above 100%.
• Keep CASY on watch rather than initiate a trade until rollout metrics show measurable labor savings or higher ticket economics from voice ordering. Without that data, the customer-side benefit is too hard to monetize and the market may be overestimating near-term upside.