AI stocks that still look like good value

Conviction: 68% · Horizon: 4Y · 2026-08-18
Ouster can scale to hundreds of millions of EBITDA by 2030

A 40% revenue CAGR with 30% EBITDA margins produces about $285M of 2030 EBITDA, which at a 25x multiple implies a $7.1B enterprise value if lidar keeps penetrating autonomy, robotics, and infrastructure.

Instrument Side Target Reason
OUST Long Forty percent annual growth and 30% EBITDA margins can produce about $285 million of 2030 EBITDA, which at 25 times is a $7.1 billion enterprise value if lidar adoption continues.
NVIDIA still looks cheap versus other AI chip leaders

NVIDIA remains the core platform of the AI build-out, yet it trades at a depressed valuation versus AMD, including a PEG ratio around 0.3x that leaves room if earnings growth holds.

Instrument Side Target Reason
NVDA Long NVIDIA is still the core AI compute platform, but it trades at a depressed valuation versus AMD, including a PEG ratio near 0.3x that is too cheap if growth holds.
Nebius power-unit pricing is understated versus AI-cloud demand

Blended revenue per MW of AI-cloud capacity could approach $20M over the next few years versus a $14M model assumption, and that pricing gap alone implies large upside in earnings power.

Instrument Side Target Reason
NBIS Long Blended AI-cloud revenue per megawatt could approach $20 million over the next few years versus a $14 million model, and that gap alone implies large upside.
Marvell is the connectivity and co-packaged optics winner

Marvell is positioned as a trillion-dollar connectivity franchise, and the Celestial acquisition strengthens its co-packaged optics offering just as GPU clusters need faster, denser optical interconnects.

Instrument Side Target Reason
MRVL Long Marvell is the connectivity franchise flagged as a trillion-dollar opportunity, and the Celestial acquisition makes it a leading co-packaged optics play as GPU clusters demand denser optical interconnects.
Applied Optoelectronics is cheap on mid-2027 optical run-rate

Near-term multiples look expensive, but a path to $5.6B of annual run-rate by mid-FY27 against a market cap still below $13.5B implies the stock is cheap on forward optical demand from AI data centers.

Instrument Side Target Reason
AAOI Long Near-term multiples look expensive, but a move toward $5.6 billion of annual run-rate by mid-FY27 at a market cap still below $13.5 billion leaves the stock cheap on AI optical demand.
Bloom Energy is scarce fast power for AI data centers

The stock screens expensive on current numbers, but few firms can deploy large amounts of on-site power to data centers on short lead times, which is the binding constraint as grid interconnects lag AI load.

Instrument Side Target Reason
BE Long The shares look expensive on current numbers, but few companies worldwide can deliver large amounts of power to data centers on short lead times as grid interconnects lag AI load.

Themes

The content on this page is for informational purposes only and does not constitute financial advice. Stoquate is not a licensed financial advisor. Always conduct your own research and consult a qualified professional before making any investment decisions.