The AI power infrastructure buildout is quietly rewriting the demand curve for wide-bandgap semiconductors. Gallium nitride (GaN) and silicon carbide (SiC) suppliers sit at the center of the transition from 48-volt to 800-volt DC rack architectures, and three small-cap names give investors direct exposure at entry prices well under $50. Each has AI-data-center leverage in its bull case, a fresh earnings report, and identifiable execution risk. Here is how they stack up in August.
Navitas Semiconductor (NVTS)
Navitas Semiconductor (NASDAQ:NVTS) is the purest AI-power play on this list. Shares closed at $12.23 on August 24, up 71.29% year to date and 90.2% over the past year, though the stock fell 16.12% in the past week as the group cooled off.
Navitas reported Q2 FY26 revenue of $10.5 million, up 22% sequentially, with non-GAAP gross margin of 39.5% and cash of $557 million against no debt. Q3 revenue is guided to $13.5 million, plus or minus $0.5 million. CEO Chris Allexandre told investors that "high-power markets grew more than 50% year-over-year" and that "AI infrastructure market will represent more than one-third of our total sales by year-end."
The bull case is architectural. Navitas is designed into the NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) MGX 800V DC rack ecosystem, and management flagged a new 1.2 kV JFET product line targeting an incremental $1 billion of SAM by 2030. Fuse's model puts a one-year base-case target of $16.18, or 32.3% upside.
The risk: this is a story stock. Beta sits at 3.882, TTM EPS is -$1.34, and quarterly revenue is still down 27.3% year over year. The 2027 hyperscaler ramp has to land.
Alpha and Omega Semiconductor (AOSL)
Alpha and Omega Semiconductor (NASDAQ:AOSL) is the mix-shift story. Shares last traded at $25.89, up 30.69% year to date but down 16.67% over the past week. Forward P/E is a striking 8, and the analyst target sits at $43.75.
The Q4 FY26 report delivered revenue of $170.4 million, with non-GAAP gross margin expanding to 23.7% from 21.7% the prior quarter. Non-GAAP loss per share was $0.13, a meaningful improvement from the $0.28 loss in the March quarter. The September-quarter guide calls for revenue of approximately $176 million plus or minus $10 million and non-GAAP gross margin of 24.5% plus or minus 1%. AI and server revenue is expected to grow more than 60% sequentially.
CEO Stephen Chang said, "Advanced computing is now a clear and growing contributor to both revenue and earnings." The joint-venture divestiture closed with the final $15 million installment, leaving cash at $180.8 million. Fuse's one-year base case projects $51.32, or 98.22% upside.
The caveat is the September consumer segment, which is guided down approximately 25% sequentially. Traditional PC weakness tied to memory and CPU shortages, plus typhoon and Shanghai flooding impacts, will pressure the quarter even as the AI mix improves.
Wolfspeed (WOLF)
Wolfspeed (NYSE:WOLF) is the post-restructuring wildcard. Shares closed at $24.76, up 42.22% year to date but down a sharp 29.18% over the past week following the Q4 report.
The numbers were mixed. Fiscal Q4 revenue came in at $150 million, with non-GAAP EPS of -$2.26 versus -$1.465 expected, missing expectations by a wide margin. Non-GAAP gross margin stayed at minus 19.9%. Guidance for Q1 FY27 is revenue of $140 million to $160 million, with gross margin expected to remain negative.
The bull case is the AI pivot. CEO Robert Feurle said AI data center revenue more than doubled from fiscal 2025 to fiscal 2026 and grew approximately 20% sequentially in Q4. Wolfspeed launched its fifth-generation SiC MOSFET, a commercially ready 10 kilovolt SiC MOSFET recognized as the top innovation at PCIM 2026, and signed an MoU with GE Aerospace. Balance sheet cash sits at $576.3 million, and management said gross-margin breakeven arrives at roughly an $800 million annual run rate. Fuse's base case is $31.76 for 28.28% upside, with a bear case at $14.60.
Wolfspeed is the highest-risk, highest-optionality name here. Materials revenue fell to $43.3 million from $78.4 million, TTM EPS is -$11.39, and analyst sentiment skews cautious with one hold and one strong sell. This one only works if AI data center demand ramps faster than the legacy materials business erodes.
What to Watch Next
All three tickers trade under $50, all three have real AI-power exposure, and all three run on cash cushions large enough to fund another year of ramp. Navitas offers the cleanest thesis, AOSL offers the cheapest multiple, and Wolfspeed offers the deepest turnaround optionality. The next catalysts arrive with Q3 earnings reports and any additional 800V design-win disclosures heading into 2027. Power and cooling suppliers keep showing up on our radar for the same reason, and we rounded up seven of them in a free report on the AI infrastructure names that aren’t chipmakers.
Contact [email protected] for any questions or corrections.