Nvidia, AMD, and Micron still dominate the AI stock conversation, but one much smaller semiconductor name has quietly stolen the performance spotlight in 2026.
Navitas Semiconductor (NASDAQ: NVTS), which trades under the ticker NVTS, recently changed hands near $17 and carried a market value of about $4.3 billion.
That makes it tiny compared with the giants of the AI trade, but its share-price move has been anything but small.
The reason is simple. Navitas is not trying to build the next GPU, but solve a different problem inside AI data centres: how to move huge amounts of power more efficiently.
That has turned the stock into one of the market’s more interesting AI infrastructure bets and valuation much harder to ignore.
The hidden bottleneck in the AI boom
Navitas makes power semiconductors. That sounds less exciting than GPUs, but it matters more as AI data centres get bigger.
Modern AI systems consume enormous amounts of electricity. That power has to be converted, stepped down and delivered efficiently inside server racks.
If too much energy is lost along the way, data centres become more expensive, hotter and harder to scale.
That is where Navitas is trying to fit in.
In March, the company introduced an 800V-to-6V DC-DC power delivery board, which converts very high-voltage power down to a level that can be used closer to the chips inside AI servers.
The key point is that Navitas says it can do this in one stage, removing the traditional 48V intermediate conversion step.
That matters because every efficiency gain counts when AI data centres are trying to feed more power into systems without wasting energy, space or cooling capacity.
That is also where the Nvidia comparison becomes more useful as Navitas is not competing with Nvidia, and it is certainly not a bigger AI business.
Nvidia remains the centre of the AI chip universe, with a market value above $4.7 trillion, while Navitas is still a small-cap name worth roughly $4.1 billion.
But in stock-market terms, Navitas has done something unusual in 2026: it has outpaced Nvidia while riding the same AI infrastructure wave.
Navitas was up roughly 148% year-to-date as of June 29, far ahead of Nvidia’s roughly 8% to 12% gain over the same broad period.
Navitas Semiconductor stock: Why Wall Street suddenly cannot stop upgrading it
The stock’s surge has not been driven only by retail excitement.
Analysts have also moved quickly to reset their expectations. Morgan Stanley lifted its price target on Navitas to $12.50 from $4.20 in May.
Baird followed with an even more aggressive move, raising its target to $20 from $9.
The revisions show that Wall Street is taking the AI power-delivery story more seriously than it did a few months ago.
The reason is that Navitas sits at the intersection of two hot themes: AI infrastructure and energy efficiency. Data centres need more power, but they also need to waste less of it.
A company that can improve conversion efficiency inside AI racks has a clean story to tell investors.
But the stock is volatile. Its 52-week range runs from $5.44 to $34.17, which tells you how quickly expectations have moved.
This is not a sleepy industrial supplier, but a small-cap semiconductor stock being repriced around a fast-changing AI narrative.
The catch: Should you actually buy?
This is where the story gets more complicated.
Navitas may be exciting, but the stock is no longer cheap. As per market data, the Navitas Semiconductor stock trades at about 92 times sales, compared with a five-year average price-to-sales ratio of 11.8.
That means the stock is trading at roughly eight times its historical valuation multiple.
That is a serious premium for a company still trying to prove how much revenue it can generate from AI data-centre demand.
The analyst picture is also more mixed than the headlines suggest.
Some firms have raised targets, but several consensus trackers still show the average price target below the current share price.
That is the tension investors need to sit with.
The post This $17 AI infrastructure stock is beating Nvidia in 2026: should you buy? appeared first on Invezz
