The Needham Fund has found a way to profit from AI in small-cap stocks. Here are its top picks
In the first half of the year, the Needham Small Cap Growth Fund outperformed the Russell 2000 Growth Index by a factor of 4
The Needham Small Cap Growth Fund is betting on AI-related infrastructure / Photo: Igor Omilaev / Unsplash
The AI boom delivered a return of nearly 87% over six months to the Needham Fund, which invests in small-cap companies. But it also created problems for some of its investments. Which three stocks generated the highest returns for the fund, and which investment turned out to be the biggest failure?
How Can Small-Cap Investors Profit from AI?
The Needham Small Cap Growth Fund is a prime example of how investors who put money into small-cap companies can profit from the massive spending on AI infrastructure. These expenditures benefit not only chip and accelerator manufacturers—players with market capitalizations in the tens of billions—but also interface developers, analytics software providers, and suppliers of passive components. This is precisely where a significant portion of U.S. small-cap companies are concentrated.
Chris Retzler has managed the fund since January 2008. He describes his approach as follows: “We tend to look for deeply undervalued opportunities at a reasonable price.”
The fund has two share classes: the retail class, NESGX (which has a higher fee), returned 86.58% in the first half of the year, while the institutional class, NESIX, returned 87.13%. By comparison, the Russell 2000 Growth Index gained 22.18% over the same period.
A year earlier, things were different: in 2025, the fund’s return was lower than the index’s—11.16% versus 12.90%.
As of June 30, 2026, the fund’s net assets totaled $419 million across 70 holdings. However, its portfolio differs significantly from the composition of the Russell 2000 Growth Index, a small-cap index comprising growth companies. More than half—58%—of the assets in Needham’s specialized fund were in the information technology sector.
This level of concentration comes at a price. As early as July, during the latest market crash, the fund lost 16.63%, and its year-to-date return had fallen to 58.1% by August 10.
The ten largest holdings accounted for 36.34% of the fund’s assets. Their combined contribution to returns was approximately 17 percentage points. The three-year beta is 1.35—this means that the fund’s price movements are more volatile than the market’s.
Which investments have been the most successful since the beginning of the year?
— Vishay Intertechnology, Malvern, Arkansas
Top 10 Stock: The company manufactures discrete semiconductors and passive components: resistors, capacitors, diodes, and MOSFETs.
On August 5, 2026, Vishay reported its second-quarter results. Its adjusted revenue reached $918.6 million, a 20.5% increase year-over-year. Earnings per share were $0.19. Gross margin rose to 22.6%, exceeding the company’s forecast. The order backlog grew 18% during the quarter to $1.9 billion, and the book-to-bill ratio reached 1.32.
Its potential growth driver is the industrial segment. Revenue in this segment jumped 30% year-over-year in the second quarter, driven by demand for smart grids, power electronics for AI, and high-voltage DC lines. According to JPMorgan’s estimates, the company’s revenue from AI will reach $150–200 million in 2026, or 4–5% of total sales.
The company faces the risk that its profit margins are lower than those of its competitors in the field of power solutions for data centers. In addition, its convertible debt increased the number of shares outstanding from 137 million to 162 million, which diluted shareholders' stakes.
The average target price for Vishay is $39, which implies 17% upside potential from current levels.
— Arteris, Campbell, California
The company develops technologies for data exchange between processor cores, memory, and AI accelerators.
On August 6, 2026, Arteris reported second-quarter revenue of $24.1 million, a 46% increase compared to the same period last year. Commitments under existing contracts totaled $135 million, up 36% from a year earlier. The annual value of contracts, including royalties, reached $99.5 million, a 44% increase year-over-year.
The company described all of these figures as record-breaking.
The company raised its revenue forecast for 2026 to $95–98 million (from $91–95 million).
The company’s growth is driven by customers’ shift to chiplets and custom AI accelerators: the quarter’s largest deals were in the data center segment, and in the automotive sector, the company received royalties from Li Auto’s mass-produced autonomous driving chips.
On the day the earnings report was released, analysts at TD Cowen reaffirmed their “Buy” rating and $40 price target. On July 16, Oppenheimer initiated coverage of the company with an “Outperform” rating and a $40 price target. In August, Jefferies upgraded Arteris’s rating from “Hold” to “Buy” and raised its price target from $35 to $50.
The current average target price for the company's shares is $41. At the close of trading on August 13, they were trading at $27.59.
But it’s important to understand that despite its record performance, the company is not turning a profit: its operating loss for the quarter exceeded analysts’ expectations, and quarter-over-quarter royalty growth slowed. The company’s CFO, Nick Hawkins, attributed this to logistics and supply chain issues at one of its clients.
— Veeco Instruments, Plainview, New York
The company manufactures equipment for laser annealing, ion etching, and lithography.
On August 5, 2026, the company reported revenue of $193.5 million, up 16.5% from the previous year and above its own forecast. Adjusted earnings per share came in at $0.33, compared to a consensus estimate of $0.28.
Analysts at Needham noted that growth came from all of the company’s business segments. The main driver was a $200 million order for advanced packaging equipment (the integration of multiple chips or wafers into a single package) received in the second quarter from several customers at once; shipments are scheduled to begin in the first half of 2027.
The company raised its revenue forecast for 2026 to $780–810 million from $740–800 million, but at the same time lowered its adjusted earnings per share guidance to $1.36–1.61 (from $1.50–$1.85). The reason is that Veeco is hiring and expanding capacity in anticipation of growth in 2027.
Veeco’s rising stock price is not solely due to its business performance. In February of this year, its shareholders approved a merger with Axcelis Technologies. The deal is expected to close by the end of 2026. However, it is contingent on obtaining approval from China’s antitrust regulator, SAMR.
Veeco’s stock price currently reflects not so much its operating results as the likelihood and timing of a deal: for this reason, on August 6, Needham maintained its “Hold” rating on the stock and did not set a price target. On August 10, Citigroup analysts raised their price target from $60 to $63, maintaining a “Buy” rating. The market consensus is $61.33. The closing price on August 13 was $53.58.
— Calix, San Jose, California
This is the worst performance among the fund's top ten holdings.
The company sells a bundle of hardware, a cloud platform, and managed services to broadband providers.
Its second-quarter revenue totaled $293.3 million, up 21% from the previous year and above the upper end of its own forecast. Adjusted earnings per share reached $0.47, a result that also exceeded the company’s own forecast.
On the surface, it was a strong report, but the stock fell by more than 12% in after-hours trading on the day of the release, and by more than 6% over the next three days following the report. Investors were disappointed that the company’s gross margin came in below the market consensus—54.8% versus 55.6%.
This was due to rising memory prices caused by high demand from data centers. Calix purchases memory chips for its Wi-Fi equipment, so the increase in their prices directly reduces the company’s profit margin.
The company expects its gross margin to decline further in the third quarter to 50.5–53.5% (compared to the market consensus of 55.7%), as part of its order backlog is priced at older rates. The company is passing on the rise in memory costs to new orders and is now adjusting these markups monthly rather than quarterly. However, these markups merely offset the additional costs. The company does not yet forecast when the margin will return to around 55%, as it cannot accurately assess the future trend in memory prices.
On July 21, JPMorgan lowered its price target for the company’s stock from $65 to $58, while maintaining its “Outperform” rating. The average target price for Calix shares is $62.33, which is about one and a half times higher than the current price. Six out of seven analysts recommend buying the stock.
Four Conclusions: The Pros and Cons of the Fund's Strategy
— Needham Small Cap Growth is not betting directly on AI, but rather on the infrastructure surrounding it: on-chip interfaces, packaging equipment, manufacturing data analytics, licensed intellectual property, and broadband networks.
— The range of returns among the fund’s largest holdings shows that its performance is driven not by a focus on a single sector, but by individual growth stories. At the same time, at least one company received a boost not from its operating metrics, but from an M&A deal.
— Portfolio growth requires high capital expenditures by hyperscalers, affordable memory prices, and a market willing to pay high valuations for unprofitable growth companies. At the same time, some holdings are suffering from the very same factor that is supporting other companies in the portfolio: for example, rising prices for electronic components are putting pressure on Calix’s margins.
— A portfolio like this, with a strong focus on IT and AI, is better suited for investors with a long-term horizon and a high tolerance for volatility than for those who expect stable returns.
