The surge in AI-related capital spending produced some of the market's most dramatic stock rallies in recent years. For professional allocators, however, the bigger edge comes from knowing when to move on - and where to redeploy proceeds next. A rules-driven AI picker that evaluates thousands of equities has been executing precisely that discipline, closing major positions after large moves and reallocating into the pockets of the market that show the strongest combination of valuation, momentum, and growth.
That model, marketed through a monthly ProPicks AI process, continuously reassesses investment candidates and refreshes a portfolio to reflect where the data points next. The stated aim is simple: capture the core of a structural theme while exiting once the risk-reward profile tilts unfavorably versus other opportunities in the strategy's ranking.
Performance and access
Since the model's official launch in November 2023, the ProPicks AI strategy has produced a cumulative real-world return of +224.21%, outperforming the S&P 500 by +146.30% over the same interval. That outperformance reflects dozens of individual position-level outcomes - some of which became notable examples of how the rotation mechanism works in practice.
ProPicks AI subscribers receive a continuously refreshed list of high-conviction names every month. The service highlights stocks that pass a quantitative screen built from historical and forward-looking data, and it uses equal weighting across selected names to maintain a transparent performance benchmark for the strategy's results.
For a limited time, an InvestingPro subscription is available at a promotional price cited as less than $7 per month - named in the promotion as the LOWEST PRICE OF 2026 as part of an EXCLUSIVE SUMMER SALE. The offering is presented as a way for readers to access the full list of that month's ProPicks AI selections.
Realized winners - server and memory trades
Two specific trades illustrate the process the AI uses to enter and exit positions: one on the server side of AI infrastructure and another on the memory side. Both trades started with clear fundamental momentum and ended when the strategy's monthly ranking process signaled that the best risk-reward had shifted elsewhere.
Super Micro Computer (SMCI) - servers
The model added SMCI at an entry price of $28.43 in January 2024. At that point, the signals supporting the buy included accelerating hyperscaler spending on AI infrastructure, rapidly expanding revenue as customers raced to deploy GPU-dense servers, and price action that the model interpreted as a sustainable uptrend backed by improving earnings delivery. Analysts of the setup also noted that forward earnings estimates were expanding in line with revenue growth, which helped keep valuation metrics within a defensible range at the time of entry.
Over the following six months, shares rallied sharply as the AI server demand wave unfolded. The strategy exited SMCI in July 2024 at $81.23, locking a gain of +185.77% for members who entered at the stated purchase price. The exit was prompted not by a single negative event but by a shift in the strategy's internal rankings: valuation had stretched relative to near-term earnings delivery, early signs of margin pressure appeared amid rising competition and elevated component costs, and other opportunities rose higher in the monthly selection process. The model rotated the capital out of SMCI at that point.
In subsequent months the stock declined materially, ultimately losing more than three-quarters of its value and falling below the original entry price - an outcome that underlined the strategy's argument for disciplined exits when the risk-reward changes.
SanDisk (SNDK) - memory
By late 2025, the AI infrastructure trade had broadened to a different supply point: NAND memory. The models identified SanDisk as a leading beneficiary of elevated hyperscaler demand for high-performance storage in a constrained supply environment. The strategy added SanDisk in November 2025 on the basis of several fundamental signals.
- Explosive revenue dynamics, with sales noted in one section as growing 10.4% while EBITDA surged 362% as memory pricing improved.
- Accelerating hyperscaler demand for NAND, creating a runway that analysts within the model expected would persist for multiple quarters.
- Pricing power driven by large customer wins and industry supply constraints, which the models flagged as a durable tailwind.
- Upward earnings momentum, with revenue and earnings beats prompting revisions to forward estimates and leaving the stock trading at a discount to those revised forward measures at entry.
From that November entry, the position delivered a +189.09% gain over the subsequent three months as the company posted strong top-line and margin outcomes. By February's rebalance, the easier upside had largely been captured and the model again rotated capital based on valuation and ranking considerations. The decision to exit noted that shares had surged over the prior year - at one point cited as a more than 1,500% advance - pushing forward multiples to levels that the model judged to leave limited margin for execution missteps. Even with reported strength - including a later-cited 61% revenue growth figure and a 78% earnings beat noted in the strategy's evaluation - the forward return available from the stock at that time had compressed relative to other candidates in the strategy.
The rotation out of SanDisk therefore reflected the same rules-based logic used for SMCI: enter when fundamentals and momentum align, exit when comparative risk-reward deteriorates versus alternative opportunities.
Broader realized winners
Beyond SMCI and SNDK, the strategy has closed other sizable winners. The promotion and performance list cites locked-in profits including:
- SanDisk (SNDK): +189.1% locked-in profit
- Victoria's Secret Co (VSCO): +113.6% locked-in profit
- Fortrea Holdings (FTRE): +76.6% locked-in profit
- InnovAge Holding (INNV): +63.8% locked-in profit
- Rayonier Advanced Materials (RYAM): +60.8% locked-in profit
- Tronox (TROX): +55.9% locked-in profit
- Cardinal Health (CAH): +54.0% locked-in profit
- Kulicke & Soffa (KLIC): +53.0% locked-in profit
Those realized gains were then redeployed into new monthly selections as dictated by the quantitative ranking process, demonstrating the portfolio's dynamic approach to allocating capital across evolving market opportunities.
How the AI selection engine operates
At the start of each month, the proprietary engine processes a multi-decade dataset across hundreds of models to score thousands of global equities. The process mixes historical financial data with valuation signals and forward-looking growth metrics. The system references more than 15 years of financial history across 150 quantitative models to generate a ranked list of candidates, from which it identifies up to 20 high-conviction stocks per strategy based on projected medium-term upside.
The portfolio is rebalanced monthly. That rebalance adds new candidates, retains top performers that continue to meet the criteria, and removes positions whose relative ranking has declined. To provide a consistent measure of model performance, each strategy applies equal weighting across selected stocks - though individual subscribers may adjust allocations to suit their own risk preferences.
The stated objective is to systematically reposition capital toward the strongest opportunities as market conditions evolve, rather than relying on ad hoc judgment about when to hold or sell winners.
Subscription note
The piece references a promotional subscription price for InvestingPro that is described as the lowest price of 2026 for a limited period. The article also notes that subscription offers are tested regularly and may vary by region.
Conclusion
The case studies of SMCI and SanDisk demonstrate a specific tradecraft: identify concentrated exposure to structural AI demand, enter when momentum and valuation are supportive, and exit when the systematic ranking indicates that the capital can be better deployed elsewhere. That combination of data-driven entry and disciplined exits produced material realized gains for the strategy since its November 2023 launch, according to the model's performance numbers. For investors tracking AI-driven themes, the approach emphasizes rotation and relative value - capturing core moves while systematically redeploying proceeds as valuations and momentum evolve.