5 AI stocks that struggled in 2025, even with the AI boom!

Some may not believe this, but there are some AI stocks that struggled in 2025 even with the AI boom. Just because a company is right at the thick of a ‘hot trend’, it does not mean execution will be successful or even more successful than its competitors. Let’s look at 5 such AI stocks.

What are the Best ASX Stocks to invest in right now?

Check our buy/sell tips

5 AI stocks that struggled in 2025!

Adobe (NDQ:ADBE)

You’d think the company behind PhotoShop (along with Illustrator, Premiere and other products) would be at the forefront of the AI resolution, especially given its generative AI tools (specifically Firefly). But this has not been the case. Despite this pivot toward AI augmentation, Adobe’s stock has underperformed severely over the past 12 months, with share price down >20%. A key reason is investor scepticism about how quickly and profitably Adobe’s AI investments will translate into meaningful revenue acceleration.

Yes the company is growing revenue and recurring subscription sales, but the generative AI landscape has become intensely competitive, with products from Microsoft/OpenAI, Alphabet, Canva, Figma and others eroding Adobe’s traditional dominance in creative workflows and raising questions about pricing leverage. Additionally, Adobe has faced legal and regulatory pressures, such as a copyright lawsuit alleging misuse of authors’ works to train AI models, which contributes to near-term uncertainty.

Marvell Technology (NDQ:MRVL)

Marvell Technology makes semiconductors and custom silicon for high-performance computing, networking and data centres — areas that are critical for supporting AI infrastructure. Its portfolio includes custom AI accelerators, co-packaged optics, and high-speed data-centre interconnects that are used by cloud hyperscalers and enterprise customers.

Despite its AI infrastructure exposure, Marvell’s stock has lagged over the past year, declining roughly ~20%. A combination of factors has contributed to this weakness. A key element was investor disappointment with periodic guidance — even when current quarterly revenue was in line with expectations, forward outlooks often failed to excite markets at the lofty growth multiples AI-linked stocks command.

Geopolitical tensions, particularly fears around U.S.–China trade restrictions and the impact on global semiconductor supply chains, added uncertainty because a significant portion of Marvell’s revenue is tied to China. Additionally, the custom silicon business can have “lumpy” orders from hyperscale customers, meaning big deals followed by quieter periods, which made near-term revenue visibility less stable.

C3.ai (NYSE:AI)

Not even the company with AI as its ticker is immune from a decline – indeed it has more than halved in 2025. C3.ai is an enterprise AI software company that sells a suite of AI applications and a platform designed to help large organisations build and deploy AI solutions. Its business model is software-as-a-service (SaaS), with most revenue from subscription licensing of tools for predictive analytics, operational optimisation and other data-centric AI workflows.

The company saw steep drawdowns as revenue declines and execution issues emerged. Indeed in Q1 of the current fiscal year, revenue declined 19%. No that’s not a typo, it fell 19% – it was not a case of revenue rising 19% but shares fell as the company guided to 39%…revenue fell 19%. Founder Thomas Siebel stepped aside due to health reasons and there was a a global sales reorganisation that disrupted deal closures. Margins compressed significantly and losses widened compared to prior periods, all of which dampened investor confidence.

Salesforce (NYSE:CRM)

Salesforce has its namesake leading customer relationship management (CRM) platform. The company purports to have integrated AI deeply into its products (e.g., Einstein AI) to automate sales, service and marketing workflows. Over 2025, Salesforce shares has lagged, with roughly a >20% decline over the past 12 months, partly reflecting macro pressures on enterprise IT spending and investor caution on how quickly its AI tools translate into material revenue growth. The stock’s 2025 retreat mirrors broader tech multiples compressing as investors demand near-term results on AI initiatives.

SentinentalOne (NYSE:S)

SentinentalOne is an AI-centric cybersecurity platform whose Singularity suite uses machine learning to automate threat detection, response and incident handling. Shares have been down materially from prior highs, reflecting cooling speculative appetite, questions around profitability and a challenging broader cybersecurity spend environment.

The company’s revenue has continued to grow — roughly 23–25 % year-over-year — and annualised recurring revenue (ARR) has crossed the $1 billion mark, showing its not the case that no one wants a piece of it. But at the bottom line, it continues to notch up significant net losses and is relying on stock-based compensation and R&D spend to fuel product development and sales efforts.

The takeaway for investors here

Some may be surprised at seeing any AI companies not do well – aren’t all AI companies automatically AI winners. But such thinking may come from a from a very understandable but flawed syllogism: “This company is in AI, AI is transformative, therefore all AI stocks should go up.” Markets don’t work that way, especially once a theme becomes crowded and expectations get extreme.

First, AI is a technology wave, not a guarantee of profits. History is full of examples where a real technological shift created enormous value overall but destroyed or stagnated many individual stocks along the way. The internet did this in the late 1990s; cloud computing did it in the 2010s. AI is similar. Some companies will become structural winners, some will be enablers with thin margins, and others will discover that “having AI” does not automatically translate into durable revenue or pricing power. The explosion of AI applications — from enterprise automation tools to consumer-facing products like an online math solver — shows how widespread the technology has become, but widespread adoption does not mean every provider captures meaningful economic value.

Second, share prices and valuations reflect expectations, not just business quality. In other words, just because it is highly valued it does not mean it is or will be an AI winner or loser Many AI-linked companies entered 2024–2025 priced for near-perfect outcomes: rapid adoption, fast monetisation, expanding margins and little competition.

When reality arrives more slowly — pilots instead of large contracts, higher compute costs, customer caution, or delayed payoffs — the share price can fall sharply even if the company is still growing and still “doing AI.” A stock can be down 30% and the underlying business can still be healthy; it just wasn’t healthy enough to justify the prior valuation.

Third, AI is expensive to build and monetise, especially for software and infrastructure companies. Training models, buying GPUs, building data centres, hiring scarce talent and integrating AI into enterprise workflows all cost real money today, while revenues often come later. Investors tend to punish companies during this “investment gap,” particularly if cash burn rises, margins compress or guidance becomes uncertain. This is why some AI infrastructure and enterprise-software names have fallen despite strong long-term narratives.

Fourth, competition in AI is brutal and still accelerating. Barriers to entry are lower than many assume at the application layer, and hyperscalers are moving up the stack while startups move down. This puts pressure on pricing and differentiation. For investors, the question becomes not “Does this company use AI?” but “Can it defend margins, retain customers and avoid being commoditised?” Stocks fall when the answer becomes less clear.

Fifth, AI stocks are not immune to macro and market cycles. Rising rates, tighter financial conditions, enterprise IT budget scrutiny and rotations away from growth all matter. When markets de-risk, high-multiple AI names often fall more, not less, because so much future success is already priced in.

So the takeaway shouldn’t be that AI is overhyped or not working. It’s that AI as a technology can be wildly successful while many AI-branded stocks disappoint. For investors looking for AI winners, the mindset has to shift from thematic investing (‘AI will win’) to selective investing (‘Which companies can sustainably turn AI into cash flow, defensible margins and long-term returns?’).