DISCLAIMER — Not financial advice. Educational content only, not an offer or solicitation to buy or sell any security. Biotech and small/mid-cap stocks are highly speculative and volatile and can result in a partial or total loss of capital. Do your own research and consult a licensed advisor where appropriate. / Contenuti a solo scopo informativo e didattico, non costituiscono consulenza finanziaria né offerta o sollecitazione al pubblico risparmio ai sensi delle normative CONSOB e SEC. Le azioni biotech e le small/mid cap sono strumenti altamente speculativi e volatili e possono comportare la perdita parziale o totale del capitale investito. Si raccomanda di effettuare sempre le proprie ricerche e, se necessario, di rivolgersi a un consulente abilitato.

Merlintrader Trading Pub
Biotech catalyst news and analysis. FDA PDUFA tracker

Merlintrader Trading Pub
Biotech catalyst news and analysis. FDA PDUFA tracker
AI disruption 2026 – winners, losers and what it really means for people
GPUs, memory and cloud infrastructure are treated like oil in the AI age, while a chunk of traditional software is suddenly being repriced as “automation-risk”. Under the surface, this isn’t just a sector rotation: it is a change in how markets value business models and human skills.
The core idea is blunt but useful: markets are paying up for the “picks and shovels” of AI – chips,
foundry, high-bandwidth memory, data centres, cybersecurity, observability – and they are putting a
heavy discount on businesses whose main value proposition can be replicated by AI agents, plug-ins and
low-cost automation. It is not just about which tickers are in fashion; it is about which roles and
workflows are becoming cheaper to automate.
1. What is actually happening in markets right now
For a while, “AI in markets” basically meant buying NVDA and a few mega-caps and calling it a day. That phase is over. With agentic AI, plug-in ecosystems and vertical copilots appearing everywhere, investors have shifted to a more brutal question that cuts across sectors: if a well-trained AI agent could do most of this work, how much am I really willing to pay for the software – or the human – doing it today?
On one side you have companies that benefit almost mechanically from AI adoption. They supply the physical and logical infrastructure: GPUs, custom chips, HBM, foundry capacity, cloud platforms, network and security stacks, data and observability. On the other side, you have business models that sit exactly where AI is strongest: repeatable workflows driven by text, forms and structured data, from CRM and marketing software to tax and accounting, legal tech, wealth management, creative tools, collaboration suites and design platforms.
As of February 2026, this is how the market is handling the trade: a clear “AI Winners” bucket and a “Losers / at risk” bucket, not in the sense that the latter will vanish overnight, but in the sense that disruption risk, margin pressure and lower long-term multiples are being priced in much more aggressively than just a year ago.
2. AI Winners vs AI Losers – the 2026 map
The table below is a compact snapshot of how a set of US-listed names is being seen today in the “AI disruption” theme. On the left are those treated as structural beneficiaries; on the right are those where AI agents and automation overlap strongly with what the product already does.
| AI Winners (benefit from AI) | Main driver | AI Losers / at risk | Why at risk |
|---|---|---|---|
| NVDA – NVIDIAWinner | Dominant GPUs for AI training & inference, plus a very sticky software ecosystem. | CRM – SalesforceAt risk | Fear that AI agents can automate large chunks of sales and CRM workflows. |
| MSFT – MicrosoftWinner | Azure, Copilot and deep AI integration across productivity, dev tools and cloud. | NOW – ServiceNowAt risk | Workflow automation and IT service management that agents could rebuild and orchestrate directly. |
| TSM – TSMCWinner | Crucial foundry for virtually all leading-edge AI chips. | HUBS – HubSpotAt risk | Marketing and sales software where AI can generate campaigns, copy and scoring almost for free. |
| AMD – Advanced Micro DevicesWinner | Gaining share in AI GPUs and data-center compute versus NVIDIA. | TEAM – AtlassianAt risk | Collaboration and project tools whose “glue work” can be handled by AI agents. |
| AVGO – BroadcomWinner | Custom AI silicon and high-speed networking for data centres. | INTU – IntuitAt risk | Tax and accounting software, a perfect target for specialised AI tools. |
| MU – MicronWinner | HBM and advanced memory, indispensable for large models and complex workloads. | BOX – BoxAt risk | Cloud storage and collaboration seen as easily replaced by AI-native knowledge tools. |
| PLTR – PalantirWinner | Data and AI platform exposure in enterprise and government use-cases. | ADBE – AdobeAt risk | Creative tools pressured by AI image, video and text generation. |
| AMZN – Amazon (AWS)Winner | Hyperscaler with AI infrastructure, managed models and proprietary chips. | FIG – Figma (design proxy)At risk | Collaborative design where AI can generate UI/UX directly from prompts. |
| GOOGL – AlphabetWinner | Gemini, Google Cloud AI and on-device AI; search and ads being rebuilt around models. | SHOP – ShopifyAt risk | E-commerce setup and operations increasingly automated by AI agents. |
| META – Meta PlatformsWinner | AI-driven ad optimisation, content ranking and open-source Llama models. | SCHW – Charles SchwabAt risk | Wealth management and brokerage exposed to fee pressure from AI-based advisory tools. |
| CRWD – CrowdStrikeWinner | Cybersecurity that uses AI to defend against increasingly automated attacks. | LPLA – LPL FinancialAt risk | Advisor platform where AI can erode the value of some standardised advisory functions. |
| PANW – Palo Alto NetworksWinner | Network and cloud security strengthened by AI analytics. | TRI – Thomson ReutersAt risk | Legal and research data increasingly reachable via LLM integrations and plug-ins. |
| DDOG – DatadogWinner | Observability for cloud stacks that are becoming more complex and AI-heavy. | LAW – CS DiscoAt risk | Legal tech and e-discovery, precisely where AI can read and summarise documents at scale. |
This is not a list of “doomed” and “invincible” companies; it is a snapshot of how disruption risk
is being priced in February 2026 for these business models.
3. Why markets love infrastructure and punish part of software
The logic behind this split is more pragmatic than ideological. On the infrastructure side, demand is structural: training larger models, serving more agents, running heavier inference and building out data centres all require more compute, more memory, more networking, more storage, more security and better observability. Whether one model or another wins, the physical and platform-level inputs are non-optional. That is why names in the “AI Winners” column are being treated almost like utilities of the AI era, with investors willing to pay high multiples as long as visibility remains strong.
On the software side, the threat is not that AI refuses to show up. The threat is that marginal cost of software work falls dramatically. If an agent can write sales emails, design campaigns, prepare tax drafts, do first-pass legal research, assemble project boards, generate user interfaces and propose basic portfolios, it becomes legitimate to question how much pricing power remains for vendors that sell those workflows as a standalone product.
This is not a binary “disappear / survive” story. It is a story of margins and bargaining power. Markets are starting to price scenarios where part of the value previously captured by these companies is transferred to users in the form of lower prices, and to AI infrastructure providers in the form of higher compute consumption. Less rent for the middle layer, more value for the layers where scarcity is real – capital-intensive hardware, trusted platforms, differentiated data.
4. Looking ahead – what this means for companies
4.1 For the “AI Winners”
For the winners, the main risk is not lack of demand but excess expectations. GPUs, memory and data-centre builds are inherently cyclical. When everyone expands capacity at the same time, pricing power eventually softens, and the narrative can turn quickly from “shortage” to “overbuild”. Hyperscalers are also working hard to reduce dependence on any one supplier by designing their own chips and balancing their vendor mix. Over a long enough horizon, that naturally redistributes some bargaining power.
That said, if agent adoption and AI workload growth continue anywhere near the current trajectory, it is hard to imagine a world where AI infrastructure becomes a dull, low-growth business. The balance between different suppliers will shift; some cycles will be painful; but the structural need for compute, memory and secure data pipes is not going away. These companies are likely to stay at the centre of investor attention, even if the story occasionally pauses for digestion.
4.2 For the “At-Risk” software names
On the at-risk side, the story is more nuanced. Some companies will manage to integrate AI so deeply that they migrate back toward the winner camp. If Salesforce, ServiceNow, Adobe, Intuit or Shopify become the natural interface through which customers orchestrate their own agents and automations, markets will reward them again with higher multiples. They will no longer be selling just “classic SaaS”; they’ll be selling AI-hosted platforms with network effects and lock-in.
Others will have a harder time. Generic legal tech, low-end tax software, undifferentiated collaboration tools and platforms that essentially mirror what “LLM + APIs + a few plug-ins” can do over a weekend are at risk of a slow erosion in revenue quality: not an overnight collapse, but quarters of price pressure, tougher renewals and customers increasingly willing to experiment with AI-native alternatives that feel cheaper and more flexible.
The corporate lesson is simple and uncomfortable: adding an “AI” bullet point to the product sheet is not enough. Products have to be redesigned around a world where every client has at least one capable agent at their disposal. The winning pattern is to become the place where those agents live and coordinate, not just another app that sends notifications and generates dashboards.
5. Looking ahead – what this means for people and work
All of this shows up as green and red on screens, but the real pressure is on the people who work inside these companies: sales reps, marketers, customer-success managers, accountants, paralegals, wealth managers, designers, developers, support teams, analysts. The market map is really a map of which human tasks are becoming cheaper to automate and which ones are becoming more valuable.
5.1 Where AI replaces and where it amplifies
AI replaces most easily the kind of work that is repetitive, rule-driven and heavily text- or form-based: writing standard emails, filling out templates, drafting simple contracts, preparing routine tax forms, summarising meetings, tagging tickets, doing first-pass research across large document sets, putting together basic slides, producing “good enough” images or layouts.
It tends to amplify, rather than replace, work that lives closer to three zones: context, decision and relationship. Context is the ability to understand what really matters in a messy situation. Decision is the willingness and skill to choose between trade-offs with real consequences. Relationship is the human side – trust, motivation, negotiation, leadership.
A salesperson whose job is mostly blasting out follow-up emails is vulnerable; a salesperson who uses AI to prepare, research, and then spends time listening to the client and shaping deals is not. A lawyer whose day is 90% copy-pasting clauses is vulnerable; a lawyer who uses AI for research and drafting but then invests time in strategy, negotiation and risk management is still very much needed. The same pattern repeats in finance, product, operations and creative work.
5.2 How the skill mix is shifting
One quiet but profound shift is that “tool skills” alone are losing their edge. Being the person who knows their way around a CRM, a creative suite, a project-management tool or an accounting package used to be a real advantage. With AI helping people learn interfaces faster and automating many low-level actions, that advantage decays.
What rises in value instead is a trio of capabilities that cut across roles.
First, operational AI literacy. You do not need to be a machine-learning engineer, but you do need to know how to work with AI on a daily basis: how to frame prompts, how to chain tools, how to sanity-check outputs, how to build small automations that save hours. People who can orchestrate agents and systems, rather than just click through static screens, become dramatically more productive.
Second, business and risk understanding. As mechanical work is automated away, the centre of gravity moves toward understanding what actually makes sense for a specific customer, regulation and balance sheet. That applies to wealth managers deciding whether a portfolio is appropriate, to tax professionals navigating grey zones, to lawyers weighing legal and reputational risk, to product managers and founders prioritising what to build next.
Third, communication and relationship skills. In a world where AI can generate reports, emails and slide decks in minutes, the scarce resource is not words but trust. The ability to explain, persuade, de-risk, calm, motivate and lead becomes more, not less, valuable. Roles that combine AI fluency with empathy and leadership will be in higher demand than purely mechanical “back-office” functions.
5.3 The uncomfortable part: some roles really are at risk
It is tempting to say that “AI will create more jobs than it destroys” and move on. Historically, that has often been true at the macro level, but the transition can still be harsh at the micro level. Highly specialised roles built almost entirely on repetitive procedures – data entry, basic back-office, first-line support, low-complexity form filling, generic document review – are under real pressure.
That does not mean the people in those roles are doomed, but it does mean that choosing not to engage with AI in 2026 is effectively choosing a higher probability of professional obsolescence in the next three to five years. The upside is that the playing field is still open: because many organisations are only now figuring out how to use AI properly, individuals who move early, experiment and learn can close much of the gap surprisingly fast.
6. In summary – a market map and a personal signal
The “AI Winners vs AI Losers” list that traders are throwing around right now is not just a scoreboard for tickers. It is a real-time map of where intelligence, capital and automation are re-drawing economic value. More power to those who own scarce physical and platform resources; more pressure on those who sit too close to what AI can do cheaply and at scale.
For companies, the message is that it is no longer enough to slap a chatbot on top of an old product and label it “AI-powered”. The products that will survive and thrive are the ones rebuilt around a world where customers bring their own agents to the party and expect them to plug in cleanly.
For people, the message is equally clear: staying on the right side of this shift means treating AI not as a gimmick or a threat, but as a daily lever – offloading mechanical work, climbing up the ladder of context and decision, investing in communication and trust.
In that sense, the winners/losers table for February 2026 is more than a hot macro theme. It is a mirror showing which kind of work the market believes will be done mostly by machines, and which kind still needs humans in the loop. Reading that correctly matters if you are allocating capital, but it matters even more if you are deciding what kind of skills you want to carry with you into the next decade.
Scanner for active traders

Try ChartsWatcher free, then unlock 10% OFF with SAVE10
ChartsWatcher is a real-time scanner for momentum traders: fast movers, unusual volume and rotations — so you can focus on the few tickers that matter right now, instead of watching hundreds of charts.
Start with the free version. When you upgrade, use SAVE10 for 10% OFF your first paid period.
Start free – then use SAVE10
No credit card required to start. Apply SAVE10 when upgrading.
Recommended platform
One platform. All your brokers.
Medved Trader connects multiple brokers in one workspace, with pro charts, hotkeys and fast execution — without changing your broker accounts.
A single cockpit for positions, Level II and multi-broker order routing, built for active day & swing traders.
Get 1 Month Free ➔
Multi-broker workflow + customizable layouts in one platform.