If you can remember back to the dawn of the internet, the discourse was filled with all kinds of utopian ideas about how it would transform the world and in many of those ways it ultimately delivered but on the commercial side, the biggest money maker and the backbone of the largest companes -- Google and Meta in particular -- is advertising. For others, it's data which usually has a roundabout way into advertising.Skip ahead 30 years and we're running through the same sequence with AI. One of the big promises is drug discovery and AI will surely help but Scotiabank talked to pharma executives and for them, the first meaningful financial impact from AI is unlikely to come from discovering new drugs. Instead, it's coming from selling existing drugs better, launching them faster and making the commercial organization more productive.That matters because commercial execution is already one of the largest costs s in pharma. The industry spends heavily on sales forces, marketing campaigns, medical-legal review, market access, patient identification and physician engagement. Those workflows are data-heavy, repetitive and measurable. AI is shaping up to be better suited to that problem than drug discovery.Scotiabank’s base case assumption is modest. They say AI can drive roughly 3% revenue accretion by 2028 across its large-cap biopharma coverage, with cost savings beginning to emerge more clearly after that as infrastructure investments mature. Even with just that, the margin uplift is huge. Across the group, that translates into an estimated $770 billion of cumulative incremental revenue and nearly $50 billion of cost savings over the next decade, before assigning any value to AI-driven discovery.The important part is that the near-term revenue contribution is coming mainly from commercial use cases but those numbers show a meaningful boost to stock prices.Sales-force tools are already helping reps decide which physicians to call on, what message to deliver and what the next best action should be. One industry executive cited in the report pointed to a potential 5% uplift in “impactable sales” for a major product from next-best-action tools. That is a meaningful number in an industry where a single blockbuster can generate billions in annual revenue.There is also a large opportunity in content creation. Pharma marketing is slow because it has to be compliant, scientifically accurate and carefully reviewed. The report notes that materials that historically took six to eight months to develop can now be produced in closer to four months using AI-enabled tools. Medical-legal review times have been reduced by around 20% in one cited example.That is where the early payoff is. Faster content, better targeting, fewer bottlenecks and improved campaign performance. The future is a Facebook feed packed with custom-tailored, AI generated pharma ads.The company examples are broadPfizer’s Charlie platform is aimed at making marketing content creation three to five times faster. Merck has used AI to ship marketing materials up to 80% faster. Johnson & Johnson is using AI tools to help reps prepare for provider interactions. Gilead has a next-best-action model for field teams. AbbVie has GenAIsys for sales-force planning and execution. Teva is using AI models to support upselling and cross-selling into pharmacy accounts.The thread running through all of this is simple: AI is being applied to revenue operations where the feedback loop is short and the outcome is measurable.That is a very different challenge from drug discovery. Discovery remains the larger strategic prize, but it is also far more difficult to underwrite. Biology is messy. Disease pathways are complicated. Datasets are fragmented and often unstructured. Testing is long and fraught with problems. Human disease biology remains poorly understood in many areas. The report notes that one executive sees meaningful discovery benefits as at least three years away, and Gilead’s commentary that roughly 90% of biological data remains unstudied is a useful reminder of the scale of the problem.Investors should be careful not to confuse long-term potential with near-term earnings power.Commercial AI does not need to solve human biology. It needs to improve the odds that the right physician gets the right message at the right time. It needs to reduce the time required to create compliant materials. It needs to find undiagnosed patients, improve market access decisions and make sales reps more productive.Those are solvable problems, and the benefits can show up relatively quickly.There is also a cost angle. AI is unlikely to lead to an immediate wholesale replacement of high-skilled commercial teams, but it can reduce the need for incremental hiring and external agency spend. One executive in the report suggested AI could meaningfully displace a large portion of the roughly $400 million his company spends on external marketing agencies. Another noted that new launches are already requiring less incremental budget and headcount than in the past.That is where operating leverage begins to emerge.The broader market has tended to value AI in healthcare through the lens of discovery platforms and biotech optionality. Some of that is warranted, but it may miss where the first real dollars are going to appear. Large pharma companies have enormous installed revenue bases, proprietary data, established sales forces and complex commercial workflows. A small improvement across that system is worth a lot.This is why AI may be more valuable to incumbents than the market appreciates. These companies do not need AI to reinvent the entire business model in order to create value. They need it to make the existing model more efficient. It's much the same elsewhere.The near-term AI winners in pharma may not be the companies with the most ambitious claims about AI-designed drugs. They may be the companies with the largest commercial platforms and doctor relationships they can exploit.Discovery is still the long-term dream. But for the next few years, the more practical opportunity is in the commercial machine.Takeaways:1) The AI dream case is there but it's years away at best, even in the most-oversold use cases2) The commercial value is real and there is meaningful upside for incumbents using it for old-fashioned sales and marketing This article was written by flc97fe4880a4b454993821fe0b770a597 at investinglive.com.