AI Can Play 'Significant' Role in Tech for Managing Diabetes, Says David Roman

Goldman Sachs analyst David Roman said AI could play a significant role in diabetes management, helping patients count carbohydrates and deliver the right insulin dose. He also said he left Goldman Sachs' annual medtech conference 'very bullish' on innovation across the healthcare technology sector, including companies like Abbott and Dexcom. The article is primarily analyst commentary, so the direct market impact looks limited.

Analysis

AI in diabetes management is less a headline than a monetization bridge: the near-term value is not autonomous insulin dosing, but lower user friction that improves adherence, retention, and sensor utilization. That matters because the economic moat in CGM is increasingly driven by ecosystem lock-in; if carbohydrate estimation becomes meaningfully easier, the addressable pool shifts from highly engaged diabetics to a broader, less disciplined cohort where churn has historically been the problem.

The second-order winner is the platform with the best data flywheel, not necessarily the best algorithm. If AI features reduce the need for manual input, device cadence and cloud data volume rise, which improves model quality and makes switching costs more punitive. That is constructive for DXCM if it can translate better UX into higher new-user conversion and lower abandonment, but it also raises the bar for competitors: smaller medtech players without proprietary datasets or payer relationships will struggle to match the combined hardware/software bundle.

The contrarian risk is that AI enthusiasm may be ahead of reimbursement and clinical validation. These features can drive engagement quickly, but payer adoption and physician trust usually lag by multiple quarters to years, especially where dosing decisions are involved. If the product narrative outruns hard outcomes data, the market could compress multiple expansion even as unit growth remains intact.

For GS, the read-through is more subtle: bullish sell-side tone on medtech innovation supports underwriting activity, conference access, and advisory pipeline, but the stock should only react if AI-driven healthtech becomes a durable capital-markets theme. The bigger opportunity is in identifying which tools become standards across care delivery and which remain features; the latter are easy to copy and won’t sustain margin expansion.

Key Decisions for Investors

• Long DXCM on 3-6 month horizon; use a staggered entry on any post-conference pullback. Base case is continued multiple support from AI-enabled UX improvements and higher retention, with upside if management proves conversion/engagement metrics.
• Pair trade: long DXCM / short a smaller CGM or diabetes-device laggard if available in the book. The thesis is that AI benefits accrue disproportionately to the incumbent with the richest user data and strongest payer distribution over the next 2-4 quarters.
• Consider a bullish options structure on DXCM, such as buying 6-month calls financed by selling a further-out-of-the-money call, to express upside from sentiment and product-cycle momentum while limiting premium burn if reimbursement/validation headlines lag.
• Avoid chasing GS on this note alone; at most, treat it as a modest positive for healthcare investment-banking activity over 6-12 months, not a catalyst for immediate multiple rerating.
• Set a review trigger around the next product/clinical data update: if AI features improve adherence but fail to show utilization or retention lift, reduce exposure because the market will likely reclassify the feature as incremental rather than transformative.