← back to rankingOSCR · Oscar Health, Inc.
Medical - Healthcare Plans · mkt cap $5.5B · calls: Q1 FY2026 vs Q4 FY2025
49.0 conviction · conf-adj 47
conf 4/10 partial
enthusiasm:18.0 · trend:-5 · quantifies:5 · impact:0 · under_radar:5 · credibility:12 · business_impact:8 · disruption:0 · commitment:0 · confirmation:6
Enthusiasm latest 6 / prev 8 (falling)
Oscar’s AI story is operational and member-facing: agentic bots, Oswell, LLMs, and automation are framed as drivers of SG&A leverage, faster service, and retention—not as a separate revenue line. Credibility is strongest in Q4 FY2025 (67% response-time cut, 86% Oswell resolution) and weaker in Q1 FY2026, where AI is still cited for margin but specifics shift to predictive drug pricing and voice agents without refreshed KPIs. No analyst pressed AI on the latest call; impact remains bundled into SG&A bps rather than isolated AI economics.
GROUNDED NEXT-FY IMPACT vs CONSENSUS
Grounded on actual base — revenue $11.7B · net income $-0.4B · net margin -3.8% · diluted EPS -1.69
These are next-fiscal-year annual uplift estimates, not next-quarter numbers.
Aggregate next-FY est. rev uplift: 0.0% · next-FY EPS uplift: % · vs analysts: inline · priced in: medium (model's call-read: high; verdict above is the hard-data one used for ranking) · confidence: 4/10
| Claim | Figure | Arithmetic | Next-FY Rev % | Next-FY EPS % |
|---|
Care-guide bot cut response times 67%
productivity · soft | 67% | 67% latency/cycle-time reduction with no disclosed care-guide opex/FTE base anywhere in the inputs; cannot convert to $ saved without inventing a cost pool. Unanchored. | | |
Oswell agent handles 86% of member questions
productivity · soft | 86% | 86% deflection/automation rate; SG&A ratio (17.5%) is disclosed but not split by function, so no member-services cost base or call volume $ to size deflection savings. Real lever, no anchor. | | |
FY2025 SG&A ratio -160bps to 17.5%
cost | -160bps | 160bps x $11,701,427,000 FY2025 rev = ~$187,222,832 gross SG&A reduction vs prior 19.1%; after-tax @21% = ~$147,906,037. Realized IN FY2025 -> already in current base, not incremental to next FY. AI is only part (also growth + expense mgmt). EPS% n/a: FY25 NI = -$443,151,000 (loss base). | 0 | |
Q1 FY2026 SG&A ratio -60bps to 15.2%
cost · soft | -60bps | 60bps YoY ratio improvement disclosed but Q1 FY2026 revenue not provided; 60bps x undisclosed Q1 revenue = unanchored $. Directionally supports the FY2026 full-year guide only. | | |
FY2026 SG&A guide 15.8-16.3% (~140bps YoY, tech+AI)
cost | -140bps to ~16.05% mid | Midpoint 16.05% vs 17.5% = ~140-145bps; ~140bps x $18,707,167,457 consensus FY26 rev = ~$261.9M pre-tax SG&A leverage; after-tax @21% = ~$207M if 100% realized. But mgmt splits this between scale-driven fixed-cost leverage and tech/AI; AI slice undisclosed. At ~50% AI attribution: ~$131M pre-tax / ~$103.5M after-tax (~28% of consensus FY26 NI $374M). rev=0 (cost-only). EPS% null: GAAP loss base; also already in published guidance/consensus. | 0 | |
Dozens of LLMs; 2 agentic AIs launching + 1 more
other · soft | dozens / 2+1 | Deployment-scale count; no run-rate revenue, cost savings, or deployment economics quantified. Unanchored. | | |
Assumptions: Next fiscal year = FY2026; sizing uses FY2026 consensus revenue $18,707,167,457 for the ~140bps SG&A guide. Incremental tax rate 21%. AI-attributed share of the 140bps assumed ~50% (mgmt language splits scale/fixed-cost leverage vs tech/AI; no exact split disclosed). FY2025 -160bps treated as historical (already in FY2025 base), excluded from forward aggregate. No quantified AI revenue claim exists (Lucie marketplace/agentic-commerce statements are forward and unquantified -> not sized). Cost claims map to SG&A; topline impact ~0. EPS% off the FY2025 GAAP base is invalid (NI = -$443M loss) per the loss-making guardrail; consensus turns positive only in FY2026 (NI $374M) and already embeds the SG&A guide, so no clean independent AI-EPS figure.
Top line: No hard adopter-side AI revenue claims. Every dollar-sized AI metric is a cost/efficiency lever (67% response-time cut, 86% Oswell deflection, SG&A ratio), not revenue. The only revenue-facing AI statements (Lucie platform broadening the individual market, agentic-commerce marketplace) are forward and unquantified -> not sizable. est_rev_uplift_pct = 0.
Bottom line: The one hard, forward, dollar-sizable lever is the FY2026 SG&A guide: ~140bps to ~16.05% = ~$261.9M pre-tax SG&A leverage on $18.707B consensus revenue (~$207M after-tax if 100% realized). At ~50% AI attribution: ~$131M pre-tax / ~$103.5M after-tax, ~$0.39/share vs consensus $0.85 EPS (~262M diluted shares) or ~28% of consensus FY26 NI $374M. But mgmt never isolates the AI slice, so 100% attribution overstates it. FY2025 is loss-making (NI = -$443M), so the EPS-uplift % is meaningless and set null per the negative-base guardrail. The 86% Oswell deflection and 67% latency cut are genuine cost levers with no disclosed cost base to convert to dollars.
[EPS uplift n/m (loss-making base)] The ~140bps FY2026 SG&A improvement is published guidance, and the 6-analyst consensus (FY26 rev $18.71B, EPS $0.85, NI $374M; FY27 EPS $1.28) is built off that same guidance. Consensus FY25->FY26 NI swing of ~+$708M on +56% revenue implies heavy embedded operating leverage; the AI-attributed half (~$103M after-tax) is material but not incremental to a guide analysts already model. No quantified AI claim points above what consensus assumes, and productivity metrics (67%, 86%) lack $ anchors to push estimates higher, so there is no gap to exploit.
MODEL CONSENSUS (impact)
partial
Cost-only AI story; FY2026 ~140bps SG&A guide is the sole sizable lever but already in consensus; EPS% suppressed by GAAP loss base.
Conflicts reconciled
- Q1 FY2026 -60bps row: X soft=false/rev_uplift_pct=0 vs Y soft=true/null -> used Y because Q1 revenue is not disclosed so the 60bps has no dollar anchor (unanchored=soft per guardrail)
- est_rev_uplift_pct: X=null vs Y=0 -> used 0 because anchored AI claims exist but are all cost levers, so revenue impact is determinately ~0
- Oswell type: X=engagement vs Y=productivity -> used productivity (automation/deflection is a cost lever)
- confidence: X=4 vs Y=5 -> used 4, lowered for the open conflicts
| Field | Opus 4.8 | GPT-5.5 |
|---|
| Rev uplift % | – | – |
| EPS uplift % | – | – |
| Priced in | high | – |
| vs analysts | inline | – |
| Confidence | 4 | – |
| Top line | Effectively none that is AI-attributable and quantified. Every dollar-sized AI claim is a cost/efficiency lever (SG&A ratio), not revenue. The only revenue-facing AI statements (Lucie platform 'bringing more people into the individual market', agentic-commerce marketplace) are forward and carry no figure -> not sizable. est_rev_uplift_pct = null. | – |
| Bottom line | The one hard, forward, dollar-sizable lever is the FY2026 SG&A guide: ~140bps to ~16.05% = ~$262M gross / ~$207M after-tax of SG&A reduction on $18.7B FY26 revenue. But management explicitly splits this between scale-driven fixed-cost leverage and 'technology and AI' — the AI slice is never isolated, so this overstates AI's contribution. And FY2025 is loss-making (NI = -$443M), so the EPS-uplift % is meaningless and is set to null per the thin/negative-base guardrail. The 86% Oswell deflection and 67% latency cut are genuine cost levers but have no disclosed cost base to convert to dollars. | – |
| Reasoning | The 140bps FY2026 SG&A improvement is published guidance, and the 6-analyst consensus (FY26 rev $18.71B, EPS $0.85, NI $374M; FY27 EPS $1.28) is built off that same guidance — the efficiency is already in the numbers. Consensus revenue growth of 11.70B->18.71B (+60%) dwarfs and already absorbs any AI efficiency. No quantified AI claim points above what consensus assumes, so there is no gap to exploit. | – |
Rows highlighted where the two models disagreed.
QUANTIFICATIONS
Care-guide response time reduction (Agentic AI bot): 67% (during peak and open enrollment period (FY2025), bottomline)
“For example, our Agentic AI bot for care guides reduced response times by 67% during peak and open enrollment period.”
Member questions handled by Oswell health agent: 86% (current (as of Q4 FY2025 call), both)
“Oswell, our industry-first Health agent now completes 86% of questions received from members with high accuracy and quality.”
Full-year SG&A ratio improvement (partly attributed to AI/technology): approximately 160 basis points year-over-year to 17.5% (full year 2025, bottomline)
“Our SG&A expense ratio of 17.5% improved by approximately 160 basis points over the prior year, reflecting continued efficiency gains through growth, disciplined expense management and AI and technology advancements across the business.”
Q1 SG&A ratio improvement (partly attributed to AI/technology): 60 basis points year-over-year to 15.2% (first quarter 2026, bottomline)
“Our SG&A ratio improved 60 basis points year-over-year to 15.2%, driven by disciplined expense management, top line growth and the growing impact of AI across our operations and member services.”
2026 SG&A expense ratio guidance (efficiencies expected from technology and AI): 15.8% to 16.3% (approximately 140 basis point year-over-year improvement at the midpoint) (full year 2026, bottomline)
“We expect the SG&A expense ratio to be in the range of 15.8% to 16.3%, representing an approximately 140 basis point year-over-year improvement at the midpoint. We continue to see the benefits of scale as fixed cost leverage and variable expense efficiencies driven by technology and AI are expected to drive further improvement in our SG&A expense ratio.”
Agentic AI / LLM deployment scale: dozens of LLMs; 2 agentic AIs launching; another in the next few months (forward from Q4 FY2025 call, bottomline)
“We have a dozens of LLMs on the back end of the business. and now 2 agentic AIs are about to launch another here in the next few months.”
PAST (realized)
- Q4 FY2025 | Mark Bertolini: We lowered administrative costs by 160 basis points year-over-year while significantly increasing membership.
- Q4 FY2025 | Mark Bertolini: our Agentic AI bot for care guides reduced response times by 67% during peak and open enrollment period.
- Q4 FY2025 | Mark Bertolini: Oswell, our industry-first Health agent now completes 86% of questions received from members with high accuracy and quality.
- Q4 FY2025 | Richard Blackley: The year-over-year improvement was driven by fixed cost leverage, lower exchange fee rates and disciplined cost management, including an increased impact from technology and AI initiatives.
- Q1 FY2026 | Richard Blackley: The approximately 60 basis point year-over-year improvement was driven by fixed cost leverage and disciplined expense management, including an increasing impact from technology and AI initiatives, partially offset by higher risk adjustment as a percentage of premium.
CURRENT (now)
- Q1 FY2026 | Mark Bertolini: Oscar is rapidly evolving our technology and deploying AI use cases at ever-increasing speed to drive growth, lower costs and help members make smart choices.
- Q1 FY2026 | Mark Bertolini: We recently launched several new transparency tools, including a real-time drug pricing feature that predicts when costs may cause a member to abandon a prescription.
- Q1 FY2026 | Mark Bertolini: We are also scaling new bilingual voice agents to support care navigation and improve speed to care.
- Q4 FY2025 | Mark Bertolini: AI is integrated across the Oscar platform, enabling teams to automate routine tasks, efficiently scale our service operations and improve decision support.
- Q4 FY2025 | Mark Bertolini: We continue to embed Oswell across our product portfolio to help members take control of their health.
- Q4 FY2025 | Mark Bertolini: We have a dozens of LLMs on the back end of the business. and now 2 agentic AIs are about to launch another here in the next few months.
FORWARD (guidance)
- Q1 FY2026 | Mark Bertolini: We will continue to add more AI solutions and health services on the Lucie platform to bring more people into the individual market.
- Q4 FY2025 | Mark Bertolini: drive operational excellence through AI and frictionless execution.
- Q4 FY2025 | Richard Blackley: variable expense efficiencies driven by technology and AI are expected to drive further improvement in our SG&A expense ratio.
- Q4 FY2025 | Mark Bertolini: And this is where our new Agentic AI tool is headed than having a marketplace where people can use the money that they receive for healthcare to buy what they want in their local market
TRACK RECORD — PROMISE vs DELIVERY
72/100 track record mixed 6 calls reviewed
Oscar talks up AI with strong point-in-time metrics but rarely commits to forward quantified AI targets; the one clear dated ops pledge ($60M 2026 admin savings tied to AI/tech) is still unverified, while a 2025 SG&A efficiency goal was met.
~$60M administrative cost elimination in 2026 from AI/technology efficiencies plus workforce actions — promised Q2 FY2025
too-early Reiterated in Q3–Q4 FY2025; Q1 FY2026 shows SG&A leverage and AI called out as a driver, but management never reported $60M realized and full-year 2026 is not closed in this transcript set.
~50 bps improvement in 2025 SG&A expense ratio vs initial full-year guidance (midpoint), citing AI/technology-driven efficiencies — promised Q2 FY2025
delivered FY2025 SG&A ratio finished at 17.5% (~160 bps YoY), inside the tightened 17.1%–17.6% range and better than the original 17.6%–18.1% guide.
Virtual urgent-care live chat to cut member response times ~90% and lift provider efficiency ~28% — promised Q1 FY2025
partial Reported as achieved in the same quarter; no earlier call set a dated numeric target to hit later.
LLM post-ER follow-up tool to lower readmissions ~10% (major health-system client) — promised Q4 FY2024
quietly-dropped Cited once as initial results; not repeated as a forward target or updated in later calls.
Oswell health AI agent to complete ~86% of member questions with high accuracy — promised Q4 FY2025
partial First disclosed as a result in Q4 FY2025; no prior quantified launch target in earlier calls.
>50% of Oscar Urgent Care onboarding and post-care instructions AI-powered — promised Q4 FY2024
partial Stated as current penetration, not a future milestone; later calls cite other AI metrics but do not track progress vs 50%.
PRICED-IN (REFINED)
MEDIUMEst. revisions rising · Fwd P/E -16.6 · EV/Sales 0.1x
Price targets stair-step higher (lastMonth 23.67 > lastQuarter 21.75 > lastYear 18.09) and consensus embeds a sharp 2026 profitability inflection (revenue ~$18.7B, EPS ~$0.85 vs 2025 losses), with ratings drifting less bearish since January—signals the market is already lifting the AI/efficiency narrative into estimates. Valuation is not stretched on hard multiples: EV/Sales ~0.1 and P/S ~0.41 are depressed for a health insurer, and next-FY fwd P/E is negative on continued 2025 losses, so the stock is not clearly “paying up” like a mature AI winner. Segment mapping is unavailable (null segmentation), so AI upside cannot be tied to a specific reported line. Net: rising revisions argue more upside is priced than not, but cheap sales-based multiples cap the verdict at medium rather than high.
COVERAGE — ENTHUSIASM TRAJECTORY + CATALYSTS
8Q4 FY20246Q1 FY20255Q2 FY20258Q3 FY20259Q4 FY20257Q1 FY2026
AI enthusiasm across 6 calls — trend ↗ rising
Operational LLM tools faded mid-crisis, then Oswell and agentic KPIs made AI a named strategic driver.
RECENT AI CATALYSTS & NEWS
BUSINESS IMPACT - QUALITATIVE MATERIALITY
7/10 qualitative impact material near-term · mixed evidence
Where AI matters: SG&A leverage and member-service automation
Deployed LLMs, agentic care-guide bots, Oswell (~86% question resolution), and voice agents are driving measurable service speed and admin-ratio improvement (e.g., ~160 bps FY25 SG&A, ~60 bps Q1 FY26), but management never isolates an AI $ slice and there is no separate AI revenue line—only bundled cost/retention levers already embedded in profitability guidance.
Caveats: SG&A gains conflate AI with membership scale and generic expense discipline—true AI attribution is unquantified; Productivity KPIs (67%, 86%) lack disclosed cost bases, so dollar savings remain soft; Forward Lucie/marketplace AI revenue claims are unquantified and not yet a business line; Peer insurers can replicate service automation, narrowing differentiation over time
AI DISRUPTION / CANNIBALIZATION RISK tailwind · 3/10
AI does not automate away regulated risk-bearing insurance; it mainly deflates Oscar’s own contact-center and admin cost stack while transparency/navigation tools are offensive retention plays, not a commoditization of premiums.
OPTIONS / MARKET STRUCTURE
option liquidity: fair
proxy inputs — dollar-ADV $152M · beta 2.341 · px $21.14
source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.
CONFIRMATION — INSIDERS · 13F · LANGUAGE
Confirming — insiders buying, institutions adding, management language 5/10 measured.
INSIDERS buying 1 open-market buy(s) vs 18 sell(s) — net accumulation
INSTITUTIONS (13F) adding as of 2026-03-31: 61 new / 74 closed positions; 182 increased / 102 reduced; institutional ownership -3.42pp; -12 net 13F holders
MGMT LANGUAGE 5/10 measured Brief AI mentions: live deployments named, but impact unquantified and future additions framed softly.
commit “Oscar is rapidly evolving our technology and deploying AI use cases at ever-increasing speed to drive growth, lower costs”
commit “We are also scaling new bilingual voice agents to support care navigation and improve speed to care.”
commit “the growing impact of AI across our operations and member services”
VERBATIM AI QUOTES
“Our SG&A ratio improved 60 basis points year-over-year to 15.2%, driven by disciplined expense management, top line growth and the growing impact of AI across our operations and member services.”
— Mark Bertolini, Q1 FY2026
“Oscar is rapidly evolving our technology and deploying AI use cases at ever-increasing speed to drive growth, lower costs and help members make smart choices.”
— Mark Bertolini, Q1 FY2026
“We recently launched several new transparency tools, including a real-time drug pricing feature that predicts when costs may cause a member to abandon a prescription. The tool instantly cross-references deductible status, local supply and pricing and guides members to lower-cost pharmacies or equally efficacious alternatives in the network.”
— Mark Bertolini, Q1 FY2026
“We are also scaling new bilingual voice agents to support care navigation and improve speed to care.”
— Mark Bertolini, Q1 FY2026
“We will continue to add more AI solutions and health services on the Lucie platform to bring more people into the individual market.”
— Mark Bertolini, Q1 FY2026
“The approximately 60 basis point year-over-year improvement was driven by fixed cost leverage and disciplined expense management, including an increasing impact from technology and AI initiatives, partially offset by higher risk adjustment as a percentage of premium.”
— Richard Blackley, Q1 FY2026
“Our SG&A expense ratio of 17.5% improved by approximately 160 basis points over the prior year, reflecting continued efficiency gains through growth, disciplined expense management and AI and technology advancements across the business.”
— Mark Bertolini, Q4 FY2025
“Member retention remains solid across the book, driven by our suite of affordable products, agenetic AI features and a superior member experience.”
— Mark Bertolini, Q4 FY2025
“Oscar investments in AI are creating efficiencies across the business as we grow. We lowered administrative costs by 160 basis points year-over-year while significantly increasing membership. AI is integrated across the Oscar platform, enabling teams to automate routine tasks, efficiently scale our service operations and improve decision support.”
— Mark Bertolini, Q4 FY2025
“For example, our Agentic AI bot for care guides reduced response times by 67% during peak and open enrollment period.”
— Mark Bertolini, Q4 FY2025
“AI is also central to our member experience. Oswell, our industry-first Health agent now completes 86% of questions received from members with high accuracy and quality. We continue to embed Oswell across our product portfolio to help members take control of their health.”
— Mark Bertolini, Q4 FY2025
“The impact of AI on our efficiency and the quality of the interactions for our members is unparalleled in this pace in my 40 years in this industry.”
— Mark Bertolini, Q4 FY2025
“Third, drive operational excellence through AI and frictionless execution.”
— Mark Bertolini, Q4 FY2025
“The year-over-year improvement was driven by fixed cost leverage, lower exchange fee rates and disciplined cost management, including an increased impact from technology and AI initiatives.”
— Richard Blackley, Q4 FY2025
“We continue to see the benefits of scale as fixed cost leverage and variable expense efficiencies driven by technology and AI are expected to drive further improvement in our SG&A expense ratio.”
— Richard Blackley, Q4 FY2025
“AI, where we're able to create a better member experience and greater stickiness, and we're seeing that on a regular basis. We have a dozens of LLMs on the back end of the business. and now 2 agentic AIs are about to launch another here in the next few months. So we're now having a lot of impact where people can access us quicker with much more accuracy and without having to wait on phones, which would also again reduces our costs.”
— Mark Bertolini, Q4 FY2025
“And this is where our new Agentic AI tool is headed than having a marketplace where people can use the money that they receive for healthcare to buy what they want in their local market, a narrow network with a plan design that changes with their life, starts to create the opportunity for lifetime value of membership and change the investment thesis that insurance companies would have in managing that membership and how we would approach it, which leads to the lifestyle products.”
— Mark Bertolini, Q4 FY2025
ANALYST QUESTIONS ON AI
Q (Q4 FY2025, Samuel Becker (Goldman Sachs, for Scott Fidel)): What are your levers — key levers to achieving EBITDA profitability without the extension of the enhanced subsidies? And what are those key headwinds or tailwinds when thinking about MLR and SG&A from 2025 to 2026.
A: Mark Bertolini: Well, there are a number of them. First, it's growth. So it's growth drives a reduction in overall percentage of costs. AI, where we're able to create a better member experience and greater stickiness, and we're seeing that on a regular basis. We have a dozens of LLMs on the back end of the business. and now 2 agentic AIs are about to launch another here in the next few months. So we're now having a lot of impact where people can access us quicker with much more accuracy and without having to wait on phones, which would also again reduces our costs. And then on the MLR front, we are constantly working on our contracts and our utilization management, and we task the team to deliver so many hundred basis points every year and opportunities to keep our trend in line with where we think the market should be. And so all of those things together, and there are a lot of levers that we manage every day through the management process are the things that we track to make sure that we commit our targets. Richard Blackley: And Sam, I just want to make 1 point really clear. Our guidance is on EBIT. So it's not on adjusted EBITDA.