How Duke Energy’s 2026 Initiatives Drive Customer Benefits and Strategic Growth

Expected impact of Pinterest’s AI personalization strategy on user engagement and capital expenditure in Q1 2026

Based on Pinterest’s Q1 2026 earnings transcript excerpts, management frames its AI personalization strategy as a compounding “relevance flywheel” that should increase user engagement while also containing AI infrastructure cost growth, using a mix of proprietary and in-house models.

---

1) User engagement: expected to improve via “more relevant and personalized” experiences

A. Engagement thesis: AI personalization increases “reasons to come back”

Pinterest says platform improvements across search ranking, content recommendations, and creative generation are converging on a “more relevant and personalized experience” that “gives users more reasons to come back,” and that is “built off” proprietary signals and unique curation behavior. ¹

B. Evidence of engagement strength tied to AI personalization (already visible in Q1 trends)

Pinterest highlights:

• “Ten straight quarters of double-digit user growth” with “deepening engagement per user,” attributing this directly to multi-year investments in AI improving personalization and curation within visual search and discovery. ²³
• Management also reiterates record-high users and emphasizes that 100% of reported users are logged in and 85% come directly to the mobile app, characterizing Pinterest as a “clear destination app.” ⁴

C. Quantification of AI personalization gains inside Q1 (search experience and creative outputs)

Pinterest provides concrete performance improvements from specific AI systems:

• PinRack (a personalized generative retrieval system) extension in Q1 improved search fulfillment by ~180 basis points and reduced advertiser CPA/CPC by ~180 basis points. ⁵
• A Q1 update to Pinterest’s search ranking model extending user context windows “30-fold” (using “up to 16 thousand user actions over a two-year period”) improved search fulfillment by ~70 basis points and saved ~390 basis points. ⁵

D. Why personalization matters to engagement (mechanism)

Pinterest explicitly lays out the engagement mechanism:

• Its taste graph captures visual intent and curation signal, and every search/click/save provides more signal to deliver more relevant/personalized experiences. ³
• Management states: higher relevance drives deeper engagement, engagement supports retention, and retention brings users back with higher intent. ³

Net expectation for Q1 2026 engagement: Pinterest’s commentary indicates AI personalization is not just incremental—it is intended to sustain and deepen engagement through higher relevance across surfaces (notably search and discovery). ¹³

---

2) Capital expenditure (capex): expected to be supported by cost-efficient AI approaches (lower-cost model strategy)

The excerpts don’t provide a numeric Q1 2026 capex forecast. However, they do provide management’s rationale for how AI personalization can scale while limiting incremental cost pressures—i.e., the strategy is designed to improve relevance and capabilities without requiring only expensive large third-party models.

A. Cost control via in-house / fit-for-purpose models and post-training

Pinterest states it does not believe the economics of relying on large proprietary third-party LLMs “make sense” for many use cases due to cost premiums and overengineering risk, and it emphasizes compact, fit-for-purpose models trained on Pinterest-specific data. ⁶

Pinterest further says for generalized LLM capabilities it uses open-source models in its own cloud environment and post-trains them on proprietary data, which Pinterest argues yields advantages including security, lower latency, and ‘a fraction of the cost.’ ⁶

Implication for capex in Q1 2026: the strategy suggests incremental compute and model spend needed to deliver personalization may be managed more conservatively through model choices and post-training, rather than scaling only with expensive third-party LLM usage. ⁶

B. Explicit example of “lower cost than leading third-party models” in creative AI (Canvas)

For Pinterest’s creative-generation AI:

• Management says Canvas operates “at an order of magnitude lower cost than leading third-party models.” ⁷
• They expect to expand Canvas to enable more creative experiences in coming months and quarters as model versions improve (including real-time, high-fidelity image editing). ⁷

Implication for capex: because Canvas is described as significantly lower-cost than alternatives, scaling personalization/creative AI capabilities may require less incremental spend than if Pinterest used only third-party model endpoints at large scale. ⁷

C. Search model efficiency and “saves” signal

Pinterest’s Q1 search ranking updates include both user-facing and economic outcomes:

• The search ranking update improved search fulfillment and produced “saves by approximately 390 basis points.” ⁵

While “saves” is not explicitly labeled as capex, it is a direct efficiency indicator tied to operational/model improvements and is consistent with the company’s cost-managed AI strategy. ⁵

---

3) Putting it together: expected impact on engagement and capex balance in Q1 2026

Engagement outcome expectation

• Pinterest expects AI personalization to continue driving engagement by producing more relevant and personalized experiences that increase repeat usage and retention behavior. ¹³
• Q1 already shows AI-driven improvements in search fulfillment and measurable advertiser outcomes from personalization systems, reinforcing that engagement improvements are likely grounded in product performance rather than marketing alone. ⁵

Capex expectation

• Although Q1 2026 capex (or guidance) is not quantified in the excerpts, Pinterest’s AI approach is explicitly designed to scale personalization while controlling unit economics—favoring in-house and fit-for-purpose models, plus post-trained open-source models, and using creative AI systems with order-of-magnitude lower costs than leading third-party models. ⁶⁷
• The presence of “saves” tied to search model updates suggests Pinterest is targeting efficiency improvements alongside relevance gains. ⁵

---

Bottom line

In Q1 2026, Pinterest’s AI personalization strategy is expected to increase user engagement by delivering substantially more relevant, personalized visual discovery and shopping experiences (mechanically tied to relevance → deeper engagement → retention). ¹³ At the same time, the strategy is structured to limit capex intensity (or at least incremental AI infrastructure cost pressure) through cost-efficient in-house/fit-for-purpose models and lower-cost AI systems (e.g., Canvas) rather than relying primarily on expensive third-party LLM services. ⁶⁷

(No numeric Q1 2026 capex impact or forecast is stated in the provided excerpts, so the capex impact can only be inferred directionally from Pinterest’s stated unit-economics and cost-efficiency rationale.) ⁶