Western Digital details LTAs workload evolution, with Agentic AI as the primary, data-intensive driver for storage via inference, and Physical AI expanding data needs through real-world and synthetic data in autonomous vehicles and robotics. Across 2027 and into 2028–2030, the bulk of workloads remains, but growth vectors continue to compound.
Generated by Dafinchi AI. Source-grounded AI analysis, not investment advice.
What workloads are expected under LTAs—Agentic AI versus physical AI—and how do these workloads differ across 2027 and beyond (2028–2030)?
Management describes that the bulk of customer workloads remain the same, but that the primary AI growth driver is shifting from training/development to inference and then to Agentic AI. Specifically, they note that “the primary drivers… over the last 24, 36 months have been model training and development,” while “growth [is] driven by inference, Agentic AI” going forward. 1
They further characterize the workload nature of Agentic AI as more data-intensive and more persistent than earlier AI phases: agents “generate data at every step of the workflow,” which increases both the volume and the amount stored over time; management calls Agentic AI a “structural and step function driver of capacity-orientated storage demand.” 2
They also tie this workload behavior to how inference drives storage: inferencing and Agentic growth generate more data to be stored for reinforcement learning and for retaining generated context, and they argue storing that context is more economical than re-running compute/memory. 3
Alongside Agentic AI, management says they are seeing the “early innings of growth… driven by physical AI” and that they are engaging with large enterprise customers in physical AI with visibility into their growth trajectories. 1
They define the physical AI workload mechanism as a further extension of the AI data accumulation cycle: autonomous vehicles, robotics/industrial automation systems, and humanoids require real-world data to train, and when that data is insufficient, they generate and store synthetic data sets—which becomes “another driver of storage demand.” 2
They also give a concrete example of how this is scaling: for an “autonomous vehicle” example, management states the increase in exabyte demand for calendar year ’27 has increased multiple fold, and they explain that these players use their collected data plus synthetic data generation (to support training via AI tools and reinforcement learning), which “requires a lot more storage data.” 4
For the near-term “calendar year ’27,” management explicitly points to physical AI exabyte demand increasing multiple fold, tied to autonomous vehicles expanding data generation and synthetic data for model training and reinforcement learning. 4
At the same time, the broader AI workload evolution described for storage demand includes Agentic AI becoming the key forward driver (inference → Agentic), with a persistent, multi-step data generation pattern that compounds storage requirements. 2
So, for 2027, the picture presented is:
In response to the specific question about how workload types differ through time under LTAs (from near-term 2027 into ’28, ’29, and ’30), management states: “the bulk of the workloads remain the same.” 1
However, they simultaneously identify growth vectors that are expected to extend beyond 2027:
Putting those together: for 2028–2030, management’s framing implies no step-change to a completely different storage workload profile, but rather continued compounding of the same AI data creation dynamics—first via inference/Agentic AI, and increasingly via physical AI’s real-world + synthetic data loops. 142
| Dimension | Agentic AI under LTAs | Physical AI under LTAs | Timing emphasis (as described) |
|---|---|---|---|
| Core workload nature | Agents “do the work, coordinating tasks… operating continuously across multistep workflows,” creating a “fundamentally more data-intensive workload” that is increasingly persistent. 2 | Real-world data is insufficient for full training, so systems generate and store synthetic data sets; this increases storage demand. 2 | Agentic AI emphasized as the forward growth driver. 1 Physical AI described as starting “early innings” but becoming “very real,” with ’27 multiple-fold exabyte growth examples. 14 |
| Data generation & storage behavior | Agents generate data at every step; both volume and stored-over-time amounts rise. 2 | Synthetic data generation and storage become an explicit driver because training requires more data than is available directly from the real world. 2 | Physical AI said to extend “’27 and beyond.” 4 |
| “Bulk remains the same” implication | The workload base does not radically change, even as the AI component evolves toward Agentic. 1 | Physical AI is an added growth vector layered onto continuing AI/inference dynamics rather than a replacement. 142 | For ’28–’30, management indicates continuity in the “bulk,” with growth from these vectors. 1 |
Under LTAs, management expects Agentic AI to be a major driver because it creates step-function, persistent, multi-step data generation that increases both data volume and retention requirements. 2
They also expect physical AI workloads—autonomous vehicles, robotics/industrial automation, and humanoids—to become increasingly important: these systems require more data than direct real-world collection provides, so they rely on synthetic data generation and retention, which further drives storage demand. 2
Across time, management’s key distinction is that for 2027 physical AI is already showing measurable, multi-fold exabyte demand growth in examples, 4 while for 2028–2030 the “bulk of workloads remain the same,” but the growth vectors (inference/Agentic AI and physical AI) continue to build. 142
Disclaimer: The output generated by dafinchi.ai, a Large Language Model (LLM), may contain inaccuracies or "hallucinations." Users should independently verify the accuracy of any mathematical calculations, numerical data, and associated units, as well as the credibility of any sources cited. The developers and providers of dafinchi.ai cannot be held liable for any inaccuracies or decisions made based on the LLM's output.
Western Digital outlines a dual-path outlook: LTA pricing remains contract-driven and predictable, particularly for nearline pricing, while offering upside in non-nearline contracts, and the exabyte growth trajectory remains robust at 25% plus, supported by 40TB ePMR ramp and upcoming HAMR and higher-capacity drives, with visibility extending through 2029–2031. Quarterly margins may vary due to LTA timing, but the long-term demand signal remains strong.
Sources used
Research questionWhat did management say about LTA pricing and exabyte growth outlook?
Answer outline
🚀 Western Digital's sales have surged over five years driven by AI demand & booming data center growth, with innovative HDD/SSD tech fueling revenue & margin gains! 📈🤖
Sources used
Research questionSummarize how Western Digital’s sales trends have evolved over the past five years, focusing on the impact of AI-driven demand and data center growth on its revenue mix. Highlight evidence of increasing product sales linked to AI workloads, changes in pricing or volume, and management’s commentary on the future financial outlook and growth strategy.
Answer outline
Western Digital's pricing strategy in Q3 2026 is shaped by structural growth in AI data and increased HDD demand. The company emphasizes predictable, value-based pricing supported by capacity leadership and long-term customer commitments.
Sources used
Research questionHow will WDC's pricing strategies be affected by AI data growth and HDD storage demand in Q3 2026?
Answer outline
Applied Materials' management outlines a robust DRAM growth trajectory for 2026, with heavy emphasis on the second half as customers expand clean room capacity. They also signal durable multiyear demand into 2027, driven by AI memory expansion and a persistent supply-demand gap, positioning Applied to gain share and sustain high visibility through 8-quarter rolling forecasts.
Sources used
Research questionWhat did management say about DRAM growth outlook and multiyear visibility?
Answer outline
Teradyne outlines a near-term CPO testing opportunity of $300 million to $700 million by 2028, with the final TAM hinging on 2027 ramp progress. The company also expresses strong confidence in sustained double-digit networking growth over the next three years, anchored by 15%–20% transistor growth and a broad platform shift across copper-to-backplane transitions, pluggables, NPO, and CPO, supported by robust switch-silicon demand.
Sources used
Research questionWhat is the potential near-term size of the CPO testing opportunity by 2028, and how confident is management in sustained double-digit networking growth over the next three years?
Answer outline
Schwab outlines why its roughly 12 million daily trades appear sustainable, attributing durability to structural shifts in participation—especially young investors—plus a durable AI-enabled research and trading workflow, and a regulatory tailwind from pattern day trader rule changes, with market volatility and broad interest providing near-term support.
Sources used
Research questionWhat factors make the 12 million daily average trades appear sustainable, and which elements (young investor growth, options trading comfort, pattern day trader rule change, AI usage, crypto trading) are most structural versus market-environment driven?
Answer outline
CrowdStrike reports record pipeline momentum and Mythos-driven urgency turning cybersecurity into a strategic AI enabler, with AI protection discussions driving demand through Q1 2027.
Sources used
Research questionWhat did management say about Record pipeline and Mythos-driven demand?
Answer outline
Management explains that Mythos crystallized a cybersecurity bottleneck into an AI adoption accelerator, fueling a record Q2 pipeline and accelerating demand across multiple modules. They highlight the AI AIDR pipeline, executive engagement, and early momentum as evidence for continued confidence in Q2 and full-year guidance.
Sources used
Research questionWhat did management say about Record pipeline and Mythos-driven demand?
Answer outline
Veeva outlines where pharma AI adoption will begin—agentic automation across clinical/regulatory workflows—and frames the SaaS-to-AI shift as MAAP (Models, Agents, Applications). Investors should look for headless agent execution and measurable productivity gains.
Sources used
Research questionWhat areas of pharma are likely to adopt AI first, and how should investors think about transitioning from traditional SaaS to AI in pharma?
Answer outline
Broadcom's Q2 2026 earnings discussion forecasts a second wave of AI demand driven by enterprises and consumers via tokens and cloud APIs, with most compute demand still routed through frontier labs. Out-year gigawatt commitments from Anthropic, OpenAI, and Meta anchor ongoing growth through 2027–2029.
Sources used
Research questionDo you expect a second wave of demand as AI expands into enterprises and cloud services, and how does that compare to hyperscaler demand in terms of gigawatt commitments?
Answer outline
This analysis explores how Qualys' AI-driven risk management solutions are poised to shape cybersecurity market investments in 2026, highlighting strategic implications and future trends.
Sources used
Research questionHow is Qualys' AI-driven risk management expected to influence cybersecurity market investments in 2026?
Answer outline
🚀 A detailed comparison of staffing challenges and labor market outlook from Korn Ferry and ManpowerGroup highlighting AI impact and sectoral trends. 🔍
Sources used
Research questionCompare challenges in staffing across various sectors and future outlook for the labor market
Answer outline
Dafinchi