In the high-stakes world of semiconductor manufacturing, where sub-2nm process nodes and advanced CoWoS packaging push physical limits, operational time is measured in milliseconds. Yet inside many of the world’s most sophisticated facilities, critical data still moves at the sluggish pace of hourly batch updates and periodic API requests.
In a fab, that timing delay carries a steep price tag. A microscopic variance during etching or chemical mechanical planarization can jeopardize an entire wafer lot. If an anomaly isn’t detected instantly, dozens of defective wafers can pass through downstream tools before anyone notices.
“The biggest challenge we see from a data perspective is that data lives in silos,” explains Phil Scanlon, Field CTO and Head of Solution Engineering (APJ & ME) at Solace. “When modernization initiatives rely on getting that data, it’s not just about having it somewhere like a data warehouse—it’s about having the real-time context of that data outside of those silos.”
Flipping the Model: Event-Driven Manufacturing
To eliminate these blind spots, forward-thinking manufacturers are shifting toward an architecture known as Event-Driven Manufacturing.
Instead of asking systems to periodically poll databases or bundle updates into scheduled batches, an event-driven framework treats every significant moment—a wafer passing a quality gate, an equipment sensor anomaly, or a raw material order—as an instantaneous “event”. Through an enterprise “event mesh,” these events are published instantly across the network. Any system, analytics engine, or shop-floor tool subscribed to the mesh can react immediately.
Rather than tightly coupling point-to-point connections that break when applications update, an “event mesh” acts like a decoupled central nervous system for the factory. Data is liberated from individual applications, enabling proactive yield protection and instant operational visibility.
This pattern is already running in advanced semiconductor packaging. Absolics, a US-Korean manufacturer of advanced packaging technology for high-performance computing, built a cloud-based event mesh on Solace to connect its manufacturing, management, and product development applications into one integrated environment. The shift replaced a patchwork of point-to-point connections with a single events layer, giving Absolics high-throughput data streaming with no downtime and the real-time visibility to make faster, better-informed decisions on the floor.
Stress-Testing Supply Chain Volatility
This need for real-time responsiveness extends well beyond cleanroom walls. Modern chipmakers operate within a volatile global landscape shaped by natural events and geopolitical tensions. Recent occurrences—from earthquakes in key manufacturing hubs like Kumamoto to maritime bottlenecks along the Strait of Hormuz and the Suez Canal—have exposed the fragility of static supply chain schedules.
When disruptions hit, static daily reports leave supply chain managers flying blind. Manufacturers need live, end-to-end visibility across their entire footprint, from specialized chemical gas shipments to substrate inventory.
To solve this, Solace frequently co-deploys alongside supply chain planning platforms like Kinaxis. While planning engines calculate complex optimization models, Solace feeds live shop-floor yield events and operational sensor data directly into the planning pipeline. This complementary pairing allows supply chain managers to automatically adjust production schedules the moment physical reality strays from the original forecast.
“Organizations that can detect, respond, and react in real time using an event mesh as the basis of the pulse of their business are better placed to win in competitive environments,” said Scanlon.
Fueling Agentic AI with Live Context
As semiconductor leaders move beyond pilot projects into deploying Agentic AI—autonomous software agents capable of making operational decisions—the data freshness problem has become impossible to ignore.
Autonomous agents are only as accurate as the information feeding them. When agents rely on stale batch data, they suffer from data drift and hallucinations, leading to risky decisions in production environments. Live, contextual event streams provide the clean operational baseline these agents require.
To help enterprises bridge the gap between AI prototypes and production, Solace recently introduced its Agent Mesh capability. It provides the necessary orchestration, security, governance, and real-time observability required to deploy autonomous agents safely at enterprise scale.
Evolution, Not a ‘Rip-and-Replace’ Nightmare
For fab executives, the prospect of introducing new software often triggers valid fears of expensive, disruptive overhauls to core Enterprise Resource Planning (ERP) or Manufacturing Execution Systems (MES).
Scanlon emphasizes that event meshing isn’t about replacing existing infrastructure. Instead, it acts as a complementary layer that unlocks data from legacy applications without disrupting core operations. This pragmatic approach is validated by Solace’s deep strategic alliance with SAP, which OEMs Solace’s streaming tech under the product name SAP Advanced Event Mesh to help industrial enterprises modernize their digital cores.
Beyond the fab floor, this architecture comes battle-tested in environments where operational downtime simply isn’t an option. Drawing on over two decades of dedicated focus on real-time data movement, Solace currently streams more than one trillion messages a day across global industrial and supply chain environments—from automotive leaders like Bosch to aerospace pioneers like Airbus, which transformed spare part tracking from hours down to milliseconds. That same financial-grade backbone underpins 75% of the world’s top investment banks, handling extreme pressure tests like Royal Bank of Canada processing 278 billion messages in a single trading day with zero outages.
Looking Ahead to SEMICON Taiwan
Ultimately, building a smarter fab isn’t about chasing tech buzzwords or attempting an overnight overhaul. As the semiconductor industry gathers in Taipei, the focus remains squarely on practical, high-impact agility.
“More data doesn’t necessarily mean better results,” Scanlon reflects. “The real differentiating factor is having real-time data available. A manufacturing plant isn’t going to throw everything away and start again—it’s about making meaningful, incremental changes. As you expose more information from existing systems into a real-time event layer, every innovation and AI model built on top becomes far more valuable.”
