The rapid expansion of artificial intelligence infrastructure is acting as the main catalyst for a deep technological transformation in semiconductor manufacturing and data center equipment.
According to recent reports from Goldman Sachs and TrendForce, annual capital expenditures on AI infrastructure will more than double over the next five years. At the same time, the market for optical connections designed to accelerate data transfer within server racks is expected to grow nearly 400‑fold.
Experts at Lithuanian company LITILIT, a manufacturer of high‑precision laser equipment, note that the massive rollout of AI networks is creating a shortage of high‑precision material‑processing tools that avoid thermal damage.
“Many advanced AI components – from semiconductor housings and glass substrates to optical interconnects and complex printed circuit boards (PCB) – require material‑processing technologies that do not cause thermal deformation. Femtosecond lasers are becoming a key tool for this transition,”
Bottlenecks in Manufacturing and the Role of Femtosecond Lasers
Goldman Sachs estimates that global capital spending on AI infrastructure will rise from $765 billion in 2026 to $1.6 trillion by 2031. The primary drivers are surging demand for compute power, construction of data centers, and the supporting energy infrastructure.
Glass Interposers and Optical Connections: Technical Details
According to TrendForce, the market for optical interconnects using CPO (Co‑Packaged Optics) and NPO (Near‑Packaged Optics) will expand from roughly $100 million in 2025 to over $39 billion by 2030.
Femtosecond lasers generate ultrashort pulses lasting only a quadrillionth of a second. This enables precise micro‑processing of fragile materials such as glass:
- Glass interposers: The laser drills microscopic holes in a glass substrate, which are later filled with copper to form vertical connections for chip‑to‑chip communication.
- Absence of thermal shock: Unlike mechanical drilling or conventional lasers, the femtosecond pulse delivers energy so quickly that the surrounding glass does not have time to heat up, preventing micro‑cracks and chips.
Market Dynamics of AI Infrastructure and Optical Interconnects (2025–2031)
| Indicator / Metric | 2025‑2026 | Forecast for 2030‑2031 | Growth Dynamics |
| AI infrastructure capex (Goldman Sachs) | $765 bn (2026) | $1.6 tn (2031) | Growth 2.1× |
| Optical interconnect market CPO/NPO (TrendForce) | ~$100 m (2025) | >$39 bn (2030) | Growth 390× |
| Main laser applications | Glass substrates, CPO, PCBs | Chiplets, photonic interfaces, quantum data centers | Shift to non‑thermal processing |
Automation and Scalable Production: LITILIT’s Solution
Despite strong demand, industrial production of femtosecond lasers remains a complex engineering challenge. Most existing systems originate from laboratory scientific setups, complicating automation and requiring highly skilled specialists during assembly.
To address this issue, LITILIT, in cooperation with the Center for Physical Sciences and Technology (FTMC) in Vilnius, has developed and patented a new technological platform (patents credited to Kęstutis Regelskis, Nerijus Steponkus, and Nikolajus Gavrilinas). The technology enables the creation of simplified‑design femtosecond lasers with a high level of manufacturing automation.
In June 2026, LITILIT began construction of a specialized laser factory in Vilnius, with plans to scale this model together with international partners.
Xpert Take
The AI data center boom is reshaping not only silicon architectures but also the market for precision industrial equipment. The shift from traditional copper routing to glass interposers and optical modules such as CPO demands a fundamental change in material‑processing tools. Femtosecond lasers have become a foundational link in the AI hardware supply chain, and automation of their production—exemplified by companies like LITILIT—will help eliminate technological barriers on the path to next‑generation data centers.
