An Australian National University spinout that treats AI's memory wall as a manufacturing problem - and builds the wires to solve it.
Modern AI runs on staggeringly fast processors. But those processors spend much of their time waiting - for data to arrive from memory and from neighbouring chips. Syenta exists to shorten that wait.
Syenta is a deep-technology semiconductor company headquartered in Sydney, spun out of the Australian National University in 2022. It does not design chips or run a giant fab. Instead, it works on the layer between chips - the advanced packaging and interconnects that stitch processors, memory and accelerators into a single high-performing system.
Its core invention is a process called Localized Electrochemical Manufacturing, or LEM. Conventional advanced packaging builds fine metal connections through many sequential steps: deposit a metal, coat it with resist, expose a pattern, etch away the excess, strip the resist, and repeat. Syenta's approach collapses two of the most important steps - deposition and patterning - into a single electrochemical move, using stamp electrodes carrying a dielectric pattern to place metal precisely where it is needed.
The result, according to the company, is micron and sub-micron interconnects laid down with roughly 40 percent fewer process steps, on packages larger than 1,000 square millimetres, at up to three times the throughput of existing fabrication approaches. Syenta describes connections up to 100 times thinner than a human hair - the kind of density that lets far more data move between chips per second.
In the company's own framing, the goal is blunt: "The compute-memory interface is no longer the bottleneck for AI." That is the promise Syenta is trying to manufacture into reality.
Over two decades, raw compute raced ahead. The connections feeding it did not. That widening gap is the bottleneck Syenta targets.
Approximate relative improvement over ~20 years, as cited by Syenta and reported in Australian tech press. Bars are illustrative, not to exact scale.
This is what engineers call the "memory wall." As AI models balloon, accelerators increasingly stall waiting for data rather than crunching it. More silicon does not help if the data cannot get in and out fast enough. Syenta's bet is that the most valuable place to intervene is not the transistor - where thousands of engineers already compete - but the interconnect, a quieter chokepoint where a manufacturing breakthrough can unlock system-wide gains.
Incumbent packaging from the big foundries and OSATs relies on multi-step deposit-and-etch flows. Syenta's differentiation is procedural: do more in a single electrochemical step.
Companies like TSMC, Intel, ASE and Amkor already offer sophisticated advanced-packaging platforms - CoWoS, Foveros, EMIB and their peers. Syenta is not trying to out-scale them at their own game. It is proposing a different physics of fabrication, one that removes several traditional steps, including much of the etching and resist work, and with them the associated cost, time and chemical waste.
That has two knock-on effects the company emphasises. First, fewer steps mean fewer opportunities for defects and lower cost per package - useful when the goal is high-volume production. Second, skipping resist and etch stages reduces the chemical footprint of manufacturing, which Syenta frames as a sustainability advantage. That angle earned it a semi-finalist spot in the 2025 SEMI S3 program for sustainability and semiconductors.
Syenta organises its roadmap around three axes - a clean way to explain a hard-tech vision that many startups never manage.
Push interconnect lines finer to raise bandwidth.
Extend across bigger packages and field sizes.
Move from demonstration toward volume manufacturing.
Syenta's technology is a foundation others build on - a way to fabricate the interconnect layer that AI and high-performance systems increasingly depend on.
The proprietary single-step deposit-and-pattern process at the heart of everything Syenta makes.
Micron and sub-micron redistribution layers that raise chip-to-chip density and bandwidth.
Interconnects across packages >1000 mm² for 3D and wafer-level system integration.
Who uses it. Syenta's customers are the makers of advanced systems - semiconductor manufacturers, foundries and advanced-packaging partners serving AI accelerator, high-performance computing and, increasingly, quantum chip makers. The same process is also applicable beyond AI, to sensors, photovoltaics, batteries and printed circuit boards, wherever fine metal patterning matters.
The business model. Syenta is a B2B deep-tech company. Rather than selling to consumers, it develops proprietary LEM tools and processes and aims to commercialise them with manufacturing and packaging partners across the AI and HPC supply chain, scaling toward high-volume production later this decade. As an early-stage company, its reported revenue is modest; the value today sits in the technology, the intellectual property and the roadmap.
Genuinely new manufacturing needs genuinely mixed disciplines. Syenta's founders come from electrochemistry, materials science, applied physics and neuroscience.
CEO Dr Jekaterina Viktorova co-invented LEM and holds a PhD in localized electrodeposition - the exact niche the company now commercialises. CTO Ben Wilkinson brings a background in applied physics, photovoltaics and metallisation. Professor Luke Connal contributes deep expertise in polymer chemistry and additive manufacturing, while co-founder Zachary Dowse, a neuroscientist and serial entrepreneur, rounds out the operating side.
That academic core is now paired with semiconductor-industry operators, including process-development leaders with long careers at companies like Intel. The board reflects the same blend of research credibility and industrial heft: Pat Gelsinger, the former Intel chief executive and now a general partner at Playground Global, joined as a director in 2026, alongside Blackbird co-founder Niki Scevak and semiconductor-manufacturing veteran Oreste Donzella.
Syenta's cap table mixes venture capital with sovereign industrial strategy - a signal that its work is seen as strategically, not just commercially, valuable.
The National Reconstruction Fund Corporation invested A$10.1 million, explicitly to help anchor a semiconductor-manufacturing capability in Australia - a country without a large domestic chip industry. That makes Syenta not only a startup but a test case for whether sovereign capability in advanced packaging can be built from a university lab.
Dr Jekaterina Viktorova and co-founders incorporate Syenta to commercialise Localized Electrochemical Manufacturing developed at the Australian National University.
Syenta demonstrates micron-scale interconnects with its single-step LEM process, drawing attention from the semiconductor sector.
The company raises A$8.8M to attack AI's memory bandwidth bottleneck, is named a SEMI S3 semi-finalist, and presents at 3D & Systems Summit, SPIE and IMAPS.
Playground Global and the NRF lead a ~A$37M round, Pat Gelsinger joins the board, and Syenta opens its first US facility at ASU Research Park in Tempe, Arizona.
The AI boom is usually told as a chip story - GPUs, HBM, ever-bigger accelerators. Syenta's pitch reframes it as a wiring story.
The compute already exists. The harder question is how to feed it. By positioning itself at the interconnect and advanced-packaging layer, Syenta sits alongside - rather than head-to-head with - the chip designers and foundries. Its technology is meant to slot into the supply chain that produces AI, high-performance and quantum systems, improving the density and bandwidth of the connections those systems rely on.
That positioning is deliberate. Advanced packaging has become one of the most contested frontiers in semiconductors precisely because transistor scaling alone no longer delivers the gains it once did. Whoever can pack more, faster connections into a package captures value across the whole stack. Syenta is wagering that a fundamentally different manufacturing method - electrochemical, single-step, lower-waste - can win a place in that frontier.
Whether it scales to volume by the end of the decade remains the open question. Hardware has no growth-hack shortcut; atoms take time. But with sovereign backing, a former Intel chief on the board, and a foothold in both Sydney and the Arizona semiconductor corridor, Syenta has bought itself a serious run at the problem.
Syenta develops advanced semiconductor packaging technology - specifically high-density, high-resolution chip-to-chip interconnects - using its proprietary Localized Electrochemical Manufacturing (LEM) process.
LEM combines metal deposition and patterning into a single electrochemical step using stamp electrodes, producing micron-scale interconnects with roughly 40% fewer process steps than conventional deposit-and-etch flows.
It was co-founded in 2022 by Dr Jekaterina Viktorova (CEO), Ben Wilkinson (CTO), Professor Luke Connal and Zachary Dowse. It is headquartered in Sydney and spun out of the Australian National University, with a US facility in Tempe, Arizona.
More than A$37 million in total, including a 2026 Series A led by Playground Global and Australia's National Reconstruction Fund, with former Intel CEO Pat Gelsinger joining the board.
It targets the "memory wall" - the growing gap between processor speed and the bandwidth of connections between chips - which increasingly limits AI and high-performance computing performance.
Share this profile, or dig into Syenta's own channels and the coverage below.
Profile compiled from public sources including Syenta, BusinessWire, Forbes Australia, Startup Daily, New Electronics, DCD, the NRF and City of Tempe. Figures are approximate where noted.