“Down the Road” is Nearer than You Think: Why Nexa3D Is Bullish on AI for 3D Printing

February 21, 2024 by Matt Kremenetsky 3D Printing 3D Printing Events Automation Exclusive Interviews Quality Control

Share this Article

At Additive Manufacturing Strategies (AMS) in New York City (February 6-8, 2024), Nexa3D CEO Avi Reichental gave a presentation entitled, “How Artificial Intelligence Will Transform AM.” The day before he gave the talk, Reichental — sitting alongside Blake Teipel, Nexa3D’s Chief Strategy Officer and former CEO of Essentium — shared his thoughts with me on that same topic in an interview:

“My prediction is that by 2030, there are going to be two kinds of AM companies,” Reichental said. “Those who fully embraced and benefited from AI, and those who are out of business. And I say that about this industry in particular, simply because it’s the industry that I’m in, but I think you can say it about any industry.”

Reichental has never been one to shy away from publicly making bold predictions. That particular prediction, though, doesn’t feel that far off to me, and one reason for that has to do with another point Reichental made towards the end of our conversation:

“AI was conceived in a conference at Dartmouth College in 1956,” Reichental noted, referring to the Dartmouth Summer Research Project on Artificial Intelligence, an event viewed as marking the origins of AI as a field of research. “AI is my age. In the beginning, it was just a concept. I remember doing AI exercises in workshops 10 years ago with cards, which illustrates very simply how the concept works — but without ubiquitous connectivity, without infinite computing power in the cloud. When we have our discussions internally now, as an enterprise, it’s almost like, forget about additive, let’s talk about digital manufacturing.”

Much like AM, AI is a technological realm that feels brand new, but really isn’t. During a panel discussion at the end of the last day of AMS, in response to a question from an audience member, one of the panelists referenced Reichental’s talk on AI and said that in his opinion, the incorporation of AI into AM won’t have an impact until somewhere “down the road.” Even if that is the case, however, the companies that do make that impact down the road will have to have started planning the drive to their destination well in advance.

It is symbolic that Nexa3D released Nexa AI, its automated workflow software platform for the XiP Pro, at the same time as the company was announcing its acquisition of Essentium, right at the start of Formnext 2023. Both news items marked Nexa’s entry into the next phase of its existence—a phase that looks like it will be defined by the company’s adoption of precisely the sort of new capabilities that allow it to lean into its core strengths most effectively.

In addition to leaning into the automation of workflow management with AI, Nexa3D is also leaning into the speed of product development enabled by its technology through leveraging Essentium’s deep expertise in materials science. Teipel explained to me how he envisions AI playing a role in the formulation of materials:

“Essentium released two material formulations last year that I was particularly proud of our team for brining to market. One is a material called Duratem, where we basically took the chemistry of ULTEM, which is a polyetherimide, and we use siloxane treatments to modify the chemistry. Basically, we took the stiff parts of the polymer chain and we broke them up with some flexible bits that creates an ULTEM, aerospace-worthy material that is now ductile and has some resilience to it, but still survives the flame, smoke, and toxicity requirements for airworthy parts.

“But it took us like 18 months to develop that material. We went through a whole bunch of custom formulations and all this deep chemistry and material science R&D to make it happen. With the use of AI, we could’ve gotten there so much more quickly. And then we could’ve immediately started focusing on the implementation of the material to save time and money for aircraft parts.”

Teipel described a similar scenario in regards to another 2023 material release, Altitude:

“Altitude is a polycarbonate, so like the same material as hard hats and safety glasses, but it’s got a UV stabilizer, it’s got a thermal stabilizer, and it has anti-microbial properties—all of these packages that we put into the material to make it useful at very low temperatures and useful for applications like drones and UAVs. The goal was to make a material that’s both low-cost and easy to use. But again, the use of AI could’ve greatly accelerated our path to optimizing the ratios between all those different properties.”

The ways in which Reichental and Teipel are envisioning the incorporation of AI into the Nexa3D workflow may be ahead-of-the-curve, but they’re in no way far-fetched. Original equipment manufacturer (OEM) Inkbit has been using AI for in-situ 3D printing quality control for some time now, while MIT researchers recently released a study concerning the use of AI for material qualification in a manner very similar to what Teipel touched upon.

But beyond any specificities, these executives just demonstrated a strategic grasp of the nature of the opportunity that is presenting itself— with Reichental astutely framing AI’s potential the same way that Microsoft is—as a “copilot”:

“In my two decades in the industry, the slow adoption or the friction or the blockage point that stands between what we as OEMs build, and how much of it our customers can successfully onboard and implement, has to do with expert, labor-intensive work in the pre-processing phase,” said Reichental. “It’s a very steep learning curve. We know of hundreds of companies over time that wanted to adopt 3D printing, brought it in-house, and within six months or a year, they push it to the side. Why? Not because the printers didn’t work, but because they underestimated the level of complexity and expertise required to adopt it successfully at scale. AI takes most of those obstacles away just in the pre-processing phase.

“Then you begin to look at the printing phase itself, and with sensors, you can collect real-time data, process it, and make your systems more and more intelligent. We’ve been printing for decades, but the printers don’t know what they print. So if, all of a sudden, the printers gain that capability — if they can learn — yields improve. The ratio of labor intensity to parts printed decreases. The learning curve shrinks. All of that delivers tangible value. We’ve been talking about wanting to democratize access for years, but we really haven’t done much to remove the biggest barriers to adoption. AI can enable you to do that, which is why we think about it as a wingman.”

There was something especially compelling about that particular way of framing the stakes involved, insofar as the message was being delivered by two executives who have clearly come to see each other as wingmen. And their experience in successfully executing a complicated merger process together may in fact provide relevant data for continuing to incorporate AI into their operations: thinking long-term, learning on the fly, not being afraid of a challenge.

The optimization of AI’s incorporation into AM may indeed be a ways down the road, but that means no one is going to get there by standing still.