Automate construction. It’s a simple sentence. Short. What it conveys may be the single most important realm of progress to uplift humanity on a global scale.
Rising housing costs, increasing natural disasters, aging labor forces… AI won’t fix that in the cloud, only manufacturing can.
Nearly every industry has been completely transformed in the past 10 years, yet construction automation still occupies under one percent of the current industry in 2024. The obvious answer for this: it’s still in development. Every facet and aspect of the technology requires improvements before it reaches significant market penetration.
Sound overwhelming? Not for Icon. Yesterday I got the opportunity to attend their 2024 SXSW presentation. I don’t know what I was expecting, but they delivered. Hardware, software, materials; they demonstrated improvements on all fronts. This is exactly the type of leadership the industry needs, companies working tirelessly to improve every aspect of their systems. Digital marketplaces, AI design software, projects reaching new heights, and even teasers for future robotics… Let’s get into it.
ICON kicked off the event with a story about cost improvements. On their first project, it was over $300/sqft just for the wall segment which included outer and inner portions of the wall, reinforcement, insulation and finishings. They gradually decreased the price with the goal of beating the standard $35/sqft cost of stick built houses (for the wall section). The progress they’re making is incredible but I don’t think analyzing merely the wall segment captures the full story. Concrete is a very heavy material. This makes 3-D printed buildings extremely strong – able to withstand incredible wind loads. The weight also increases the foundation requirements, an added cost which I don’t believe was reflected in their estimate. There could be other discrete cost adjustments associated with printing vs. stick built construction outside the wall structure.
With their brand new system, they estimate costs could potentially be reduced to just $24 per square foot of wall segment. This difference may be enough to compensate for increased foundation requirements but until the project is completed, we won’t know. Construction estimates are notoriously unreliable.
Carbon X was announced at this event as well. Though sparse on details, they announced a public MIT Whitepaper, claiming the lifetime carbon emissions of a 3-D printed house or lesser than any comparable method.
The primary Takeaway from the study is that 3-D printed homes have between 2%- 6% lower lifetime carbon emissions than a stick built home. Reading the study I was surprised to see they gave both the same 75 year expected service life. I’m glad they did the math this way, though I hope a printed home would have more longevity than its stick build equivalent. Here is the recipe from the study, I will make some assumptions on what the ingredients could have been below.
- Type 1L cement: 15.56% wt, 517 lb/CY
- Supplementary cementitious materials: 5.69% wt, 189 lb/CY
- Mineral fillers: 9.87% wt, 328 lb/CY
- Aggregates: 58.01% wt, 1927 lb/CY
- Liquid admixtures: 0.84% wt, 28 lb/CY
- Mineral admixtures: 0.66% wt, 22 lb/CY
- Water: 9.36% wt, 311 lb/CY
SPECULATION DISCLAIMER
- Typically I do my best to report on things which are verifiably factual to preserve the longevity of my content as a historical reference, however today, I will breach my fact preferring ethos to speculate on what the CarbonX recipe might be. I am not a civil engineer, nor a material scientist, so you can consider this a layman guess.
First of all, it’s incredible to see only 15.56% portland cement, I have never seen this number so low in any 3DCP recipe anywhere. Putting our Sherlock Holmes hat on, let’s try to figure the rest out. Supplementary cementitious materials are likely certainly fly ash or slag based on the kg C02 eq/lb reported. In the MIT study these are listed as .008 kg C02 eq/lb but I do not subscribe to the calculations used to estimate fly ash or slag production because they are a byproduct of other carbon intensive processes which do not have the total production capacity necessary to transition the world to construction automation with these materials. If fly ash or slag were ever produced intently beyond its current capacity as a byproduct the CO2 impact would be astounding. Let it be noted that I pay no heed to any environmental calculations, I am not convinced protecting the environment is as simple as the management of one single molecule. True environmental protection would require an omniscience far beyond human capacity thus any ‘guidance’ from whatever powers that be are likely to be revised in the future as we learn more about the full picture.
It’s likely the mineral fillers are finely ground fly ash as well given the low kg C02 eq/lb reported and the qualifying statement in Table 1 of the MIT whitepaper regarding energy required for grinding the material. Silica fume is also a pretty good guess for this material as it also has a lower kg C02 eq/lb. Other common mineral fillers are limestone powder or quartz powder but both require too much energy to fit the bill of .0012 kg C02 eq/lb.
The aggregates are sand and perhaps small gravel likely all below 4mm perhaps as low as 2mm. Liquid admixtures are probably an accelerant often mixed with the material near the exit of the printer head or extruder to assist with buildability without limiting flow through the small diameter (2-4 inches) concrete hose. Mineral admixtures typically refers to fly ash, silica fume, slag, metakaolin or rice husk ash.
Next, we move on to CodeX. Think of it as Icon’s version of the App Store, but instead of digital apps, they sell 3-D printed houses. Currently, there are a plethora of options in well thought out categories like Texan, storm resistant fire, resistant tiny homes, and even some multi million dollar compounds. Architects and designers will be able to submit their designs and receive commissions each time their design is used. You can check out their existing models for yourself via their website in this link below, mind you the intro video is surprisingly long before the “start exploring” button appears, so be patient.
It’s unclear whether these models will ever be available for printing on non-icon printers. I expect they will continue with their Apple style ecosystem and keep everything within their walled garden.
This wasn’t the only software innovation icon had to announce that day,, Vitruvius.ai is their brand new AI powered design tool for both traditional buildings and 3-D printed houses alike. Although most of the verbiage on the website is limited to residential projects, I decided to challenge the system by asking for a 3D printed four unit apartment complex above a commercial retail operation with a restaurant and two stores and I was astounded to see it delivered not one, but three different renderings of what that 3-D printed building may look like. It’s my understanding the model was trained only on residential projects so this was a great surprise. CEO Jason Ballard noted they had the option to turn creativity for the AI system off, and improve the floor plan results in relation to the printed buildings, however eliminating creativity yielded far less exciting results, so they left some in.
Some of the challenges are obvious based on the data being limited to existing residential homes. It’s incredibly rare to see round angles on the house yet all the best 3-D printed building designs implement this feature. Likely because of this, the floorplans are hard, right angles or 45° angles never rounded. Another challenge was with the commercial building, obviously outside the scope described for this AI software. In the current stage, the first floor commercial operation, though accurate in the renderings, showed up as a typical residential building with kitchen, bedrooms, and bathrooms on the floor plan. The trajectory for AI is blistering fast. I’m sure with the right data sets, this software will improve very quickly. With enough models for curved floor plans in CodeX, Vitruvius.AI will get the data it needs to fix this problem. It was funny to see Icon announce CodeX and foreshadow its obsolescence all in one presentation.
Finally, we get to the Phoenix printer addressing many of the pain points in icons gantry style printer the Vulcan. This was the most exciting part of the presentation. You all know I’m hardware oriented, and nothing is more exciting than the robots that build. Gantry systems significant set up time, downtime, calibration challenges, and their size makes them difficult to transport, not to mention building at height becomes a wobbly experience. Icon has announced the Phoenix printer; a boom system on tracks, with a self leveling system on the printhead. lasers to identify its position and compensate for any sway in the boom. This new system has a much faster set up and take down time and with self driving, it would be able to navigate itself from one job site to the next. With a total height of 70 feet, Icon claims that it should be capable of printing two-story, perhaps even three-story structures. My favorite feature on this new system is an automatic reinforcement placing apparatus, though not yet demonstrated.
That wasn’t the end of icon hardware announcements, though the last one is incredibly sparse on the details. There was a brief mention of task bot systems developed for special use cases. We could speculate this could be flooring, painting, lighting, or scanning. At the most advanced level, someday, even plumbing or electrical work may be taskbot applications.
The event was also used to award Initiative 99 winners from around the globe, in two categories, student and professional. About a quarter of the million dollar prize purse has been awarded to the six winners, with the rest reserved for future phases of this project.
Make sure to sign up for the newsletter below in order to stay updated on the latest automate construction news. The new bill banning TikTok grants the President power to shut down any website which does not comply with their expectations, but even if every platform I operate on gets shut down, I’ll still be able to continue pursuing the mission to automate construction with you if you choose to share your email with me.