Lumafield is CT scanning for the industrial manufacturing masses
While the annals of history are rightly littered with the names of the inventors of wondrous technologies and breakthrough products, the true value of these innovations is only fully realized when they become widely accessible. Electricity, computers, genetic sequencing and thousands of other major advancements have only had their extensive effect on society after they became naturally ubiquitous.
Products that at first seem to have utility only to a small set of customers suddenly connect with markets no one had ever thought about. It’s a pattern I’ve seen the last few months across at least satellites and space, biotech, web3, enterprise infrastructure, autonomous vehicles, and silicon. Once a capability exists, the number of applications always seem to expand as we learn more about what it offers and how it performs.
That’s true, but one of the critical factors which explains this expansion in applications is price. It’s not just that applications show up — they become enabled, both from lower costs as well as continuous improvements. It’s a story we’re going to focus on today in the context of multi-spectrum computed tomography (CT).
CT scans are popularly known from the medical world, where they are used to provide physicians with information on bones, blood vessels and other structures in the human body. A CT scan is composed of a series of X-ray photographs targeting the same location from different angles which are then stitched together using algorithms to create a three-dimensional representation of the object under study.
It’s a great example of intersecting hardware and software capabilities. Carefully calibrated X-ray machines have to be coupled with intricate computer vision software to orchestrate a high-resolution and readable image. As the capabilities of both hardware and software improved, CT scans consequently became richer and clearer, making them more useful.
While they may be most recognized from our experiences at hospitals, CT scans are also used in some industrial applications, like manufacturing. However, they are extremely expensive, and the software to manipulate them relies on decades-old user interfaces. As a consequence, the vast majority of industrial inspection is still conducted exclusively on the exterior of objects through metrology (think calipers to measure external dimensions), and more recently, with 3D scanning.
However, quality control is only obtained by knowing the insides of parts, where flaws may be introduced due to design, manufacturing processes, materials science, or repeated use. CT scans allow everyone from designers to quality assurance teams to peer into a product and evaluate whether it meets specifications. The result is more rapid iteration, better quality, and less rework, ultimately saving costs for consumers.
Industrial CT scanners can easily reach into the millions of dollars, and that’s before the specialized labor for operating this equipment is included. This isn’t a capability that a company can just grab off the shelf and place on the assembly line, but rather a long-term capital investment that requires extensive planning.
We have been interested in improving the access to this technology for some time here at Lux Capital. We’ve seen how developments in computer vision can transform industries like construction, mobility and biothereapeutics, and we’ve also seen how much manufacturing needs to be modernized. In addition, we also know that access to observable imaging data allows us to simulate in-silico, leading to significant reductions in future development cycles and product improvements. Capabilities are lacking here, which we saw as an opportunity for the right startup.
So when we met Eduardo Torrealba, the founder of Lumafield, we knew we had discovered someone who had a vision for what was possible at the nexus of product design, manufacturing and computer vision. He wanted to bring CT scanning to all companies, with easier-to-use hardware and software coupled with a financial model that allowed any company — big to small — to afford this capability in their processes. We co-led the company’s seed round in 2019 with Kleiner Perkins and our partner Bilal Zuberi joined the board. We also invested in its Series A round led by DCVC, and now, we’re ready to talk about what Lumafield is building as it emerges from stealth.
It’s already been making a splash: Scan of the Month, its surreptitiously branded site showing off what the combination of its Neptune scanner and Voyager software platform can do, has been an internet viral sensation. You can see all the innards of AirPods, Game Boys, and instant cameras to see what’s possible with the company’s technology. Lumafield already has several of the biggest brands in the world as customers using its system in production, with many more now on the wait list.
Lumafield innovated in three major directions to make CT scanning a routine part of manufacturing and product design. First, its hardware scanner is compact, packed with the best scanning tech, and easy to use. Second, it has coupled its scanner with a cloud collaborative software platform that will allow whole teams to work together with scans as part of their workflow — bringing scanning to the center of the manufacturing process rather than on the fringes.
Third and perhaps most importantly, Lumafield is innovating on the CT scanner business model and embedding financing into its price. Rather than forcing companies to outlay for expensive scanning hardware, the company is offering the hardware and software together for a price as low as $3,000 per month. It’s an affordable price point that should open up scanning to a whole new spectrum of businesses.
The result is that Lumafield goes far beyond industrial inspection to offer repeatable scanning so that designers are empowered to tinker and debug their ideas faster. Perhaps most importantly, this capability will encourage designers to experiment with new processes and materials in a way they might be hesitant to do today. One of the hindrances with moving toward global sustainability is the anchor of previous experience in manufacturing. Lumafield offers designers a chance to prove that better alternatives are working, and approach manufacturing from a much more sustainable light.
We’ve gone on record that we will no longer tolerate the decline of American manufacturing. We’re investing in companies that can transform how we manufacture so that the future is bright for industry bringing great products to all of us. That includes Lumafield and innovative manufacturing and tech companies like Desktop Metal, Shapeways, Hadrian, Veo Robotics, Formic, and several others.
Lumafield is a reminder that some technologies like CT scanning are not done innovating. Even decades after its invention back in 1967 by Godfrey Hounsfield, there is more work to be done to force costs lower; utilize software to access, manipulate, share, and analyze data; and drive overall improvement in product development and production. When it comes to the capability of CT scanning for industrial manufacturing, we’re just getting started.
Life and longing in Miami and New York City
Cory Van Lew’s “Delta, 2022” series on display at the Institute for Contemporary Art, Miami. Photo by Danny Crichton
Juxtapositions. That’s all I could think about when I saw news that a man, who government authorities have now identified and charged, had thrown smoke bombs and shot dozens of bullets at harried morning commuters at the 36th Street N/D/R subway platform in Sunset Park, Brooklyn.
I was sitting in Miami for our semi-annual offsite, enveloped in radiant and serene heat — humidity cooly low — watching the news eviscerate the headlines and cable news televisions. New York City had been cold, wet and dreary for weeks, and the prospects of a bit of soleil certainly brightened the domineering grey.
But as with so much about securities, the insecurities of our world are never all that far away. Outside the temporary paradise of Miami, the world’s troubles continued unabated. Millions of Chinese citizens remain hermetically sealed in their apartments to combat Covid, even as food rations run scarce, triggering potent memories of the worse episodes of the Middle Kingdom’s history. Russia continued its destruction of Ukraine, France once again brought a far-right leader to the second round of a presidential election, and Chile’s constitutional rewrite is potentially teetering into failure.
Constant vigilance is the tragedy of our world and the everlasting duty for all of us. Except in Miami. It’s an escape — but it’s not Mars. And it’s the center of the world, a world that’s burning and sinking and flooding. As Joel Stein profiled in my favorite piece so far this year:
The last time Miami was relevant, it wasn’t important. In the 1980s, Miami provided nothing more than drugs, clubs, pastel blazers, jai alai gambling and, most notably, a hit TV show about all four. But now Miami is the most important city in America. Not because Miami stopped being a frivolous, regulation-free, climate-doomed tax haven dominated by hot microcelebrities. It became the most important city in America because the country became a frivolous, regulation-free, climate-doomed tax haven dominated by hot microcelebrities.
It’s an apt description, and while the respite and offsite work was pleasant, it’s so great to be home again in New York City where reality exists, for better and for worse.
“Redlines” in war are meant to be objective and unambiguous tests for a country to respond to another nation’s action. Redlines though are often ambiguous, used poorly, and their deterrence effect is often diminished by a lack of credibility. How should leaders — from politicians to entrepreneurs — think about redlines? To answer that, I was joined by Josh Wolfe and General Tony “T2” Thomas, the 11th Commander of U.S. Special Operations Command (USSOCOM) and venture partner at Lux Capital, to talk all things redlines from our hotel lobby in sunny Miami.
Listen to the Episode on Anchor, or click through to links to Apple Podcasts and Spotify.
Lux Recommends
Our scientist-in-residence Sam Arbesman read Chuck Klosterman’s new book The Nineties. Klosterman, who is known for his extensive pop culture essays including the hit Sex, Drugs, and Cocoa Puffs and coverage of the music industry, dives into all the highlights of what is increasingly looking like a unique interregnum between the Cold War and the post-9/11 world. Klosterman delivers a tour-de-force retrospective of a decade that fewer and fewer have experienced — or even can imagine.
I have been interested in societal collapse and disasters for a while both here at Lux Capital and at TechCrunch. Adam Van Buskirk has an insightful critique in Palladium that “Collapse Won’t Reset Society.” Rather than changing everything, collapses triggered by pandemics, famines, or ravaging climates often strengthen existing governance rather than weaken it.
What are the lessons of Desert Storm? Christopher A. Lawrencepublished an exchange with DIA analyst William (Chip) Sayers about the planning leading up to the early-90s conflict, and what analysts got right — and how much they never knew.
Finally, Sam recommends Erik Hoel’s article “How to prevent the coming inhuman future.” Hoel writes, “So, in some ways, keeping humanity human should be as central a pillar to longtermism as minimizing existential risk (the chance of Earth being wiped out in the future), both because of the innate moral value of humanity qua humanity and also because for inhuman futures we cannot make moral judgements regardless.”
That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.
Forcing China’s AI researchers to strive for chip efficiency will ultimately shave America’s lead
In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.
In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.
Right now, pathbreaking AI foundation models follow an inverse Moore’s law (sometimes quipped “Eroom’s Law”). Each new generation is becoming more and more expensive to train as researchers exponentially increase the number of parameters used and overall model complexity. Sam Altman of OpenAI said that the cost of training GPT-4 was over $100 million, and some AI computational specialists believe that the first $1 billion model is currently or will shortly be developed.
As semiconductor chips rise in complexity, costs come down because transistors are packed more densely on silicon, cutting the cost per transistor during fabrication as well as lowering operational costs for energy and heat dissipation. That miracle of performance is the inverse with AI today. To increase the complexity (and therefore hopefully quality) of an AI model, researchers have attempted to pack in more and more parameters, each one of which demands more computation both for training and for usage. A 1 million parameter model can be trained for a few bucks and run on a $15 Raspberry Pi Zero 2 W, but Google’s PaLM with 540 billion parameters requires full-scale data centers to operate and is estimated to have cost millions of dollars to train.
Admittedly, simply having more parameters isn’t a magic recipe for better AI end performance. One recalls Steve Jobs’s marketing of the so-called “Megahertz Myth” to attempt to persuade the public that headline megahertz numbers weren't the right way to judge the performance of a personal computer. Performance in most fields is a complicated problem to judge, and just adding more inputs doesn't necessarily translate into a better output.
And indeed, there is an efficiency curve underway in AI outside of the leading-edge foundation models from OpenAI and Google. Researchers over the past two years have discovered better training techniques (as well as recipes to bundle these techniques together), developed best practices for spending on reinforcement learning from human feedback (RLHF), and curated better training data to improve model quality even while shaving parameter counts. Far from surpassing $1 billion, training new models that are equally performant might well cost only tens or hundreds of thousands of dollars.
This AI performance envelope between dollars invested and quality of model trained is a huge area of debate for the trajectory of the field (and was the most important theme to emanate from our AI Summit). And it’s absolutely vital to understand, since where the efficiency story ends up will determine the sustained market structure of the AI industry.
If foundation models cost billions of dollars to train, all the value and leverage of AI will accrue and centralize to the big tech companies like Microsoft (through OpenAI), Google and others who have the means and teams to lavish. But if the performance envelope reaches a significantly better dollar-to-quality ratio in the future, that means the whole field opens up to startups and novel experiments, while the leverage of the big tech companies would be much reduced.
The U.S. right now is parallelizing both approaches toward AI. Big tech is hurling billions of dollars on the field, while startups are exploring and developing more efficient models given their relatively meagre resources and limited access to Nvidia’s flagship chip, the H100. Talent — on balance — is heading as it typically does to big tech. Why work on efficiency when a big tech behemoth has money to burn on theoretical ideas emanating from university AI labs?
Without access to the highest-performance chips, China is limited in the work it can do on the cutting-edge frontiers of AI development. Without more chips (and in the future, the next generations of GPUs), it won’t have the competitive compute power to push the AI field to its limits like American companies. That leaves China with the only other path available, which is to follow the parallel course for improving AI through efficiency.
For those looking to prevent the decline of American economic power, this is an alarming development. Model efficiency is what will ultimately allow foundation models to be preloaded onto our devices and open up the consumer market to cheap and rapid AI interactions. Whoever builds an advantage in model efficiency will open up a range of applications that remain impractical or too expensive for the most complex AI models.
Given U.S. export controls, China is now (by assumption, and yes, it’s a big assumption) putting its entire weight behind building the AI models it can, which are focused on efficiency. Which means that its resources are arrayed for building the platforms to capture end-user applications — the exact opposite goal of American policymakers. It’s a classic result: restricting access to technology forces engineers to be more creative in building their products, the exact intensified creativity that typically leads to the next great startup or scientific breakthrough.
If America was serious about slowing the growth of China’s still-nascent semiconductor market, it really should have taken a page from the Chinese industrial policy handbook and just dumped chips on the market, just as China has done for years from solar panel manufacturing to electronics. Cheaper chips, faster chips, chips so competitive that no domestic manufacturer — even under Beijing direction — could have effectively competed. Instead we are attempting to decouple from the second largest chips market in the world, turning a competitive field where America is the clear leader into a bountiful green field of opportunity for domestic national champions to usurp market share and profits.
There were of course other goals outside of economic growth for restricting China’s access to chips. America is deeply concerned about the country’s AI integration into its military, and it wants to slow the evolution of its autonomous weaponry and intelligence gathering. Export controls do that, but they are likely to come at an extremely exorbitant long-term cost: the loss of leadership in the most important technological development so far this decade. It’s not a trade off I would have built trade policy on.
The life and death of air conditioning
Photo by Ashkan Forouzani on Unsplash
Across six years of working at TechCrunch, no article triggered an avalanche of readership or inbox vitriol quite like Air conditioning is one of the greatest inventions of the 20th Century. It’s also killing the 21st. It was an interview with Eric Dean Wilson, the author of After Cooling, about the complex feedback loops between global climate disruption and the increasing need for air conditioning to sustain life on Earth. The article was read by millions and millions of people, and hundreds of people wrote in with hot air about the importance of their cold air.
Demand for air conditioners is surging in markets where both incomes and temperatures are rising, populous places like India, China, Indonesia and the Philippines. By one estimate, the world will add 1 billion ACs before the end of the decade. The market is projected to before 2040. That’s good for measures of public health and economic productivity; it’s unquestionably bad for the climate, and a global agreement to phase out the most harmful coolants could keep the appliances out of reach of many of the people who need them most.
This is a classic feedback loop, where the increasing temperatures of the planet, particularly in South Asia, lead to increased demand for climate resilience tools like air conditioning and climate-adapted housing, leading to further climate change ad infinitum.
Josh Wolfe speaking at Nvidia Auditorium at Stanford. Photo by Chris Gates.
Josh Wolfe gave a talk at Stanford this week as part of the school’s long-running Entrepreneurial Thought Leaders series, talking all things Lux, defense tech and scientific innovation. The .
Lux Recommends
As Henry Kissinger turns 100, Grace Isford recommends “Henry Kissinger explains how to avoid world war three.” “In his view, the fate of humanity depends on whether America and China can get along. He believes the rapid progress of AI, in particular, leaves them only five-to-ten years to find a way.”
Our scientist-in-residence Sam Arbesman recommends Blindsight by Peter Watts, a first contact, hard science fiction novel that made quite a splash when it was published back in 2006.
Mohammed bin Rashid Al Maktoum, and just how far he has been willing to go to keep his daughter tranquilized and imprisoned. “When the yacht was located, off the Goa coast, Sheikh Mohammed spoke with the Indian Prime Minister, Narendra Modi, and agreed to extradite a Dubai-based arms dealer in exchange for his daughter’s capture. The Indian government deployed boats, helicopters, and a team of armed commandos to storm Nostromo and carry Latifa away.”
Sam recommends Ada Palmer’s article for Microsoft’s AI Anthology, “We are an information revolution species.” “If we pour a precious new elixir into a leaky cup and it leaks, we need to fix the cup, not fear the elixir.”
I love complex international security stories, and few areas are as complex or wild as the international trade in exotic animals. Tad Friend, who generally covers Silicon Valley for The New Yorker, has a great story about an NGO focused on infiltrating and exposing the networks that allow the trade to continue in “Earth League International Hunts the Hunters.” "At times, rhino horn has been worth more than gold—so South African rhinos are often killed with Czech-made rifles sold by Portuguese arms dealers to poachers from Mozambique, who send the horns by courier to Qatar or Vietnam, or have them bundled with elephant ivory in Maputo or Mombasa or Lagos or Luanda and delivered to China via Malaysia or Hong Kong.”
