On September 8th we celebrated our 25th anniversary here at Computer-Aided Products. In my most recent post, I spent some time thinking about the forces shaped the company into what it is today. This is a continuation of the journey CAPINC has been on.
So, what shaped the company from the early days to now? There were two especially important factors. The first was almost getting fired! The second was my Army experience.
My first job was in the R&D group of an automotive OEM plastic parts producer. This gave me an excellent grounding in plastic product design and development. I moved on to an office machine manufacturer. Over time I was put in charge of the development of many of the plastic parts for an office machine, including working with the Industrial Designer, molders, moldmakers, etc.
With a major trade show coming up, we commissioned an industrial model shop to make a show model of the exterior package of the new machine. Prints were late from the Industrial Designer, changes were required by engineering, and the shop was heavily booked. As the critical trade show date approached, we still had no model! It finally got to the point where the R&D Director told me very pointedly that if the model was not ready in time for the show, I would be fired.
The shop did come through and built a very nice model, albeit with only a couple of anxious hours to spare before the deadline. While I escaped being fired, this was what we jokingly called in the Army a “Non-Career Enhancing Event.” Along with many others I was laid off a few months later.
This experience left a very strong impression on me. I resolved that if I ever worked for a supplier, I would make sure that the people who hired us got promoted. It is important to think about how suppliers can help their customers become more competitive all day, every day.
My Army experience and training was valuable in too many ways to count. But compared to purely civilian-trained managers, probably the most important skill I learned was the difference between “managing” and “micro-managing.” If you have an entire Cavalry platoon that is in in contact only by radio, you must not only have confidence in their training and equipment, but must have given clear orders on what mission is to be accomplished.
Many businesses have lengthy “Mission Statements.” I’m in the habit of asking people at these companies, “Do you know what your company’s mission is?” Very, very few folks have ever been able to answer the question.
So at CAPINC, I made the mission statement so simple that everyone can remember it, always.
“Improve Engineering Productivity.”
On September 8th we celebrated our 25th anniversary here at Computer-Aided Products. I spent some time thinking about what forces shaped the company into what it is today.
After working for four years as an engineer and four years in the Army, starting in 1985 I spent four years at an industrial model shop. At that time most products were still designed in 2D – not always with CAD – and prototypes were machined and fabricated largely by hand. We made some halting steps to modernize, including, eventually, a three axis milling machine and CAM software.
One day in 1988 I got a call from the model shop manager at Polaroid. “Have you heard of a new process?” he asked, “Where they shine a laser on a liquid to make models?” I had not.
In those pre-internet days information about new technology was not easy to come by. Eventually I found an article in the library about this new technology called “Stereolithography.” It so happened that the company had a sales office in Nashua, NH so we had a demonstration.
While novel, the early machines were expensive, slow, and had very limited material properties. But the biggest barrier to introduction was the lack of 3D solid models, or 3D models of any sort. As a job shop, we would be faced with having to create models, and it wasn’t clear customers would pay for them.
But automation of the entire product development process seemed to have a lot of potential. This led to a six month stint as a consultant for a company developing a competing rapid prototyping technology. But when they ran out of funding, there wasn’t enough business in this infant field to sustain a living.
I started surveying people I knew in industry about other technology opportunities. One of them, Rick Quinn, suggested I find a PC-based 3D CAD system. At the time most companies were using 2D CAD, and most were on UNIX workstations.
More surveys of customers gave this idea momentum, and I knew just the system to pick. Although AutoCAD had huge market share in the 2D PC CAD market, they didn’t have 3D. But a year or two earlier I’d purchased my own license of CADKEY to learn 3D modeling.
Manchester, CT-based CADKEY had developed a capable 3D wireframe modeler that ran well on PC’s. In the days of DOS memory limitations this was no trivial achievement. And it so happened that one of the local dealers was leaving the business. The area sales manager was happy to sign me up, even though there were already six dealers in the area and little marketing support: technology providers are inclined to believe that more dealers will automatically bring more sales.
Like most small business startups, the company was “bootstrapped.” I began operations in a spare bedroom with $7,000 in cashed out educational benefits, and a pension fund. This meant I could afford the fastest PC around: a 16mhz 80386 system with add-on math coprocessor. Graphics were provided by a Metheus 1004 add-on graphics card with amazing 1024x768 screen resolution and 16 colors! Final addition was a mouse, which also required an add-on card. My display was a 19” CRT that weighed 56 lbs!
With a background in manufacturing, job shop business and molding, my first customers were often mold or model shops and several of these customers (Tech NH and Mack Prototype) are still customers today!
How did the name “Computer-Aided Products” come about? The dealer I purchased CADKEY from myself had a very unwieldy name, and one that did not lend itself to abbreviation. And I expected that as the technology marched on something like “Dana’s CAD” wouldn’t accurately portray our product offerings. So Computer-Aided Products lent itself to “CAP.” We would incorporate nine months later.
Read CAPINC's 25th Anniversary - Part 2
One year ago I wrote about the potential for an emerging “Prosumer” market for 3D printers. The term can apply to these machines in at least three ways: as Toffler’s “Proactive Consumer,” as a “Professional Consumer,” or as a “Producing Consumer.”
The Replicator Z18 is priced at $6,500. Considering these machines will likely depreciate as fast as computer equipment, this is well above the budget for any but the most well-heeled or committed consumer. We added MakerBot after their acquisition by Stratasys and have all their models. We’ve been testing them for business use. The results of those tests will be reported later, so for now let’s focus on what the Z18 is and what it offers.
The Replicator Z18 has an immense workspace in comparison to most Maker-grade 3D printers. The build envelope is 12” x 12” x 18” (305mm x 305mm x 457mm). Consider that the Stratasys Fortus 250, a best-seller in the commercial market, has a workspace envelope of 10” x 10” x 12”.
Large workspaces have been offered previously in maker-grade printers. But they did not have heated build chambers, so builds that actually used the space suffered from severe warping and delamination. The patent for heated build chambers is owned by Stratasys so their acquisition cleared the way for this feature.
The Z18 has one disappointment for commercial users: it is PLA (Polyactic Acid, used by most maker machines) only, with only one extruder so a soluble support material cannot be run. PLA is a useful material for lots of “looks-like” models and limited functional testing. The lower heat deflection temperature is a concern and you won’t want to be leaving your PLA models in a closed car in the summer, and the material is usually more brittle than ABS. On the flip side, it smells better, it’s considered more earth-friendly, and can be pigmented. It can be glued, though not as readily as ABS.
For commercial users the lack of a soluble support material is a bigger obstacle. Many injection molded, cast, and even machined parts have internal passages that require support material and will be extremely difficult to clean by hand. This hand labor is costly and can result in damage to the model.
Maker forums are full of conspiracy theories about how Stratasys may have limited the functionality of the system to prevent competition with the more expensive Stratasys-brand printers. The strongest evidence against this is the near-total disconnect between the needs and expectations of consumers and commercial users. The Z18 reflects what makers want on a grand scale. And some commercial users (artists and industrial designers, for example) may also find the system meets their needs. But for commercial users, who need production-level accuracy and strength with minimal hand labor, the system has a more limited role.
Expect our detailed test results in September.
Why do patents exist? Patents give inventors a temporary monopoly on an idea. In a free market system, monopolies are considered a bad thing. But patents are granted to encourage innovation: if an inventor could not reap any reward from new ideas, who would pay for research?
But it’s not enough for an idea to be new to you to be patentable. There is some body of “prior art” in existence to solve a particular problem. The new idea must not only be new, but “non-obvious” to “a person skilled in the art.”
The RepRap project was based almost entirely on using the expiring patent(s) for the FDM process(es) of additive manufacturing. So did the RepRap project “rip off” Stratasys? No: patent protection expires in 17-20 years and the idea(s) can be exploited by all comers.
A recent blog article called “Has MakerBot Become TakerBot?” expressed the idea that MakerBot (now a division of Stratasys) was taking the ideas of customers and attempting to patent them. In investigating the matter, I found the root of the issue to be a near complete lack of understanding of how patents work.
There are three misconceptions that seem to have led to the conclusion. The first is that patent applications document the state of art in order to explain the novel claim: this explanation is just that, not part of the claim. The second is that patent claims can be (and often are) very narrow. Finally, the length of time from which new ideas are documented until a patent application is filed is often months and sometimes, years.
The blogger commented: “OpenBeam, maker of delta-style 3D printers, has been hard at work developing an automatic bed levelling process for some time. The OpenBeam Kossel Pro machine uses a probe attached to the extruder head similar to that described in the MakerBot patent.”
The operative word here is “similar”. This article once again demonstrates the near-complete lack of understanding of many of those in the maker community with industrial equipment. We used to have an FDM2000 (introduced in 1997) in the office but ran out of space. So here is a picture of the head of a Stratasys uPrint (introduced 2009). This head resembles the head of Stratasys machines going back five to ten years.
The yellow arrow points to a probe that is indexed up and down by the toggle bar behind it. When a print sequence is initiated, the head indexes to a number of X,Y positions on the build platform and the Z drive is engaged until the probe is contacted. The machine uses this data to insure the build platform is present, flat, and level.
So on the face of it, this long-existing Stratasys technology “uses a probe attached to the extruder head” to perform this self-leveling routine. Where’s the outrage? The idea of using a probe attached to the head is old news – 10 to 20 years old. That doesn’t mean there are not new, patentable ways to accomplish this – because, luckily for the folks at OpenBeam, the claims must be quite narrow.
Finally, there are individuals claiming that ideas or concepts they posted on Thingiverse, MakerBot forums, or elsewhere, were claimed by MakerBot as their own two or three weeks after the information was posted.
No one who has ever walked around with a patent notebook (with each page dated and signed, reviewed weekly by supervisors) would seriously consider that a corporation would be able to copy an idea to make their own in a few weeks. Those patent applications probably reference work that was conceptualized as much as two years ago.
Last year, Stratasys sued Afinia for infringing on four patents. This was notable as it was the first time Stratasys sued anyone over intellectual property. But the maker community is going to have to get real. Cloning ideas for fun is OK; cloning ideas for profit, they’d better be yours.
The purchase of MakerBot by Stratasys for over $400 million in stock (and up to $200 million in performance bonuses by the end of 2014) demonstrates the potential and interest in 3D printing. To put the deal in perspective, MakerBot has sold about 40,000 machines with an average sale price below $2,000. But the deal was valued at roughly $10,000 per system sold.
The Maker Movement really resulted from three factors: the wide availability of sophisticated but obsolete electronic and electromechanical devices; the Open Source / Free Software movement; and the development of publicly-available flexible manufacturing spaces called "Fab labs" first developed at the MIT Media Lab.
The expiration of the earliest Stratasys patents on FDM led to the creation of the RepRap project, which utilized open-source hardware and software tools to make an open-source version of 3D printing available. Soon a cottage industry developed for extruders, frames, motion control, filament, and front-end software. As demand for these products grew, an entire infrastructure sprang up.
Given the wide range of electronic, fabrication, and software expertise required to build a working RepRap printer, it wasn't long before demand for kits developed, starting as plans and proceeding to ready-to-assemble kits of parts and instructions.
MakerBot started as one of a number of small kit manufacturers with roots in the RepRap project. After developing several kits, they merged with another company - and absorbed $10 million in venture capital. This led to the release of their first fully-assembled system, the first of the Replicator series, and the first cries of "sell out" by none other than MakerBot co-founder Zach Hoeken, who called for the system to remain true to Open Hardware roots.
But Bre Pettis, one of the remaining founders, disagreed, noting in his blog: "cloning is not cool". The Replicator was released with some architecture "closed" while other parts of the machine were the same as previous open hardware versions.
Pettis highlighted one of the realities of present-day manufacturing: without Intellectual Property protections, new ideas would simply be cloned for profit by others. The innovator may be denied any compensation for their efforts. For casual or part-time hobbyists who might benefit from low-cost clones, this might be a good thing, but for innovators, a strategy to package technology in a way that increased barriers to copying is the only way they might be able to recoup their investment in time and resources.
Ironically, the acquisition of MakerBot by Stratasys will make their technology more useful while less "open." The soon-to-ship $6,500 Z18 has a heated build chamber and new extruder technology. Stratasys has a patent on heated build chambers so users can expect higher quality, more reliable models. While still a single extruder, PLA only system, the Z18 may be the first legitimate entry in the Prosumer 3D printing marketplace.
How we chose our latest service vehicle
With SolidWorks and Stratasys service staff and consultants traveling all through New England, CAPINC has a small fleet of sedans and light trucks to help minimize our transportation costs. The newest addition to this fleet is a 2014 Chevrolet Cruze Diesel.
The Cruze is a mid-sized sedan built in Lordstown Ohio and other plants around the world. The 2.0 liter turbodiesel is supplied by GM Germany, while the six speed automatic is made by Aisin in Japan. As a result, while the car is “American” the American content is 50% plus 16% Mexican content. The diesel comes with a high level of standard equipment: the price difference compared to a similarly equipped gas model is between $2,000 and $3,000.
Especially in Northern New England, most of our driving is on highways and there diesels really shine: the EPA highway rating is 46mpg. The diesel model has some of the aerodynamic features of the Eco model, with under-car panels to smooth airflow, a small rear lip spoiler, and active grill shutters. These shutters remain closed unless additional cooling air is needed. Reducing airflow into the engine compartment helps reduce drag.
Having been a Cavalry platoon leader with nine diesel combat vehicles, I had a few concerns about diesels: noise, cold starts, and sluggish performance. Chevrolet has addressed these issues effectively. The typical “diesel clatter” is noticeable at idle from outside the car, but around town and on the highway it is muted. GM upgraded the engine with ceramic glow plugs and even at temperatures below zero, the glow plugs only need a few seconds to start the car. While the horsepower rating is modest (150hp) this engine has torque of 250 ft lbs and a special “torque boost” mode that will allow up to 280 ft lbs for 10 second periods. The car accelerates from 0-60mph in 8.6 seconds and easily keeps up with traffic at all highway speeds.
The best-selling diesel passenger car in the US is the Volkswagen Jetta TDI, which also features a 2.0 liter turbodiesel. Dimensionally the cars are nearly the same length and width, while the Cruze is a few inches lower. The Cruze is also a little heavier but has a larger fuel tank (15.6 gallons). We buy U.S. assembled cars and while the Passat TDI is now built in a factory in Tennessee, the Jetta Diesel is built outside the US and was thus not considered.
Now that we’ve lived with the Cruze TD for a few weeks, we’ve formed some strong impressions. First, the interior is excellent and the front seats extremely comfortable. Second, the factory navigation system and “Infotainment” system is easy to use and very well integrated relative to some other factory installations. (We purchased that option as some states require “hands free” cell phone use and we want to minimize distraction). Mileage is living up to expectations, with a range of 31 to 51 mpg in city, commuting, and highway use. Ride is well controlled without being harsh.
We normally keep company cars for the five year depreciation period, during which we rack up 120,000 to 180,000 miles. Along with the operating economy and fuel range, we expect the resale value of this car to be higher than the four cylinder gas version.
Some of the upgrades we’ve added so far:
General Altimax Arctic winter tires
We equip all our company cars with winter tires because of the immense gains in stopping and handling over all season tires. We normally purchase Nokian (and prefer the new Hakkapeliitta R2) or the Michelin X-Ice but these tires were sold out. The Altimax Arctic is the Gislaved Nord Frost 3 – a well regarded performer at very competitive cost.
WeatherTech Floor Liners.
Most OEM rubber mats have flat sides, dumping the pool of ice, snow and salt that melted off your boots onto the edges of the carpet. Weathertech molds liners with substantial edges that contain the mess and protect the car. They now use digitizers to scan each car model to provide a custom fit.
Exploiting the fame generated from design of a 3D printed plastic gun, a DMLS (Direct Metal Laser Sintering) service bureau has now produced a metal 3D printed gun. Solid Concepts, a California-based service bureau, used the EOS process to build the parts. The press release indicates the gun was built in Texas, likely due to California's complex-if-ineffective gun laws.
DMLS uses layers of metal powder. A relatively high-powered laser is focused on a region, fusing it. A recoater then applies another layer of powder and the process is repeated. The primary differences with plastic layers processes lies in the power required and the thin layers (about 20 microns, less than one one thousandth of an inch), and the very high cost of the systems. But even very tough-to-machine alloys and materials like Cobalt steel and Titanium can be used, so the primary applications for this process thus far have been for medical and aerospace products with complex geometry.
The gun produced was a classic M1911 pistol, designed by John Browning and the U.S. military's official sidearm for nearly 80 years. According to the company, over 30 parts (of a total of 50+) were made using DMLS, including the rifled barrel. Unlike Cody Wilson's Defense Distributed, Solid Concepts has a Federal Firearms License for manufacturing, and did not release any designs.
So will 3D printed metal guns catch on? The economics don't favor it anytime soon. Using a $600,000+ industrial system to produce a part you can buy for $50 seems like a "get rich slow" program to me. If you are in the market, I'd suggest purchasing one instead from Connecticut-based Colt Manufacturing, New Hampshire based Sig Sauer, or Massachusetts-based Smith & Wesson. Or a Veteran such as this M1911A1 produced by the Remington Rand typewriter company in WWII.
Although we’ve been selling and supporting 3D printers for nearly 10 years and now have many customers with multiple systems, a lot of our new business comes from businesses who are buying their first 3D printer. For these new buyers, the intertwined risks of investment and technology can be paralyzing. At the front of their minds is often the idea that “We don’t want to buy something that’s obsolete.”
Well – relax! Because any 3D printer you buy is already obsolete. So make sure the investment makes business sense anyway.
Last week I purchased an Apple iPhone 5. I’m well aware that Apple is introducing the 5S and 5C models in mid-September. And there is no question that the 5S is “better” than the 5. But the new features are not something I use all day, every day. The cost of the new phone will likely be higher, it wasn’t available for at least a month, and there is some risk of teething problems attendant in the introduction of any new, complex product. For me, the costs of missing customer calls on my aging Android phone, the increased utility of some of the apps shared with an iPad, and the proven reliability of the product made it the right business decision – even though it is “obsolete.”
The consumer products industry in general and the fashion industry in particular are built upon a strategy of planned obsolescence: the time frame for becoming obsolete is built into the product. More complex and expensive products like automobiles use a strategy of “Continuous detail improvement” so that, while the old product remains usable, new product features are regularly introduced so that the new model becomes increasingly attractive. The vehicle we purchased in 2006 remains perfectly useful – even though it does not have any means to hook up an MP3 player and relies on radio or CD’s. The 2009 model has an MP3 connection and can also play DVD’s – but no Bluetooth integration like the latest models. So all three models perform the basic task of transportation, and newer models probably last longer, but those without the latest features are less appealing to consumers.
Far more likely in the 3D printer industry is what might be waggishly called “Unplanned obsolescence.” Innovations in the industry have disrupted established markets a number of times. In 2004 Stratasys introduced the Dimension systems at a price point of $30,000, supported by local dealers. This was about 40% of the price for similar capability the year prior. Over 300 systems were sold in the first year at a time when most manufacturers sold dozens of systems a year.
Rapid obsolescence comes with a significant danger to system manufacturers: prospective customers will delay purchase. So system manufacturers have adopted more flexible architectures that accommodate new materials or electronics, and/or implemented extremely generous trade-in policies. Customers can generate sufficiently high returns on investment to justify initial purchase, without being penalized for early adoption.
The explosive growth of the low-cost “maker-grade” systems has called into question the value of professional systems, which now comprise the market space from $10,000 up. But this ignores the very real difference in capability between even the very best maker-grade systems and the least-expensive professional systems. Maker systems thus far have supported hobbyists interested in exploring a new technology but without any business objective to support.
By contrast, professional systems support paid employees developing products where time-to-market is king. The systems are expected to run uninterrupted for days to support rapid product development.
In our experience, prospective customers tend to assume that price points will decline over the long term. While this may be true, in the short run, system manufacturers generally use a different pricing strategy. They establish price points ($10,000, $25,000, $50,000, and so on) that recognize the various levels of authority to “sign-off” such an investment within corporate design environments, and continue to add features over time to keep the offering competitive. So five years from now, a $25,000 printer may be faster and more capable – but there will still be offerings at that price point.
One frequent topic of conversation as we plan for the future is “How do we make our customers most productive?” Two factors that significantly impact these planning discussions:
1) Measuring CAD user productivity is difficult!
2) All CAD users consider themselves "Experts"
At SolidWorks World each year, there is a Model Mania competition where users are timed on a standard task, including changes and sometimes an FEA run. CAPINC’s Engineering Manager Jason Pancoast is a two-time worldwide winner and runner-up a third year; we believe a VAR must be able to use the tools to their fullest potential to be able to teach them.
The breadth and depth of tools in SolidWorks can be overwhelming to new users. There are comprehensive tools for solid modeling, surface modeling, sheet metal design, drafting, assembly, and data import/export. Scratch the surface only a little and you add assembly animation with interference checking; tolerance analysis, rendering, and costing. Dig a little deeper and you add FEA (Finite Element Analysis) and PDM (Product Data Management).
SolidWorks includes online help and tutorials, and there are number of tools that augment these offerings, from for-fee testing and video to free YouTube videos, dealer telephone support, and programs like CAPUniversity. So with the financial commitment, time away from project work, and general nature of classroom training, it should be dead in the 21st century – right?
Not So Fast!
There are a number of factors that are making product development professionals reconsider more formalized training regimens:
1) Engineering staffs are at an all-time low, everyone is expected to do more
2) Product development lead times are never fast enough
3) Quality requirements are greater than ever
4) Almost everyone faces worldwide competition
Tutorials help users understand the tools. They seldom address what might be called “Design Strategy”. Designing a project from concept through production requires consistent practices throughout. There is one area where classroom training delivers real benefits.
The primary benefit of classroom training, though, seems to be the old adage that “You don’t know what you don’t know.” The classroom training environment remains unmatched in that questions and comments by others prompt additional insights into how to use the tools.
If someone commented “Well, Dana, you sell classroom training – you’re not exactly unbiased.” And that is fair commentary: we recently spent $100,000 on new Hewlett Packard workstations to equip four training centers, so we have a significant investment in that environment.
What we see is a real interest in making users more productive, for all of the reasons listed above. One trend we’ve noticed is that training is often a leading indicator of the economy: if business is on the upswing, companies spend more on training, while if the economy is slowing down, we see a decline. Based on that theory, we can expect continued slow growth in the economy into 2014.
The most significant trend we’ve seen lately, though, is an uptick in customized training. Increasingly, mid-sized and larger groups are signing up for intensive sessions tailored to their specific design environment. So based on these customer preferences, at least for the next few years, classroom instruction will remain an important tool in optimizing your product development process.
Commercial 3D printers have been aimed at corporate and professional users since their invention in the late 80’s. Like any other semiconductor-based technology, price/performance has improved over the years. For a given build chamber size, prices are now approximately one quarter of what they were in 2004 but these systems are still out of reach for anyone but commercial users.
This changed in 2005 when Dr Adrian Bowyer started a project at the University of Bath to create a “Self Replicating Prototyper.” The project was called “RepRap” and used a process called “Fused Filament Fabrication” to avoid a trademark dispute over the use of FDM (Fused Deposition Modeling). FFF is a similar approach and relied on that fact that certain key Stratasys patents had expired. These machines were (and remain) open source systems that relied on crowd sourced hardware and software.
The RepRap project developed four different models that cost as little as $350 in kit form. There are now at least 25 different suppliers of kits or systems, including fully assembled models. Some have branched off into proprietary hardware or software in an effort to create intellectual property or differentiate themselves. They top out in price at around $3,500.
The Proactive 3D Printer Consumer
With the least expensive professional models costing $10,000 or more, here comes the “Prosumer market.” Prosumer is a term coined by Alan Toeffler in 1980 as a contraction of “Proactive” and “Consumer” to mean someone who actively participates in the design and improvement of a product they use. It has since taken on multiple meanings first as “Professional Consumer” (notably in the camcorder and digital camera market) or someone who desires or requires product features of professionals while using the product personally. The term has since been used in a third way: “Producing Consumers” or those who are actively customizing mass-produced goods for their own needs.
From a marketplace perspective, system marketers are trying to fill the gap between the $2,500 “maker” systems and the $10,000 and up commercial systems. One obvious candidate is MakerBot, recently acquired by Stratasys in a stock deal. Using Stratasys extruder, software, and build chamber technology, MakerBot could potentially provide a system in this price rang that would be sold directly by the manufacturer.
While most of the systems on the market are based on FDM technology, there are others. Form One is a Cambridge, MA based startup building a $3,500 system based on UV cured photopolymer. These systems have only recently started shipping so there isn’t much user feedback as yet. In addition, 3D Systems has filed a lawsuit against them alleging patent infringement. No matter the outcome, others will attempt to fill this market segment.
The Maturing Marketplace
Currently there are almost no systems available in the price range from $2,500 to $10,000. Don’t expect this vacuum to remain for very long. Whether these newly offered prosumer systems will be commercially successful is another question entirely. They may prove too expensive for consumers while proving too slow or having properties insufficient for commercial use. But there will be new such entries, without question.