The current competition in the rocket launch industry illustrates what happens when a disruptor enters an industry and how the incumbent tries to adapt to the new competition. In the early 2000s, the American launch industry was controlled by two aerospace giants, Boeing and Lockheed Martin, which were locked in a bitter legal battle. The two companies eventually resolved their legal wrangles by forming a 50-50 joint venture in 2006 known as the United Launch Alliance (ULA). For the next decade, the ULA enjoyed a monopoly on national security launches by the Air Force and the National Reconnaissance Office (NRO).
A legal challenge to this new monopoly was filed by a tiny startup out of California known as Space Exploration Technologies (SpaceX), which had outlandish plans to colonize Mars. SpaceX hadn’t even managed to finish building its prototype Falcon 1 rocket, much less launch it, so it wasn’t even regarded as a viable competitor by the US government and its legal case was dismissed. Many in the aerospace industry at the time thought that SpaceX was a joke. Everyone knew that only large, established aerospace companies with deep pockets and long-established ties with the government could secure the necessary funding to develop a new rocket. It would take billions of dollars and have to outsource hundreds of parts from aerospace companies with decades of experience creating components that could handle the rigors of space.
Nobody would have guessed when Elon Musk founded SpaceX in early 2002, that a Silicon Valley entrepreneur who had no experience in the aerospace industry would turn the entire industry on its head within a decade and a half by breaking all the rules for how an American aerospace company was expected to operate. With a budget of under $100 million, SpaceX managed to design and launch four rocket prototypes until it finally got one successfully into orbit in September 2008. Twenty-one months later, which is an amazingly short period of time in the aerospace industry, it launched its new medium-class Falcon 9 rocket, which would take over the global rocket industry over the next decade. SpaceX managed to design and successfully launch its Falcon 9 rocket for a total cost of $300 million, which was roughly what the ULA charged for each of its rocket launches at the time. A 2011 NASA study calculated that developing the same rocket using traditional cost-plus contracting would have cost $3.977 billion.
SpaceX was able to achieve such low costs by hiring a bunch of engineers “with a proven history of building and breaking things,” and telling them to build and test in order to iterate rapidly in their designs. The whole company was organized with a very thin hierarchical structure, so that anyone could talk to anyone else in the company to accomplish their goals, and critical decisions could be made as quickly as possible. The engineers were also given a mandate to build anything themselves if they could do it cheaper than buying it from an outside source.
Elon Musk was deeply disturbed by how much rockets cost, because he saw it as a barrier to ever colonizing Mars. To prove that getting to space shouldn’t cost so much, SpaceX transformed the rocket industry by offering to launch its Falcon 9 rocket for a base price between $49.9 and $56 million in 2010. SpaceX’s launch prices were so much lower than any other company, that SpaceX reportedly had a backlog of 50 launches worth $4 billion in 2013. The company didn’t have the capacity to execute most of those launches, but it demonstrated how dramatically SpaceX undercut the competition.
Today, SpaceX is able to launch a rocket once every 2 weeks and it controls over 60% of the global commercial launch market, which is a startling reversal because the US had zero percent of this market in 2010 and the Russians controlled almost 60% of it. SpaceX now has 100 launches on its manifest worth $12 billion, meaning that new customers will have to wait 4 years in the queue before SpaceX can launch their satellites.
SpaceX is able to produce and launch its rockets so cheaply, that it has now undercut the Russians in offering the cheapest way to get satellites into space. Its cost to put a kilogram into low earth orbit (LEO) and geosynchronous transfer orbit (GTO) is lower than any other rocket in the world. NASA has traditionally paid an average of $23,750 to put a kilo into GTO, but SpaceX can do it for as low as $6,078 per kilo.
According to a recent FAA report, SpaceX has forced the entire rocket industry to drop its launch prices by 10% to 15%. As SpaceX has perfected the ability to propulsively land the first stage of its rockets, it can now reuse its first stages, which dramatically lowers its costs, since the first stage represents 60% of a rocket’s total cost. In contrast, its competitors have to throw away their first stages with each flight, which makes space flight very expensive.
SpaceX’s latest version of its Falcon 9 rocket, the Block 5, is designed so that its first stage can be used for 10 or more flights before needing refurbishment and SpaceX hopes that it will last over 100 flights before needing to be retired. This increased reusability has allowed SpaceX to lower the base price of its Falcon 9 rocket from $62.5 to $50 million and offer a new Falcon Heavy model which uses 3 cores for $90 million if reused, instead of $150 million if expended.
SpaceX also hopes to eventually be able to propulsively land its second stage, which represents 20% of Falcon 9’s costs. Many in the aerospace industry consider this to be a virtually impossible feat, considering the extra heat shielding and fuel that it will require, but SpaceX has already achieved a number of feats which were previously considered impossible. It has designed payload farings with thrusters and parafoils which will glide down so that SpaceX can theoretically catch them in a 3700 m2 net with shock absorbing arms on a ship in the ocean. It is now getting within a couple hundred meters of catching the fairings in its net. If SpaceX can figure out to reuse every part of its rockets and spread the cost of manufacturing a new rocket over 100 launches, then it calculates that it can lower its launch price down to just $5 million, which would revolutionize access to space.
Achieving rocket reusability has required thousands of engineering changes that would have taken decades to implement in a traditional rocket company, but SpaceX has a practice of continually improving its rocket designs. Within a decade of successfully launching its primitive Falcon 1 rocket in 2008, SpaceX’s rockets have evolved to become the most advanced in the world, with the most advanced avionics capable of propulsive landing and the highest thrust-to-weight ratio of any rocket engine.
To understand how extensive these changes have been, look at the dramatic changes which occurred between Block 4 and Block 5 of the Falcon 9. A new black thermal protection material was added to the Block 5’s interstage, raceways and landing legs, so that they won’t need to be repainted after every flight. The fuel valves have been made more durable. The painted aluminum grid fins that used to melt on reentry have been replaced with unpainted titanium that is designed to withstand 1000°C temperatures. The composite heat shield around the 9 engines on the base of the rocket has been replaced with a titanium one which has active water cooling at critical points. Plus, the octaweb, the structure holding the 9 engines, has been upgraded from a 2000 to a 7000 series of lithium-aluminum alloy which is stronger and it is now bolted together instead of welded, so it is easier to take off the engines for inspection and replacement. There is a new inertial measurement system, engine controllers and flight computer to make the avionics more fault tolerant. The composite overwrapped pressure vessels (COPV) made carbon fiber with an aluminum liner have been redesigned to be stronger. Finally, the Merlin 1D engines have been upgraded to provide 190,000 lbf of thrust at sea level, which is 8% more than the previous version.
No other rocket company before SpaceX managed to make these kind of changes so quickly to their designs. SpaceX’s founder, Elon Musk, created two dot.com startups (Zip2 and X.com) in the Silicon Valley before creating a rocket company and he brought the same kind of rapidly iterating engineering used by software programmers to rocket design. The goal was not to create the perfect design after hundreds of hours of meetings and long analysis, but rather try out new designs with lots of testing to see what works and then rapidly iterate on a design to perfect it.
SpaceX’s goal is to get to Mars as cheaply as possible, not try to justify ever greater costs through cost-plus contracts with the government and secure big contracts to build en masse before prototyping like traditional aerospace companies. It had the good fortune to have sympathetic administrators at NASA who shared its goals and funneled contracts to SpaceX to keep it afloat after it nearly went bankrupt in late 2008.
SpaceX couldn’t afford to pay the inflated prices for space-grade components made by the rest of the aerospace industry, so it generally either makes its own or adapts standard off-the-shelf parts. Rather than use radiation-hardened electronics, SpaceX uses 3 flight computers with standard processors for each of its engines and compares the results of each computer to make sure that radiation and the extreme stresses of space flight haven’t caused a failure in the calculations. This same idea of redundant fault tolerance guided their decision to use nine small engines on the Falcon 9, rather than a few powerful engines like on other rockets, because it allows an engine to fail and the other engines to compensate in order to complete a mission. SpaceX uses Linux and other standard software which is well tested and programs in C++, rather than in specialty languages like Ada used by most aerospace companies, because it is cheaper to find C++ programmers and easier to reuse and test its modules throughout their software.
By building its own components, SpaceX has been able to innovate in ways that ULA cannot. Dan Rasky, who helped invent the PICA heat shield at NASA, recounts how efficiently SpaceX operates. Rasky was asked by Musk in a meeting whether it would be better for SpaceX to manufacture PICA or hire an outside company. Rasky argued that SpaceX could make it more cheaply than an outside company. Used to operating at NASA, Rasky expected his recommendation to be confirmed by a follow-up study and a whole series of meetings. Instead, Musk immediately assigned a team to learn how to manufacture PICA. Not only did SpaceX’s team manage to do far more economically, but they managed create their own version called PICA-X, which significantly improved its properties. They also improved on their own version of NASA’s SIRCA heat shield, called XIRCA, which can be bent like a blanket to protect moving parts like the hinges in rocket fins, but it also can be used to protect electronics. The expertise that SpaceX gained in manufacturing their own heat shields gave SpaceX the confidence to redesign the BRF’s Starship to belly flop like a skydiver during reentry in order to reduce the velocity without expending fuel. Because SpaceX did its own R&D, it is confident that its PICA-X and XIRCA will be able to handle a large number of reentries before it needs to be replaced.
Many companies which do their own R&D get stuck in technological deadends by doggedly pursing the same solutions even when there are better alternatives. SpaceX appears to be able to avoid the deadend trap. It spent a huge amount of time learning how to make the BFR’s fuel tanks out of carbon fiber, and it reported some success in its efforts. Then, it unexpectedly announced that it would be switching to a 300 series stainless steel alloy that it calls SX500. Carbon fiber is stronger than steel at room temperatures, but SpaceX figured out how to cold-form this alloy, so that it is slightly stronger than carbon fiber at cryogenic temperatures and significantly stronger at high temperatures. Furthermore, SpaceX says that this alloy can be polished like a mirror to reduce the high-speed friction and generated heat during reentry. SpaceX is now setting up its own foundry in Hawthorne to manufacture SX500.
SpaceX has been criticized for making such radical changes in the design of the BFR when presenting its annual updates to the public in 2016, 2017 and 2018, but what these changes show is that the company is capable of quickly altering its designs as it learns more. Boeing’s SLS and Northrup Grumman’s OmegA show what happens when rocket companies get trapped in a traditional design.
Traditionally, rocket companies have not publicized their prices, so that their customers were at a disadvantage when contracting for their services. This secretive approach allowed rocket companies to virtually name their price in contract negotiations. SpaceX revolutionized the rocket industry by publicly stating in 2005 that the price of launching their Falcon 1 rocket would be $5.9 million, which was far lower than another other rocket company. This practice of publishing a price list eliminated favoritism and gave governments and telecommunications companies a means to plan their future costs, which in turn has made more space projects possible.
What is particularly interesting has been the response of the United Launch Alliance (ULA) to the disruption caused by SpaceX. Ever since its founding as a joint venture between Boeing and Lockheed Martin, ULA has enjoyed a monopoly on launches for the US government, and its high prices have reflected its lock on the market. Because its contracts with the government were so lucrative, it largely gave up competing in the commercial satellite launch market, which was served by Arianespace’s Ariane 5, International Launch Services’ Proton-M and Roscosmos’ Soyuz-2.
Global commercial launch market, source: Wikipedia.
ULA gave up competing in the commercial market, partly because its costs were much higher than Russia’s, but also because competing would have meant undercutting the monopoly prices that it was charging the US government. The US government wasn’t going to launch its national security payloads with a foreign launch company, so ULA had no reason to change its high prices and no incentive to innovation. The arrival of SpaceX, however, introduced for the first time a domestic competitor which could compete for its lucrative government contracts.
How ULA has responded is a textbook example of how incumbents react to disruption. The first step is to reassure jittery investors and the market that the incumbent has a plan to restructure its own business so that it can compete with the disruptor. In October 2014, ULA announced that the company would be restructuring to cut the cost of its launches in half. It cut its workforce and it has started hiring young college graduates and encouraging its older workers to retire, which has hurt morale at the company and reduced the expertise of its workforce.
Another part of ULA’s response to SpaceX has been to use its lobbying power and that of its two parent companies to convince the government to continue giving it launch contracts. Musk alleged in January 2015 that there was a revolving door between the Air Force and Boeing and Lockheed Martin and that Air Force officials who expected to one day work in these companies were slow walking SpaceX’s certification for national security launches, which allowed the ULA to maintain its monopoly. He pointed out on Twitter that one Air Force procurement official who negotiated lucrative rocket launch contracts with the ULA later became a vice-president at Aerojet RocketDyne, which supplies rocket engines to the ULA.
After a number of delays, the Air Force eventually certified SpaceX for national security launches in May 2015 and SpaceX won its first Air Force contract in April 2016 for an $82.7 million launch of a GPS satellite, where the offered amount was so low that ULA didn’t even bother competing. It has since won another 96.5 million launch contract from the Air Force, and was recently awarded a $130 million contract to launch a classified satellite on its new Falcon Heavy rocket.
SpaceX’s prices for these government launches are a fourth of what ULA has traditionally charged. In 2015, the ULA claimed in a congressional hearing that it charged an average price of $225 million per launch, but that average included a $6.6 billion backlog in old orders from the Air Force to launch 42 cores of its legacy rockets and it only included $1 billion for one EELV Launch Capability (ELC) contract, that the Air Force pays the ULA to maintain its readiness to launch multiple types of rockets. Most of those back orders were never launched so they shouldn’t be included in the average and the ELC costs have to be included every year, not just once. The Government Accountability Office reported in 2015 that the government pays $360 million per launch and over $350 million of that goes to ULA. In 2014, the GAO estimated that the ULA will receive $422 million per launch in 2020.
SpaceX’s Falcon 9 is not large enough to carry the same payloads as the ULA’s Delta IV-Heavy, but it was a direct competitor to ULA’s Atlas V. Once it became clear that SpaceX would not be excluded from bidding for Air Force contracts, ULA announced that it would be lowering the base launch price of their Altas V from $180 to $109 million, which was still considerably more than Falcon 9’s base price of $62 million at the time.
SpaceX charges $25-35 million over its base price for government contracts because government agencies have extra requirements compared to commercial customers. Even in fulfilling these requirements, ULA charges almost double what SpaceX charges. In 2018, SpaceX won a bid for 3 launches of GPS III satellites in the later half of 2020 for the Air Force for the price of $96.7 million per launch. Meanwhile, ULA won a bid at the same time for $175.5 million per launch. In other words, SpaceX will charge $34.7 million over its base price compare to the ULA charging $66.5 million over its base price for what should be roughly the same services. On top of that, the ULA is charging the Air Force $867.0 million in 2019 to maintain its launch readiness. Given that five Altas V and three Delta IV-Heavy launches are expected in 2019, that means an additional $108.4 million per launch.
Although SpaceX can now compete for new contracts from the Air Force and NRO, it was initially locked out from most national security launch contracts, because the Air Force signed a deal in December 2013 with ULA to block buy the launches of 35 rocket cores (23 Atlas V and 4 Delta IV-Heavy) between 2014 and 2019 for a total price of $9.5 billion. After adding in $4.41 billion for 5 annual ELC contracts to maintain its launch readiness, the ULA will have charged a total of $397.5 million per rocket core or $515.3 per million rocket launch. By signing this long-term block buy, the Air Force and NRO only planned to allow SpaceX to compete for 14 launches and this number was then cut down to 7 or 8 due to budget cuts.
The 5-year block buy with the ULA had been in the works for a long time. In 2009, the Commander of the Air Force Space Command Air Force and Director of the NRO determined that short-term contracting for launch services needed to be replaced by long-term contracting to launch 8 booster cores per year over a 5 year period, in order to “help stabilize” the launch industrial base, which they believed to be “unstable”. In March 2011, a program was setup to carry out the 5-year block buy under the Evolved Expendable Launch Vehicle (EELV), which was a program set up in 1995 to ensure that the government would have more affordable and reliable access to space
In response, the Government Accountability Office issued a highly critical report of this plan in September 2011, noting that it would lead to the government buying more launches than it needed and rockets that weren’t used would require storage and later updating when eventually used. It also noted the block buy was “based on contractor data and analyses in lieu of conducting independent analyses” and the contractor data needed to “negotiate fair and reasonable prices” was “lacking.”
The Air Force and NRO did take the GAO recommendations into account and investigate the costs of rocket components. They claimed that the EELV and the 5-year block buy saved American taxpayers $4.4 billion. That number seems to be based on subtracting the final price from the initial projection in 2011 that 40 rocket cores would cost $15 billion, but the government could have saved a lot more by not signing a long-term contract that locked SpaceX out of the majority of the market. If the goal was to save money as claimed, then the Air Force should have logically waited until it had two competitors in the market before signing any long-term contracts.
What likely happened is that the ULA foresaw that SpaceX would soon be certified for national security launches, so it lobbied the Air Force to give it a long-term contract to guarantee its high prices and lock out its new competitor for the next five years. In response, SpaceX filed a lawsuit against the Air force in April 2014 in order to be allowed to compete for the launches in the 5-year block buy. In January 2015, SpaceX and the Air Force worked out a secret settlement to the suit, that allowed all parties to save face. The Air Force agreed to hasten the certification process and make it easier for new competitors to get certified. Plus, it promised to offer more launches for competitive bidding, while not modifying its contracts with the ULA. In return, SpaceX stopped turning the public spotlight on the cozy relationship between Air Force officials and their military contractors. Several months after the settlement, SpaceX’s CEO, Gwen Shotwell, made very complementary remarks about the Air Force’s cooperation in the certification process when grilled during congressional hearings.
Currently, the ULA still wins new contracts because the government has decided that there needs to be at least two launch providers, and ULA simply can’t launch for the low prices of SpaceX, so it has been able to charge over double the price of SpaceX. It is hard to say how much of ULA’s price advantage is the result of its superior lobbying, but the government now maintains that it is necessary to have two competing launch providers, whereas it didn’t believe that to be the case in 2006 when it allowed Boeing and Lockheed Martin join together to create a monopoly in the national security launch market.
The problem for ULA is that its ability to charge higher prices will be coming to an end in the 2020s, once the market expands from 2 to 4 competitors. Northrup Grumman (which recently bought Orbital ATK) is now developing its OmegA rocket and Blue Origin is working on its New Glenn rocket. ULA is also working on its next generation rocket, the Vulcan. ULA still has the advantage of greater lobbying power, which is reflected in the fact that the Air Force recently announced that it would be giving ULA $967 million to develop the Vulcan, whereas it will only give Northrop Grumman $792 million and Blue Origin $500 million to develop their new rockets.
However, the Air Force plans to only select 2 companies in 2020 to receive future launch contracts between 2022 and 2025. One of those companies will surely be SpaceX, since its Falcon 9 Block 5 and Falcon Heavy are so far ahead of the rest of the field, especially when considering that SpaceX will have the most reliable and best tested technology of any of the next generation rockets, which are important factors in national security missions. If ULA isn’t selected over Northrup Grumman and Blue Origin in 2020, it could potentially be a death sentence for the company.
Company history and political influence suggest that ULA will beat Northrup Grumman and Blue Origin, which have never been through the process of getting certified for national security launches. However, Northrup Grumman announced the design details of its OmegA rocket in April 2018, which was 5 months before ULA announced its design for the Vulcan in September 2018 and decided that Blue Origin’s BE-4 methane engine would power its first stage. Blue Origin might seem like a long shot, but the founder of Amazon, Jeff Bezos, who is the richest man in the world, has been pouring roughly $1 billion per year into the company and its New Glenn rocket promises a reusable first stage with an advanced methane engine and propulsive landing and a hydrogen engine for the expendable second stage.
The one thing going in ULA’s favor is that the Air Force is scheduled to decide before any of the 3 next generation rockets are scheduled to have their maiden flights in 2021. In other words, ULA might be facing years of delay in deploying its new rocket, but it might not be obvious to an outside observer, whereas the first rocket to successfully launch is an easy metric to judge. The fact that ULA might go bankrupt as a company if it doesn’t get selected also helps ULA, because the Air Force knows that Jeff Bezos will continue to fund the New Glenn no matter what and Northrup Grumman has deep pockets to keep the OmegA alive. On the other hand, ULA probably won’t disappear since its parent companies will probably step in to save it, so it could limp along to compete in the next round of Air Force contracts in the mid-2020s even it is not selected in 2020.
Just as important is the fact that the Air Force wants the companies that it selects to also be competitive in the commercial launch market, because that will keep their prices low. Although the ULA talks a good game about cutting its costs, it frankly doesn’t have a rocket design that can compete, nor is it a company accustomed to not charging monopoly rents. Blue Origin’s New Glenn is a much better rocket to launch commercial satellites and take tourists to space than the Vulcan. OmegA is an even more outdated rocket design than the Vulcan in some ways, since it will have no reusability at all, but Orbital Sciences Corporation managed to function for decades on very few launches per year from NASA before it was absorbed into Orbital ATK and then bought by Northrup Grumman, so the OmegA might be able to survive without national security launches far better than ULA’s Vulcan.
ULA has not been unaware of the technical advances happening at SpaceX and Blue Origin. It foresaw that its market was going to be disrupted, but it committed the classic mistake of incumbents in being plagued by indecision and failure to fully embrace the new technology and engineering practices that are disrupting its business.
A couple factors in particular are disrupting the rocket industry:
- Rapid reusability with propulsive landing dramatically reduces the costs, making single-use rockets obsolete. A single-use rocket simply can’t compete economically with the first stage on the Falcon 9 Block 5 or the future New Glenn that are designed to be reused 10 times before refurbishment.
- Methane (CH4) is a superior rocket fuel to kerosene (RP-1), H2 and solid state fuels for both technical and economic reasons. Methane burns cleanly so it facilitates rocket reuse, it is cheaper, it can be made anywhere in the solar system where there is CO2 and water, and it is cheaper to store than H2. Kerosene and solid state fuels leave a lot of residue that has to be cleaned, so it isn’t good for rapid reusability. Both of them produce black carbon (soot) which stays in the stratosphere for up to 4 years, so they create radiative forcing which causes global warming. Solid state fuels leave toxic residue and can’t be shut off in the event of an abort and can’t be used for propulsive landings. Liquid H2 leaks and needs to be stored at much lower temperatures which complicates the engineering and increases the cost of the tank, rocket engine and the launch. H2 also enbrittles storage containers and fuel lines, so there are questions how many times an H2 rocket can be reused.
- Rapid innovation with continual improvements and startup company culture. Rocket development in the 1950s and 60s was based on rapid innovation, but the idea of experimentation and innovation was lost with commercialization and the need for greater reliability. New space companies like SpaceX, Blue Origin, Rocket Lab, Onespace and Ventions (formerly Astra Space) were inspired by the rapid innovation and horizontal structure of Silicon Valley startups. Making continuous improvements in design, eliminating bureaucracy, outside contractors and design requirements from outside agencies, and having the engineers working next to manufacturing in one location and manufacturing in house has allowed SpaceX to cut costs and rapidly innovate.
In addition, there are a number of technical advancements that are making it possible to better design rockets and innovate faster. Advancements in processors, robotics and artificial intelligence have made it possible to propulsively land a rocket. Human pilots never would be able to make all the necessary adjustments in milliseconds to land a rocket propulsively. Advances in 3D printing, computer modeling and CAD have made it possible to design new types of engines and launch vehicles faster and cheaper than before. New materials like carbon fiber, PICA and SIRCA heat shields, lithium batteries, composites and advanced metal alloys are making it possible to design better rockets, but companies like ULA, Orbital ATK/Northrup Grumman, Aerojet Aerodyne, Arianespace and Roscosmos are still using the same materials and often the same designs which date back to the 1960s.
ULA started designing its next-generation rocket in 2014, but it didn’t decide which engine to select–Blue Origin’s methane BE-4 or Aerojet Rocketdyne’s kerosene AR1–until late September 2018. For over four years, ULA had to design two different first stages for two different engines, when it was clear even back in 2014, which would be the superior engine. Not only did the BE-4 have better technical specs and was based on a superior fuel, but it was also farther along in its development and it had a surer source of funding to continue its development.
The new CEO of ULA, Tony Bruno, was convinced that the BE-4 was the better option and it promised to cost 60% of the price of the AR1, but he had to convince his board and the parent companies Boeing and Lockheed Martin. The board was only willing to authorize quarterly expenditures on the BE-4 engine and decided that the safest path was to also design for Aerojet Rocketdyne’s AR1 as a backup option. The AR1 didn’t have a clear source of funding to continue its development, and it was reportedly years behind schedule, but it took ULA over 4 years to finally drop it when it should have been dropped the moment Blue Origin started to successfully test fire the BE-4.
Although ULA had far more funding than SpaceX, it didn’t have the corporate culture or leadership to do much in-house R&D outside its limited core competency. ULA was essentially conceived as a cash cow for its parent companies, who didn’t want to invest too much into its R&D. Bruno tried to convince Boeing and Lockheed Martin to let him use ULA’s profits for more R&D rather than having to turn them over to the parent companies.
Because ULA outsourced its engines, it was hobbled in its ability to do R&D for propulsive landing. Back in 2014, when ULA was in talks with Blue Origin to use it BE-4 engine, it should have also been in talks to license or partner in the development of Blue Origin’s propulsive landing system for first stage reuseability. At the very least, it should have realized that it had missed the technological boat and tried to license the tech once Blue Origin and SpaceX started demonstrating the concept in late 2015, since it was clearly going to dramatically reduce launch costs and undercut ULA in the future.
ULA had been studying reusability for quite a while and decided back in 2008 that the best course of action was to only reuse the first stage’s engines, which represent two thirds of the cost of a first stage. While that wasn’t necessarily a bad decision in 2008 given the state of avionics at the time, it was clearly the wrong decision in 2015, when ULA announced that they would implement a system in their next generation Vulcan to detach the engines from the first stage with an inflatable heat shield to protect it during reentry, then use parafoils to glide them down to Earth and catch them with helicopters. The rub was that they decided to not bother implementing engine reusability until 2024, four years after they planned to first launch the Vulcan. Even today, after SpaceX has demonstrated the business case for propulsive landing and full reuseability, ULA still hasn’t changed its target date for partial reusability in 2024.
Tony Bruno argues that ULA’s plan for partial reusability makes more sense economically than full reuse through propulsive landing, because fuel doesn’t need to be expended to land the first stage. For larger payloads and payloads to GTO, there is not enough fuel left for propulsive landing and ULA would rather recover 2/3 of the cost of the first stage for 100% of launches, than recover 100% of the first stage some of the time.
Bruno’s arguments don’t hold much water. First of all, the ULA’s way of recovering the engines means that it won’t have rapid reusability because the engines will need to be flown to a recovery barge, shipped back to the factory and inspected, reattached to a new first stage and then shipped back to the launch site. In contrast, SpaceX lands its rockets either back on the landing site or on a floating barge which ships it back to the launch site. Then, all the first stage needs is a quick inspection before being reflown, so the turn around time and the costs are minimal. Vulcan’s system of flying to an exact spot with a parafoil and catching the first stage’s engines with helicopters will probably be more prone to errors than propulsive landing, so the recovery rate probably won’t be as high. It also has to be done in good weather over friendly international waters.
ULA is planning that the Vulcan will have a launch cost of $100 million, which means that even with reusability of its first stage engines after 2024, it will cost roughly:
$100M – ($100M x 60% x 2/3) = $60M
The problem is that SpaceX’s Falcon 9 will probably cost between $15 and $25 million per launch, if it has achieved full reusability of its payload fairings and it is getting 10+ launches of its first stage before refurbishment. Its RP-1 fuel only costs roughly $300,000 per launch, plus a couple million dollars in other launch costs, so most of its costs per launch will be replacing the second stage, which costs 20% of the rocket or roughly $12 million. SpaceX will be able to streamline the fabrication of its second stage and reduce its costs if the company is launching the Falcon 9 hundreds of times per year by the mid-2020s. Over 200 launches per year is likely if SpaceX controls virtually all of the global commercial launch market and implements its plan to place 12,000 satellites into low Earth orbit to provide a global internet service that it dubs “Starlink.” Even if Starlink falls through, SpaceX will still be launching roughly 50 times per year at a third of Vulcan’s theoretical cost of $60 million per launch. It is not hard to see that the Vulcan with first stage engine reuse simply won’t be competitive.
The second problem is that Bruno is assuming that the Vulcan will be competing with today’s Falcon 9 block 5 or Falcon Heavy. However, it will be competing with SpaceX’s future Big Falcon Rocket (BFR), which will be large enough deliver large payloads to GTO with full reusability of the entire rocket. SpaceX’s Starship (formerly named the Big Falcon Spaceship) will have cargo doors, rather than the expendable payload fairings found on the Vulcan/Centaur, which means an additional savings of $5 – $10 million per launch. If the BFR is 100% reusable and has as fast of a turn-around time as planned, then it potentially could cost one fiftieth of the price of the Vulcan per launch. SpaceX has set 2020 as its target date for the BFR’s first launch, which is comparable to the timeline for Vulcan’s first launch in 2021. Even taking into account Elon Musk’s tendency to announce overly optimistic timelines, the BFR should come into service at roughly the same time as the Vulcan, so they will be direct competitors.
The difficulty for incumbents like ULA is that the technological decisions that are deemed the most risky, often end up being the necessary choices to compete in the new environment. ULA’s decision to avoid developing propulsive landing makes sense from the point of view of minimizing R&D costs and risking less in a new technology which might fail. If the inflatable heatshields, parafoils and helicopters don’t work to recover the first stage engines, then ULA won’t have bet the company and it won’t have missed any major deadlines, since it is a secondary feature to be implemented after the Vulcan/Centaur is already in service. Similarly, the decision to go with traditional heat shielding rather than using NASA’s new PICA shielding makes sense from the point of view of avoiding risky new materials which might potentially fail, but SpaceX is able to use an innovative spaceship design in its Starship that bleeds off a lot of velocity in reentry and saves fuel because it was willing to adopt PICA and then improve it by learning how to manufacture it.
SpaceX’s willingness to invest the R&D to develop its own in-house manufacturing of so many components has now given it a tremendous competitive advantage over other rocket companies. Because SpaceX manufactures so many of its own components in house, it is able to update the design of its components more easily and quickly, because it doesn’t have to coordinate with external suppliers. Changing the design of one component that effects another component doesn’t devolve into the hassle of contract renegotiations over changed parts. Because almost all the engineering and manufacturing happens in Hawthorne, California, it is easy to call a meeting of all the people involved and make a design change that effects multiple components.
In contrast, the Vulcan will be designed and assembled with parts from many different suppliers. The two methane engines in the first stage will come from Blue Origin, the solid fuel boosters from Northrop Grumman, the avionics from L-3 Avionics, the payload fairings from Ruag, and the hydrogen RL10 engine for the second stage from Aerojet Rocketdyne. This outsourcing was even worse in the past. Close to 4000 companies worked on the Atlas rocket program. Not only does this out-sourcing substantially raise the costs because aerospace companies charge an arm and a leg for their components, but it also makes it very difficult for ULA to change the design for continuous improvement of its rocket, so it can’t keep up with SpaceX’s pace of innovation.
If ULA were in an ordinary industry, it would already be facing bankruptcy. Just consider the fact that ULA says that it charges $350 million to launch its Delta IV-Heavy, but it is probably closer to $400 million once all the extra fees are tacked on for compliance with government requirements and above $500 million once the extra cost of “launch readiness” is included. In comparison, SpaceX doesn’t get any government money to maintain its “launch readiness” and it just signed a $130 million contract with the Air Force to launch the Falcon Heavy, which can carry 125% weight more to lower Earth orbit (LEO) and 93% more to geostationary transfer orbit (GTO) than the Delta IV Heavy. No company operating in a remotely free market stays in business for long when its competitor does twice as much work for a fourth of the price.
The very things that make aerospace companies so uncompetitive, also make them so resistant to being challenged by the free market. The fact that ULA has to send a portion of its profits to two bloated aerospace companies and subcontracts with so many other big aerospace companies raises its costs, but it also gives it enormous political clout. When SpaceX goes before congress, it will have the 2 senators from California on its side since it manufactures in Hawthorne, is building a new factory in the port of Los Angeles and launches in Vandenberg Air Force Base. It might also get the two senators from Texas to show up since it has a test site in McGregor and launch site in Boca China near Brownsville. There is some chance that a Florida senator might pop in because SpaceX launches from Cape Canaveral, but little chance that the senators from Washington will support it, because Boeing employs so many more engineers in the state. SpaceX generates a lot of public support, but it is unclear how much influence that has in the halls of congress.
In contrast, if ULA has to face congress, it alone probably doesn’t have any more political influence than SpaceX, but its parent companies Boeing and Lockheed Martin have enormous pull in Washington. They have the lobbying power to let politicians know exact how many jobs in their districts are a stake. ULA, Boeing and Lockheed Martin together spend 17 times more on lobbying the federal government and give 11 times more than SpaceX in political contributions to federal candidates. Together they represent 35 times more jobs and control 81 times more revenue than SpaceX.
All the suppliers and subcontractors of the ULA also represent a potent lobbying force. As both the suppliers of components for ULA rockets and its potential rivals for launch contracts, the support from Northrup Grumman and Blue Origin will probably depend on the issue at hand. However, much of Blue Origin’s credibility as an aerospace company now rests on supplying engines for ULA’s Vulcan. If Blue Origin sends lobbyists to argue on behalf of ULA, politicians will pay attention because they know that the company is owned by the richest man in the world, who finances its operations by selling stock from the fourth largest corporation in the world in terms of market capitalization. If congress ever demands that the Air Force start awarding its launch contracts based on the lowest available price, then it is likely that ULA’s protests will be backed up by the lobbying power of Northrup Grumman, Blue Origin, Aerojet Rocketdyne and possibly even Amazon.
In the current political environment, ULA appears to have little to fear, despite the fact that it forces the American taxpayer to pay inflated prices for its services. Washington is governed by a bipartisan consensus in favor of military Keynesianism and political influence buying. The fact that a recent audit found that the Department of Defense made $21 trillion in unaccountable transactions between 1998 to 2015, doesn’t seem to have put any damper on congress’s willingness to lavish pork on the military-industrial complex. This bipartisan consensus, however, may not last forever if the Bernie Sanders wing of the Democratic Party comes to the fore. The number of Democrats in the US congress refusing to accept corporate PAC money increased from 4 to 40 in the last election cycle and they would see the inflated launch prices that the ULA charges as money better on fixing the lead pipes in Flint, Michigan and switching to 100% renewable energy.
Only recently has competition with SpaceX become a serious problem for the ULA. Its greater problem has been the vagaries of congress in funding projects at NASA and the space section of the Air Force. As the building of the Space Launch System (SLS) and Orion has ramped up to go to Mars under Obama and to the Moon under Trump, it has shrunk the funds available for other projects that employ ULA rockets. Of course, that shift directs more money towards its parent company, Boeing, so ULA can’t complain too loudly.
Republicans are offended by the fact that NASA studies how humans are changing the climate, so they are shifting funds away from Earth Observation, whose satellites, such as the Ice, Cloud and land Elevation Satellite-2 (ICESat-2), are often launched by ULA. President Trump also tried to remove the funding for the International Space Station (ISS) in 2019, which also threatens ULA, since its rockets have twice launched to the ISS when Orbital’s rockets were delayed and its rockets will launch Boeing’s CST-100 Starliner to the ISS as well. An even greater threat to the ULA has been the fact that some Republicans in the House have lost confidence in the space division of the Air Force, which is why they want to create a separate Space Force, that Trump has also picked up as his pet project. Funding for space at the Air Force was reduced by 7.3% in 2019.
The vagaries of politics can be seen in ULA’s varying levels of funding. According to Govini, ULA’s contracts from the US government grew from $1.1 billion in 2011 to over $3.6 billion in 2014. Then, it fell to $1.8 billion in 2016, and then grew to $2.2 billion in 2017. ULA received a total of $14.6 billion in launch contracts with the US government between the fiscal years of 2011 and 2017, which represented 48.9% of government spending on launch vehicles during that time period. ULA controlled 95% of government spending on medium-lift vehicles, but its funding has not been steady.
Living in a protected government market where the ULA can charge prices well above its international competition from Russia and Europe means that ULA also has to live by the vagaries of US politics. One of the reasons why there has been so much consolidation of US military contractors is the fact that larger companies have greater lobbying power and more ability to survive a government budget cut and the long waits between funding cycles.
Being the disjointed child of the two largest government aerospace contractors has helped protect the ULA politically, but it has also hobbled the company. If the ULA were an independent company, it could lobby against Boeing’s SLS/Orion as a boondoggle wasting tax payer money. Even better, it could be designing its own super heavy-lift and space craft like SpaceX and Blue Origin are doing to challenge Boeing. Not living under the controlling thumb of two aerospace giants would free it up to take risks like trying to design its own engines, avionics, playload fairings, etc, or even branch into new areas like satellites as SpaceX is doing to challenge the satellite divisions of Lockheed and Boeing.
What the case of ULA demonstrates is how incumbents often fail to respond adequately to technological disruption. Incumbents often see the new technology coming, yet that foresight doesn’t guarantee that they will make the right decisions to deal with it. Kodak anticipated digital cameras long before any of the companies which now make digital cameras. It was a Kodak engineer who created the first digital camera in 1975, yet the company never properly embraced the new technology until it was too late out of fear of cannibalizing their existing sales.
In a similar way, ULA anticipated reusability and even wrote a paper for how to implement the reuse of its first stage rockets in 2008, but the company made no effort to implement it. Then, it finally incorporated reusability in its next generation of rockets, but in a limited way, that involved the least change to its current design and it is only plans to implement it 9 years after its competitor started demonstrating the concept.
ULA gave up trying to be competitive on a world stage and it retreated into a government protected market where it enjoyed monopoly rents. This decision made it even more vulnerable to being disrupted when SpaceX appeared on the horizon, employing disruptive technologies (advanced avionics and propulsive landing providing reusability, methane engines, PICA heat shields and higher pressure engines with greater thrust-to-weight ratios), new engineering practices (continuous iterative improvements with testing, engineering working closely with manufacturing in one location, in-house design rather than out-sourcing, cheaper in-house and off-the-shelf components with redundant fault tolerance, and horizontal rather than hierarchical decision making) and new business practices (an aspirational goal beyond short-term profit, public pricing, public involvement in advances and growing the market by driving down prices).
ULA should have moved as quickly as possible to copy both the disruptive technology and the disruptive practices of its new competitor, but its culture and its business practices allowed only a partial embrace of disruption. ULA recognized the benefits of reusability, but it was unwilling to embrace propulsive landing that would make it possible, because it involved engineering new avionics and engines that it had always outsourced in the past and were outside its core competency. Instead, it only plans to implement reusability in a way that doesn’t involve major changes to its rocket design so it doesn’t have to undertake risky and costly R&D. It recognized that methane was a superior rocket fuel that would enable a high performing engine with fewer drawbacks than the existing fuels, but it took a wait-and-see attitude and dithered for years before finally committing to its use.
ULA had the opportunity to partner with another company, Blue Origin, to develop these new technologies, but instead it chose to treat Blue Origin like another outside contractor providing it with components, rather than embracing the new tech. ULA’s inability to fully embrace the disruption, but stay as much as possible with its old design and the old practices which made it successful in the past is a classic response of incumbents. There is a certain logic to this conservative response, since it is often better for a company to stick with what it knows, rather than do poorly what it doesn’t know. However, this response is a recipe for failure in the long term, when the competition is already functioning at a third of the incumbent’s costs and is planning to get to a tenth or maybe even a hundredth of its costs.
The way that ULA responded to the threat of disruption is illustrative of incumbents which enjoy privileged positions in their markets. First, it used its political influence to obtain a long-term contract that would lock out its competitor for the next 5 years. That strategy of manipulating the market worked to some degree, although the disruptor was able to use the courts and public pressure to crack open the Air Force’s bidding process so that it started to win some contracts after it was initially locked out. ULA’s response was not that different from the way that hotels, taxi drivers and the auto dealers have used their influence with local governments to try to exclude Airbnb, Uber and Tesla from their markets.
The second response of the incumbent was to announce a restructuring of its business to respond to the disruptor. This is commonly done to reassure investors that they shouldn’t lose faith in the incumbent as viable business. In ULA’s case, it isn’t a publicly traded company, but it had to reassure stockholders in its two parent companies. A new CEO was selected in 2014 who promised to cut its launch costs in half in order to better compete with SpaceX. In ULA’s case, the disruptor wasn’t the immediate threat, but rather political decisions in government funding, but the disruptor was used as a convenient justification for cost cutting. ULA reduced its number of executives by 30%, laid plans to reduce its number of launch sites from 5 to 2 and laid off workers.
Restructuring and cost cutting rarely works, because it usually doesn’t address the core problem when the disruptor has a significantly better product. In fact, laying off workers can demoralize the company’s workforce and make it harder for the incumbent to adopt the disrupting technology and/or business model that is necessary for its long-term survival.
The third response of incumbent is to develop new products with the disrupting tech. Many incumbents try to maintain their existing product lines, while introducing the disrupting tech in a new product line, that they try to isolate as much as possible. IBM used a separate engineering group to produce its original PC line in 1981 and tried to separate the members of that group as much as possible from its mainframe and minicomputer divisions, in order to be able to develop the PC using new rules, that would allow it to compete with Apple, Commodore and Radio Shack.
Tom Mueller, SpaceX’s chief architect of its rocket engines, provides an insightful summary of how the incumbents responded to the arrival of SpaceX:
And you know, we were ridiculed by the other big companies in the launch vehicle business. At first, they ignored us; and then they fought us; and then they— we found out; they found out that they couldn’t really win in a fair fight because we were successful and we were, you know, factors of two or three or probably even five lower costs than what they can do. So then it becomes an unfair fight, where they, you know, try to destroy you politically, and use other means. And then at some point, they figure out that they’ve got to do what you’re doing. So there’s a lot of talk at these other companies about how they’ll make reusable rockets; recover the engines, recover the stages, come up with a much lower-cost rocket so that they can compete. You know, there’s no way that they ULA would have considered buying engines from Blue Origin except for the pressure that SpaceX is putting on them. There’s no way that the French would have quickly abandoned the Ariane 5 and moved to the Ariane 6 design because— except for the pressure we’re putting on. So we’re really changing the world… Anyway, it’s great that we’re changing the paradigm and causing everybody else to think differently about how this is done.
What is interesting about ULA’s response is the fact that it isn’t using the skunkworks approach as IBM did. It is under pressure from Congress to stop using the RD-180 engine made in Russia, so it needs to replace the Atlas V with its new Vulcan rocket. It is adopting a disruptive methane engine and the idea of reusability, but in a cautious evolutionary way, so that it won’t have to change the design of its rocket too much and can avoid costly and risky R&D on new tech. This approach also avoids creating a potential competitor to the SLS being developed by its parent company, Boeing. Some NASA administrators are already admitting that the SLS is so over-budget and behind schedule that it might be scrapped if SpaceX’s BFR and Blue Origin’s New Glenn become viable replacements.
What is also interesting about ULA’s response to disruption is that it has been allowed to dither in deciding its course and hedge its bets on adapting to disruption over a time frame stretching from 2014 when it started planning the Vulcan to 2024 when it plans to have reusable engines. Most incumbents operating in freer markets don’t have a decade to respond and they don’t have the luxury to only partially implement the disrupting tech and not significantly change their engineering and business practices.
In all likelihood, ULA will be one of the two companies selected by the Air Force in 2020, so its business will be secure until at least 2025. In fact, the EELV 2 rules created by the Air Force and NRO to foster more competition and lower launch prices will probably end up protecting ULA. When the Air Force decided to implement the EELV 2, it was clear that there would soon be four viable competitors in the launch market, so their rules will limit the competition and keep out the challengers Blue Origin and Orbital ATK/Northrup Grumman. One of the companies selected by the EELV 2 in 2020 will get 60% of the contracts and the other will get 40%, so the Air Force has essentially guaranteed that ULA will still have 40% of the government launch market even though its Vulcan will be outdated and too expensive.
A cynical observer might believe that the Air Force deliberately created the EELV 2 to protect ULA, but the Air Force argues that that the market is only large enough to support to two competitors who cater to both the commercial and governmental launch markets. Also, the EELV 2 is helping to finance the development of new rockets by Blue Origin and Northrup Grumman, plus it is paying for the development of Aerojet Rocketdyne’s RD1 engine, so it is expanding the pool of potential competitors in the future. Much of this money is being wasted. Northrup Grumman’s solid-state boosters may be good technology for ICBMs which need to be ready to launch at a moment’s notice, but the OmegA will be even more outdated than ULA’s Vulcan. The Air Force is investing in yesterday’s rocket technology by financing the developing of the kerosene AR1 engine. SpaceX already has a highly performant kerosene rocket, and methane promises to be a much better rocket fuel.
Nonetheless, if both SpaceX and Blue Origin have functioning next-generation rockets by the mid-2020s, which appears likely at this point, then ULA will be driven out of business. ULA appears to be gambling that the US government and its bloated parent companies will always protect it, but the reality is that it needs to get to full reusability if it wants to guarantee its long-term survival. At this point, it looks like Nokia–a company which had 60% of the smartphone market in 2006 before the arrival of the iPhone and Android, but it waited too long to fully commit to any one course (Linux, open sourcing Symbian S90 or Windows Mobile), so it ended up with none of the market.
ULA’s only realistic hope in the long term is to fully embrace propulsive landing and full reusability, but it is facing a catch 22. If it announces tomorrow that its Vulcan is being radically redesigned for full reusability, then it won’t be selected in 2020 as one of the two companies under the EELV 2, so it needs to continue with the conservative design of the Vulcan to reassure the Air Force and NRO that its rocket will be ready in 2022 for the next round of launch contracts.
There are generally three outcomes for incumbents facing disruption. Most incumbents end up buying the companies which threaten them with a disruptive technology or business model, which is generally the best solution, since the incumbents generally don’t have the kind of corporate culture or leadership that would allow them to easily develop the disruption, but they do have the experience and resources to successfully market it. The other outcome which happens less often is that the disruptor buys the incumbent, as happened when Compaq bought Digital Equipment Corporation. The third outcome to disruption is that an outside company will buy the incumbent as a way to get into the market. Often the buying company thinks that this will create a synergy with its existing products, but it can end in disaster if the buying company fails to address why the incumbent was losing market share, as happened when HP bought Palm and when Microsoft bought Nokia.
What is interesting about ULA’s case is that none of the three normal outcomes to disruption appear likely, so there is no easy escape route for the company. If ULA were being disrupted by normal startups, it could buy them, but both SpaceX and Blue Origin are controlled by founders who have visions for sending humans into space, so they have no interest in cashing out and certainly not to a company like ULA which has historically been guided by short-term profits and limited innovation.
Ideally, Blue Origin or SpaceX would buy ULA, because it has what these companies lack–the experience of launching reliably, dealing with government bureaucracies and maintaining deep ties with the Air Force and NRO to win government contracts. Boeing, however, will not want to help create a stronger competitor to its SLS/Orion and both Boeing and Lockheed Martin will want to keep a foothold in the growing space market. SpaceX is threatening to disrupt both Boeing and Lockheed in satellites as well, so neither company will want to help SpaceX get a better foothold as a national security contractor.
It is hard to see any outside companies being able to buy ULA, not only because it has to be an American company, but it has to be a company whose business bears no threat to Boeing and Lockheed. Aerojet Rocketdyne recognized that it needed to combine both rocket and engine manufacturing in order to compete with SpaceX, but it is so financially weak, that it couldn’t offer more than $2 billion for ULA, which is basically ULA’s annual revenue. The most obvious candidate would be Northrup Grumman, but it is too much of a competitor to ULA’s parent companies, and it is hard to see how it would design a next generation rocket any better, given what it is currently designing in the OmegA.
Given how hobbled the ULA is, it might be better for ULA to be split up and return to its two parent companies, because at least one of the halves would have a better chance of securing the kind of deep investment from its parent that is needed to implement propulsive landing and full reusability, but the financial situation of ULA will probably have to turn dire before a breakup takes place. At that point, both Boeing and Lockheed may decide that they have to stay in the launch business, just like Sony, HTC, LG and Motorola/Lenovo have stayed in the smartphone business despite losing money on every phone they sell. However, they also might look at the launch prices that SpaceX is charging in the future, and decide that it is not worth investing any more in such a low-margin business, just like IBM decided to get out of the PC business and then the x86 server business because it foresaw little chance of future profits on such thin margins.
Given its available options, ULA’s best chance for survival is to start a skunkworks now which works on propulsive landing. It should determine whether propulsive landing can be added to the Vulcan, or whether it requires a whole new rocket with smaller methane engines than the BE-4. Either way, it should be designing a fully reusable rocket with propulsive landing to be ready for the next round of the EELV starting in the mid-2020s.
The other possibility is to license propulsive landing tech from Blue Origin, but this will become increasingly difficult the closer that the New Glenn is to flying and can demonstrate that Blue Origin can replace ULA as the second Air Force/NRO launch provider. ULA would do better to form a partnership with Blue Origin right now to develop propulsive landing, but ULA will have to accept humiliating terms as a junior partner, since it brings less to the table. Blue Origin has already demonstrated the technology on small rockets. It also has the money, engineering talent, startup ethos and leadership needed to develop the next generation of rockets.
Disruption has happened in many industries, and rarely have the incumbents handled it without significant turmoil. It is hard to say at this juncture, what will happen to an incumbent like ULA which can’t take any of the normal routes (buying the disruptor or being bought by the disruptor or by an outsider trying to get into the market), but also has time to adapt to since it probably will receive military contracts that secure its business until 2025. Because ULA occupies such a unique position, what course it chooses and how long it delays in deciding its course will be fascinating to watch.