Last week, we attended the Pipeline Pigging and Integrity Management Conference and Exhibition (“PPIM”), the pipeline industry’s only forum devoted exclusively to maintenance, inspection, and integrity evaluation. We wanted to see the extent to which the industry was embracing machine learning to manage and protect its most valuable assets – pipelines. In this report we detail our findings, but before we dive in below is a brief summary of our last report.
In our report ESG Investment for the Petroleum and Gas Pipeline Industry, we detailed that:
- Sophic Capital client OneSoft Solutions (OSS:TSXV, OSSIF:OTC) is an Environmental, Social and Governance (“ESG”) firm making the fossil fuel pipeline industry safer for workers, communities and the environment;
- U.S. pipeline failures may be occurring, in part, because legacy data analysis techniques depend upon Microsoft Excel, which only typically analyze 5 to 10% of total pipeline data collected;
- OneSoft has a machine learning, predictive analytics solution that analyzes 100% of the pipeline data — faster, more accurately, and at a lower cost;
- Six Fortune 100/500 companies, including Phillips 66 (PSX:NYSE) and an unnamed industry Supermajor, have licensed OneSoft’s solution following extensive validation efforts;
- OneSoft’s current clients operate about 7% of the U.S. piggable pipeline market, and OneSoft has almost half of the addressable U.S. piggable market in its sales funnel;
- OneSoft has: several opportunities to capture the remaining addressable U.S. piggable market; a plan to target the non-piggable market (developing new software modules for direct assessment); options to add new revenue generating functionality to its solution; and potential to expand beyond its current target market, and;
- OneSoft has a sizable competitive moat, making it difficult for new entrants (of which there currently are none that have developed and commercialized a machine learning and advanced data science platform to aggregate and analyze data across the industry) to catch up and compete.
Digital Transformation in the Pipeline Industry Has Been Slow but is HappeningAlthough the pipeline industry understands that machine learning can predict failures faster and more accurately at significantly lower costs, adoption has been slow. Pipeline operators are certainly becoming more aware of machine learning. We saw several1 at OneSoft’s booth, engaged in discussions and/or watching a video of OneSoft’s machine learning pipeline integrity management solution.
Pipeline inspection gauge (“PIG”) tool vendors also recognize that machine learning is the industry’s future. We spoke with three large, industry-leading tool vendors (two of whom we also spotted at OneSoft’s booth), and all of them told us that they had either had or were in the process of developing machine learning solutions for their own tools. What differentiates OneSoft is that its machine learning solution is tool-agnostic, meaning it can ingest and process data that any PIG collects. This is highly advantageous because OneSoft’s platform has by now aggregated tens of thousands of miles of pipeline data from numerous tool vendors and has learnings from more than 50 million features – something that no single PIG vendor could accomplish. We also met some services companies that process PIG data. These firms were becoming aware of machine learning but seemed attached to their own legacy processes, for now.
The benefits of machine learning have yet to reach pipeline workers. We spoke with several professionals who dig up and repair pipelines. In our discussions, industry messaging about improving safety and eliminating all pipeline incidents had certainly reached these front-line workers. However, the industry has not communicated with them the benefits of machine learning and how it can help achieve the zero-failure objective. When we described the technology, these folks quickly grasped machine learning’s value proposition.
Machine Learning Courses and Presentations at PPIM 2020
Machine learning’s prominence at PPIM 2020 was highlighted in the forum’s educational sessions. A Practical Application of Machine Learning to Pipeline Integrity was a two-day course that we attended. This is the first time that we’ve seen a formal instruction dedicated to machine learning for pipeline integrity management. And for anyone who believes that the pipeline industry is for Luddites, these folks could easily ace all three CFA exams2 with the amount of linear algebra, calculus, and statistics covered. After spending a morning looking at bell curves, vectors, and integral signs, we couldn’t understand why these intelligent engineers have been slow to adopt machine learning. So we asked them:
“We’ve actually been trying to do machine learning for 12 years to help manage our network. We need it because the environment where we operate is unique in the world and very corrosive to our pipes.” I approached this person to ask why they developed their system in-house. He replied that no machine learning solution existed when they first initiated their internal program, so they attempted to build their own. When asked if it met their needs, he replied “no” and said that was why he and two colleagues were taking the PPIM course.
“We need to do some background work to learn more.” This engineer said his firm had a small pipeline network that was expensive to maintain. They had too many false digs (he guessed around 30 to 40% but didn’t know for sure), meaning their work crews would dig up a site but couldn’t identify the issue they expected to find. His team understood the concepts about data science and wanted to see how they applied to pipeline integrity management. Oddly enough, this team was also considering developing their own system in-house rather than seek a third-party solution. Why? They don’t want to share their data with tool vendors and get locked into a solution dependent upon a vendor’s PIGs. As mentioned before, OneSoft’s machine learning solution is tool-agnostic, and many in the industry (including 6 Fortune 100/500 customer) view OneSoft as a neutral third-party.
“Machine learning is the future for this industry.” This person was interested in machine learning for non-piggable pipeline, which comprises about 2.1 million of the USA’s 2.7 million miles of pipeline. Interestingly, this person had tried to contact C3.ai for information about its machine learning platform but said the company was not forthcoming with any information and speculated that perhaps they still had none. He had never heard of OneSoft Solutions, so I gave him the booth number. He said he’d have a look. We asked if software development was core to his company. His response: “It will be.”
Regulatory Environment Will Likely Help Drive Machine Learning Adoption
Over 59% of America’s gas transmission pipelines were installed before 1970. Age isn’t that critical, however, if pipelines are regularly inspected and well maintained. Given that: a) incumbent, Excel-based analysis techniques only examine 5 to 10% of PIG data AND b) PPIM continues to grow year-over-year as the key forum to share information about new pigging tool technologies and integrity management processes, we suspect that pipeline integrity methodologies are still in great need for improvement.
New PHMSA (the government agency that regulates America’s pipeline industry) rules could expand OneSoft’s U.S. market by 20%. The rules now include gathering pipelines, (i.e., those pipes that transport crude and natural gas from the wellhead to a transmission mainline or storage facility). Based upon our discussions with industry experts at PPIM 2020, this new rule could increase the total number of regulated U.S. pipelines by 20%. Services companies told us that pipeline companies were preparing for these regulation changes and looking for ways to be more proactive, rather than reactive.
PHMSA’s changes will affect the integrity management plans of pipeline operators. The most significant changes call for: a) increased management of high consequence areas (like freeways and bridges) that would suffer severe consequences in the event of an incident; b) preventative measures to minimize the chance of an incident, and; c) prescriptive data gathering and assessment. Operators will need to collect, store and analyze more high-fidelity data. There are numerous tool vendors, survey firms and services companies that are more than qualified to do this. However, PHMSA plans to mandate new requirements for risk assessment validation and the ability to identify the likelihood of threats and implied consequences across each pipeline segment. This requires integrating and processing big and disparate data sets, exactly what OneSoft Solutions excels at.
Proprietary Machine Learning Solutions Exist, BUT…
… proprietary machine learning solutions have limited markets. Pipeline operators tend to avoid single sourcing tool vendors because they want to retain price competitive advantages. Perhaps more important is that they don’t want to hand over data or control over their pipeline assets to any third party. Pipeline operators therefore multisource pigging services and do not share their data with alternate PIG vendors.
This bodes well for OneSoft Solutions because it is not a tool vendor. OneSoft’s ingestion algorithms work with any PIG data set provided from a pipeline operator, with some data sets dating back as far as 30 years. While OneSoft keeps the learnings from all these data sets (which it can then share based on data-anonymous basis with all users of its solution) pipeline operators maintain control over all their data.
And here is the biggest “BUT”: Although the large tool vendors we spoke with are developing proprietary machine learning solutions, they knew about OneSoft. One representative told us that OneSoft had the “most complete machine learning solution”. He was impressed with OneSoft‘s current client base (he named Phillips 66) and said it was “a pure big data company”. When we asked if he knew anyone else competing with OneSoft, he said there were some firms trying to replicate OneSoft‘s success (he couldn’t recall their names) but OneSoft was “miles ahead”.
What It All Means
Although the pipeline industry is lagging when it comes to adopting machine learning / predictive analytics, the operators understand how this new technology will help achieve the industry’s goal of safeguarding the environment, protecting communities, and eliminating worker injuries and deaths. For years, the industry has been exploring how to improve its integrity management processes but has only recently began to consider machine learning as the catalyst for such change. Based upon the PPIM courses that we attended and informal chats we had with industry professionals, we’re confident that machine learning will be the technology that eliminates pipeline incidents not only in the United States but across the world.
What It All Means to You
OneSoft Solutions is a pureplay machine learning firm working to eliminate pipeline failures in the oil and gas industry through better detection of potential threats. OneSoft is targeting the U.S. first since it is the largest and most regulated pipeline market. Recent new regulatory rules in the U.S. could spur increased demand for OneSoft’s solutions, even though machine learning adoption in the industry is still in its infancy. OneSoft has no direct competitors yet, and the tool vendors we spoke with that are developing proprietary machine learning solutions spoke favourably about OneSoft. The Company’s balance sheet is strong, and management has a large vested interest (about 34% ownership) to make OneSoft succeed. OneSoft has an impressive customer list, a sales pipeline that is tracking almost half of the current addressable U.S. piggable market, a plan to pursue the non-piggable market, and other catalysts that could drive adoption of OneSoft’s solutions. We believe that 2020 could be the year for OneSoft to scale rapidly.
OneSoft Solutions was the only pureplay machine learning company that we saw at PPIM 2020, thus reinforcing its competitive moat. Although three large tool vendors we spoke with have or are working on their own solutions, they are disadvantaged because pipeline operators don’t single source a vendor or share their data. This limits the data they have access to, which severely limits the usefulness of their proprietary machine learning solutions. This also applies to the large pipeline operators whose strategies are to develop their own in-house solutions. OneSoft Solutions, on the other hand, has designed its system as a neutral service provider that can ingest and analyze vendor-agnostic data and share learnings with clients without sharing any data, a concept that is highly embraced by most pipeline operators and PHMSA.
Regulatory, social, and financial pressures are driving the oil and gas pipeline industry to seek new solutions to protect communities, workers, and the environment. OneSoft’s customers have validated the Company’s machine learning solution and now use it as their fundamental integrity management technology to better mitigate risks associated with pipeline failures.
OneSoft (OSS:TSXV, OSSIF:OTC) nicely dovetails into the Environmental, Social, and Governance (“ESG”) investment theme that is globally trending and growing. OneSoft has a solid balance sheet to execute its current business plan and grow into the ESG theme by addressing the regulatory, social, and financial issues that the U.S. oil and gas pipeline industry is contending with.
For more information about OneSoft Solutions visit www.OneSoft.ca.
The information and recommendations made available through our emails, newsletters, website and press releases (collectively referred to as the “Material”) by Sophic Capital Inc. (“Sophic” or “Company”) is for informational purposes only and shall not be used or construed as an offer to sell or be used as a solicitation of an offer to buy any services or securities. In accessing or consuming the Materials, you hereby acknowledge that any reliance upon any Materials shall be at your sole risk. In particular, none of the information provided in our monthly newsletter and emails or any other Material should be viewed as an invite, and/or induce or encourage any person to make any kind of investment decision. The recommendations and information provided in our Material are not tailored to the needs of particular persons and may not be appropriate for you depending on your financial position or investment goals or needs. You should apply your own judgment in making any use of the information provided in the Company’s Material, especially as the basis for any investment decisions. Securities or other investments referred to in the Materials may not be suitable for you and you should not make any kind of investment decision in relation to them without first obtaining independent investment advice from a qualified and registered investment advisor. You further agree that neither Sophic, its, directors, officers, shareholders, employees, affiliates consultants, and/or clients will be liable for any losses or liabilities that may be occasioned as a result of the information provided in any of the Material. By accessing Sophic’s website and signing up to receive the Company’s monthly newsletter or any other Material, you accept and agree to be bound by and comply with the terms and conditions set out herein. If you do not accept and agree to the terms, you should not use the Company’s website or accept the terms and conditions associated to the newsletter signup. Sophic is not registered as an adviser or dealer under the securities legislation of any jurisdiction of Canada or elsewhere and provides Material on behalf of its clients pursuant to an exemption from the registration requirements that is available in respect of generic advice. In no event will Sophic be responsible or liable to you or any other party for any damages of any kind arising out of or relating to the use of, misuse of and/or inability to use the Company’s website or Material. The information is directed only at persons resident in Canada. The Company’s Material or the information provided in the Material shall not in any form constitute as an offer or solicitation to anyone in the United States of America or any jurisdiction where such offer or solicitation is not authorized or to any person to whom it is unlawful to make such a solicitation. If you choose to access Sophic’s website and/or have signed up to receive the Company’s monthly newsletter or any other Material, you acknowledge that the information in the Material is intended for use by persons resident in Canada only. Sophic is not an investment advisor nor does it maintain any registrations as such, and Material provided by Sophic shall not be used to make investment decisions. Information provided in the Company’s Material is often opinionated and should be considered for information purposes only. No stock exchange or securities regulatory authority anywhere has approved or disapproved of the information contained herein. There is no express or implied solicitation to buy or sell securities. Sophic and/or its principals and employees may have positions in the stocks mentioned in the Company’s Material and may trade in the stocks mentioned in the Material. Do not consider buying or selling any stock without conducting your own due diligence and/or without obtaining independent investment advice from a qualified and registered investment advisor. The Company has not independently verified any of the data from third party sources referred to in the Material, including information provided by Sophic clients that are the subject of the report, or ascertained the underlying assumptions relied upon by such sources. The Company does not assume any responsibility for the accuracy or completeness of this information or for any failure by any such other persons to disclose events which may have occurred or may affect the significance or accuracy of any such information.
The Material may contain forward looking information. Forward-looking statements are frequently, but not always, identified by words such as “expects,” “anticipates,” “believes,” “intends,” “estimates,” “potential,” “possible,” “projects,” “plans,” and similar expressions, or statements that events, conditions or results “will,” “may,” “could,” or “should” occur or be achieved or their negatives or other comparable words and include, without limitation, statements regarding, projected revenue, income or earnings or other results of operations, strategy, plans, objectives, goals and targets, plans to increase market share or with respect to anticipated performance compared to competitors, product development and adoption by potential customers. These statements relate to future events and future performance. Forward-looking statements are based on opinions and assumptions as of the date made, and are subject to a variety of risks and other factors that could cause actual events/results to differ materially from these forward looking statements. There can be no assurance that such expectations will prove to be correct; these statements are no guarantee of future performance and involve known and unknown risks, uncertainties and other factors. Sophic provides no assurance as to future results, performance, or achievements and no representations are made that actual results achieved will be as indicated in the forward looking information. Nothing herein can be assumed or predicted, and you are strongly encouraged to learn more and seek independent advice before relying on any information presented.