Good morning, everyone, thank you for joining us for Bentley Systems Financial Results webcast and conference call for the third quarter ending September 30th, 2020. I'm Carey Mann, Bentley's Officer for Investor Relations. On the call today, we have Bentley Systems Chief Executive Officer, Greg Bentley, and Chief Financial Officer, David Hollister. Before we begin, allow me to provide a disclaimer regarding forward-looking statements. This call, including the question and answer portion of the call, may include forward-looking statements related to the expected future results for our company and are therefore forward-looking statements. Our actual results may differ materially from our projections due to a number of risks and uncertainties. The risks and uncertainties that forward-looking statements are subject to are described in our earnings release and other SEC filings. Today's remarks will also include references to non-GAAP financial measures.
Additional information, including reconciliation between non-GAAP financial information to the GAAP financial information, is provided in the press release and supplemental slide presentation. This webcast will be available for replay on Bentley Systems Investor Relations website at investors.bentley.com. Greg will begin with an overview of Bentley Systems. David will then take you through a review of the financials, and then Greg will review business developments before we proceed to Q&A. With that, I'll turn the call over to Greg.
Good morning. I'm going to briefly reprise our roadshow presentation from September in order to set the stage for our first quarterly update, which will be next. First, in introducing Bentley Systems, infrastructure supports our economy and environment. That's supported by the work of infrastructure engineering. They depend upon infrastructure engineering software. We go so far as to think of BSY as the infrastructure engineering software company. That's because it's been our mission for these 36 years since our founding, during which we've developed the most comprehensive portfolio for going digital in infrastructure engineering. We've invested everywhere in the world, become dependably growing and compounding, including through 2020, by virtue, in particular, that the majority of our revenues, 87%, are recurring, the majority of those revenues renewed annually, paid in advance. The majority from accounts with whom we have enterprise relationships.
They spend at least $250,000 per year with us, and our EBITDA margins are comfortably ahead of 30%. It's particularly important to understand our end market footprint, if you like. They are the sectors of infrastructure. I'm going to show the proportion of our revenues in each sector, each segment within a sector, based on accounts specific to those sectors or products specific to those sectors. The first of those, commercial and facilities. That would be vertical infrastructure buildings, if you like, which is the proportion shown here. The next sector, industrial and resources. The remainder of our revenue are segments within the sector we call public works and utilities. You see that those include roads and bridges, utility transmission and distribution, municipal and mapping, rail and transit, water and wastewater. Together, that's public works and utilities.
We have also accounts and products that are generalized, that are not particular to one of these sectors, and that makes up the balance of our revenue footprint. The portion in public works and utilities, including ports and airports alone, is 3% of GDP, and we believe we're the market leader in roads and bridges and utility transmission and distribution, and rail and transit, in water and waste, and in the disciplines of structural and geotechnical engineering, and in the sector of public works and utilities overall. How big a market can that be? It happens there are 22.5 million infrastructure engineers in the world. They recently have spent $8.3 billion annually on software, which works out to $369 per such engineer. While there are only half as many product engineers in the world, they spend more on software, in fact, $1,250 per year.
Our total addressable market, if and I think when each infrastructure engineer would spend as much as already do the product engineers, will be $29 billion. We think of that product engineering market in investment terms as the PLM segment, where the L is for lifecycle. Indeed, expanding toward our TAM can depend on that. In our company's history, we started in 1984 with the modeling of infrastructure. We had one product, MicroStation, but it was the platform for what are now all of our specialized products for modeling by sector and discipline. We began developing and acquiring the simulation engines that infrastructure engineers depend upon. Since 1999, we've also been in the business of the collaboration systems for project delivery, our ProjectWise environment. In 2009, it happened that we had a very comprehensive portfolio for project delivery.
New capital projects stopped, as you can recall, in the great financial crisis, and we realized we were missing an opportunity to add value during the operations and maintenance life cycle of those same projects as they became assets. We have since then invested in our AssetWise portfolio, representing now the proportion shown here of our revenues. The majority of our revenues recently are still modeling and simulation applications. These generally are on-premise seats that we connect together with cloud services to keep them current. Our systems are faster-growing. Of course, they're strictly by subscription, and Azure managed services or still some on-premise hybrid, but increasingly native SaaS. We became, in 2019, one of the top 25 users of Azure for the purpose. That continues now.
If you like the consecutive nature of these functions, if we introduce now what we call digital twins, a cloud service to connect together an evergreen environment for the full life cycle of infrastructure assets, then the functions of modeling and simulation especially can be repeated and add value to maintain adaptive fitness for purpose for infrastructure assets, whatever would come along, including resilience challenges like a pandemic. These 36 years have made us the most comprehensive across sectors and disciplines. That matters because most substantial infrastructure projects require all of the sectors and disciplines to work together interoperably. Uniquely, we cover the whole infrastructure life cycle, and geographically, the majority of our business is outside the U.S. The majority of our colleagues are in the world where the opportunities for infrastructure are greatest. By commercial model, we still sell perpetual licenses.
Our principal competitor, Autodesk, does not, is the reason we do. We have some episodic professional services, but more and more of our professional services are part of success plans, we call it, that are part of our enterprise relationships and recurring we build. Our select subscriptions are for those with perpetual licenses, but more than just maintenance. Our enterprise subscriptions are either enterprise ELS, enterprise license subscriptions, or E365, which we'll look at at greater length today. We offer quarterly and monthly term licenses. That's the recurring majority of our revenues. Within our ARR, most by far renews annually. As I say, we have term licenses that are monthly and quarterly. The E365 enterprise program is based on daily consumption, and of course, that would be subject to the greatest volatility, which we'll report on today.
Our growth strategies, we primarily grow within our existing accounts because, after 36 years, there are few new names among infrastructure engineering organizations. We have a priority to force multiply our inside sales efforts to grow in smaller and medium accounts in particular. We focus on Asia, we're bringing to market new cloud services jointly developed and jointly sold, in particular with Microsoft and Siemens. Acquisitions are programmatic for us, we're particularly investing, and we'll report today on investments in digital integrators to help to create and curate infrastructure digital twins. Now David Hollister will report on our third quarter of 2020 financial metrics.
Thank you, Greg, and good morning, everyone. I'm first going to discuss our third quarter 2020 results, and then we'll offer some views on our full year 2020 performance expectations. I'll start with third quarter revenues, which grew 8.8% year-over-year to reach $203 million. This is fairly consistent with our year-to-date growth of 9%. Breaking that down further, subscription revenues, which are over 85% of our total revenues, grew 11.6%, while our perpetual licenses and services revenues declined by 7% and 3.5%, respectively. These more episodic perpetual license and services revenues are typically, for us, where we'll feel softness in the face of macro headwinds and cyclicality. To a lesser degree, we'll also see any macro-induced softness manifest in certain of our shorter-term subscriptions, and in particular, our E365 daily consumption-based subscriptions.
Greg will further discuss, while we're resilient and growing, we're cautious about the lingering effects of the pandemic on capital project starts, in particular within the commercial and facilities sector and industrial and resources sector, where the combination of these end markets represent about one-third of our revenues. The public works and utilities sector, our largest end market, representing about two-thirds of our revenues, while not unaffected, remains relatively robust for us. You will have noted that our last 12-month recurring revenues, which include primarily our subscription revenues, but also will include certain services revenues delivered under contractually recurring success plans, increased by 11%. This is consistent with continued strong performance in our net recurring revenue retention rate of 110%. Our deep and long relationships with our accounts, again, are proven with our 98% account retention rate.
Growing and expanding with those accounts continues to be our most prevalent source of growth and an area where we continue to invest with our user success adoption initiatives. A significant KPI for us is our annual recurring revenue, or ARR. As of September 30th, 2020, our ARR has grown by 9% on a constant currency basis since the same time last year. Our GAAP operating income was $5.3 million for the third quarter 2020, compared to $41.4 million for the same period last year. There is a significant distortion in these results, and I will next highlight several of these items. Firstly, our GAAP results include a charge of $26.1 million for costs directly associated with our IPO in September. Uniquely, our IPO did not include the issuance of any new shares for Bentley Systems.
Normally, IPO costs are netted against the proceeds from primary share issuances and do not flow through the issuer's operating results. With no primary shares being issued in our IPO, these same costs were charged to operating expense. Next, during the third quarter 2020, but in advance of our IPO, we issued a one-time stock bonus award, essentially to all Bentley Systems colleagues. These awards vested upon completion of the IPO and resulted in $15.4 million charged to operating expense. In addition, during the third quarter 2020, we initiated and approved a restructuring plan. You'll see this labeled as a realignment plan in our financial statement footnotes. As a result, we accrued and recorded a $10 million charge to operating expenses in the third quarter 2020, almost exclusively related to severance benefits. The plan impacts about 100 of our 4,000 colleagues and is not a cost savings motivated plan.
We're investing and will continue to invest any savings to stimulate several of our growth initiatives you'll hear Greg speak about momentarily. Our GAAP net income of $5.8 million for the third quarter was similarly distorted by these significant and unusual items. Accordingly, we also present our non-GAAP adjusted EBITDA, which normalizes for these and other items to facilitate useful analysis and comparison. Our adjusted EBITDA for the third quarter was $73.6 million, up 39% year-over-year and up 43% year-to-date 2020 versus last year. Continues to be an unusually profitable year for us as our cost savings initiatives undertaken as a precaution against pandemic uncertainties continue to generate the expected savings, contributing to unusually strong levels of profitability even beyond what we would have expected in an ordinary year and even beyond a concerted reinvestment of certain of these savings into the growth initiatives we're highlighting today.
Certain of these 2020 cost savings are temporary in nature, while we expect a portion of the savings to permanently benefit our cost structure. Our resulting year-to-date adjusted EBITDA margin of 32.5% is nearly 800 basis points of improvement over the same year-to-date period last year. Of course, due to the expected return of certain 2020 cost savings, this level of margin and the pace of margin expansion is not sustainable. However, we've historically delivered and expect to continue to deliver steady margin expansion from a normalized baseline of approximately 100 basis points per year for many years to come. We also present adjusted net income aimed at similarly normalizing outlier items to facilitate analysis and comparison. Adjusted net income for the third quarter was $51.4 million, up 31% year-over-year.
Before I turn to our capital structure, I'll offer a quick diagnostic on our unusually high 62% effective tax rate for the third quarter. This anomaly results from the non-deductibility of much of the $26.1 million in IPO costs we expensed during the quarter. In addition, as a public company, certain of our executive compensation for all of 2020 becomes non-deductible, and upon our IPO in the third quarter, we trued up our tax provision accordingly. These permanent differences were partially offset by some windfall deductions for stock-based compensation. However, the overall effect was an abnormally high third quarter 2020 tax rate. Turning to our balance sheet capital structure and liquidity, I highlight that we finished the quarter with approximately $138 million in cash and cash equivalents and $590 million of long-term debt. Net leverage based on trailing 12 months adjusted EBITDA was thus 1.8 x.
Our debt levels reflect borrowings incurred to support the payment of $1.50 per share special dividend declared and paid in the third quarter 2020. Adding this to our ordinary recurring quarterly dividend of $0.03, we paid $397 million in dividends in the third quarter 2020. Obviously, in the third quarter, we completed the initial public offering of our Class B common stock. Our secondary selling stockholders completed the sale of 12.4 million shares, including 1.6 million shares issued pursuant to the full exercise of the underwriter's overallotment option. As mentioned, there were no new primary shares offered in our IPO, and accordingly, the company did not receive any proceeds from the issuance of shares in the IPO. Regarding cash flow, our cash flow from operating activities year-to-date third quarter 2020 was $176 million, up about 49% compared to the same period last year.
We don't present a non-GAAP free cash flow metric, but it might be useful to know that our year-to-date Q3 2020 operating cash flows reflect the payment of approximately $4 million of the $26.1 million in IPO costs, with the remaining $22 million outflow expected to be reflected in the fourth quarter. In addition, the year-to-date third quarter 2020 operating cash flows reflect the payment only of about $400,000 of the $10 million in restructuring charges, with again, the remaining $9.6 million of these costs expected to be reflected in our Q4 operating cash flows, with some of this potentially deferring into the 2021 period. Turning now to our expectations for full year 2020. We expect total revenue in the range of $790 million-$800 million. This represents growth of 7.2%-8.6%. We also expect ARR growth of 7.5%-9% on a constant currency basis.
Given our strong profitability performance, we now expect adjusted EBITDA in the range of $250 million-$265 million, representing growth of 33%-41%. Here are some additional modeling details you might find useful. We expect full-year interest expense of approximately $7.5 million, which would decline in the event of any follow-on offering to the extent debt is reduced by the proceeds from the issuance of primary shares. We expect our effective tax rate for 2020 to be 23%-25%. Normalized for the unusual IPO-related activity I've mentioned, our effective tax rate is expected to be approximately 20%, a rate we target prospectively. We expect fourth-quarter diluted weighted shares outstanding of approximately 304.7 million and full-year diluted weighted shares outstanding of approximately 299.1 million.
We expect full-year outlays for capital expenditures, including minimal amounts for certain capitalized software development activities, to remain less than 2.5% of sales, as they've consistently been in the past. Overall, we're pleased with our performance. While we're cautious given our visibility into some marginal pandemic-induced softness in usage of our applications relative to last year, these are concentrated in certain sectors as Greg will further discuss, overall, our business and revenues have demonstrated relative resiliency, we continue to grow. We're prudently managing our costs and delivering improved profitability despite these macro headwinds and have steadfastly reinvested in future growth initiatives, including our normal acquisition cadence. We remain intent on continued investment, as clearly demonstrated by our just-announced $100 million corporate venture capital initiative, Bentley iTwin Ventures.
Greg, I think now you're going to comment on new business developments and provide some further commentary on tone of business.
Thank you, David, for covering the financial results and our range of financial performance expectations for 2020 as a whole, obviously based on our assessment for this fourth quarter that we're now almost halfway through. Had we instead been a publicly traded company for the whole of the year, we would have first provided such annual guidance along with the operating results for the previous full year in March. Though, as you know, I mean, we decline to ever provide financial guidance for individual quarters. In such a volatile year as this has been, I believe we would have been updating those annual expectations for you upon such occasions as this after the end of at least the second quarter, as well as now. I am momentarily going to cover at some length our current soundings regarding the tone of business.
This will also be more protracted than I expect to normally be the case because of the confluence of exogenous crosscurrents now unfolding around us. I want to share an abundance of the underlying observations to which we're managing so that under these volatile circumstances, you can be best prepared to draw your own interpretations. I will conclude with some plans and announcements for going forward. First, I'd like to cover our BSY corporate news since the IPO in September. There is more such news than will tend to be the case for most such quarters because we've just completed our annual Year in Infrastructure conference, and we tend to bunch up announcements to coincide with that. Of course, our conference format was all virtual for the first time.
This enabled us to expand attendance by a multiple of the 1,500 or so thought leaders that we could invite to our traditionally alternating Singapore and London venues. It also will enable you, through our bentley.com website, to browse and at least sample the presentations, which I heartily recommend. The conference is annually anchored by our Year in Infrastructure awards for advancements in going digital. You may recall that we featured all of the 2019 award winners in our S1 document. This press release announces the 2020 winners, the presentations for each of them are accessible now on demand. For me, the keynote highlight was my interactive discussion with Microsoft CEO Satya Nadella about infrastructure digital twins and our expanded strategic alliance initiatives to accelerate not only technical but also go-to-market collaboration.
We discussed why and how the events of 2020 to date have created imperatives for going digital to improve infrastructure projects, operations, and especially their resilience. Joint initiatives with Microsoft include our ProjectWise 365 with Microsoft Teams, city-scale digital twins for smart cities, and our iTwin platform integration with Azure Digital Twins and Azure IoT. The resonant theme of the Year in Infrastructure 2020 was that infrastructure digital twins are now practical and versatile, thanks to our iTwins platform and to our applications, which are increasingly iTwin-enabled. Rather than for Bentley colleagues to evangelize the benefits, this year, the digital twin presentations were made by user organizations covering a full range of use cases, as you see here and as you can watch online.
Our determined emphasis now is to get across that digital twins are an accessible technology and strategy even for smaller projects, assets, and organizations, and that it's easy to get started. To summarize the current point of departure in propagating digital twin aspirations across our user base, within the about 400 projects nominated by our accounts for these Going Digital Awards, it must be acknowledged that these are considered their exemplary projects by them rather than merely being representative. The following proportions credited these digital advancements in 2020 by comparison to prior years. 17% credited 4D construction modeling, up from 4% in 2019. This is uniquely enabled by our iTwin-enabled SYNCHRO software. 35% credited reality modeling, up from 28% in 2019. This is enabled by our ContextCapture software and cloud services, which produce 3D reality meshes from photographs and scans, typically from drones.
23%, up from 20%, credited a connected data environment by which we mean combining ProjectWise and AssetWise. 27%, up from 23%, credit iModels, used to semantically align deliverables. Often, iModels are behind-the-scenes enablers, which users don't see for digital workflows. 13%, up from 7%, credit asset performance modeling, in which modeling and simulation applications are reused during the life cycle of asset operations. Finally, 33%, up from 24%, credit digital twins, although that can be interpreted as aspirational in many cases, if not literally iTwins platform cloud services yet. At the top of the scale, with greatest potential benefits from infrastructure digital twins, we released at the conference our 2020 Bentley Infrastructure 500, the unique global ranking of the largest owner-operators ranked by their balance sheet infrastructure asset value net of depreciation. Shell is number three.
Shell's QGC in Australia was the winner in the juried category of utilities and industrial asset performance for their connected data environment, which over the last five-plus years has taken full advantage of ProjectWise and AssetWise, and now the iTwin platform for this world-class mega project and resulting assets. I find it particularly significant that this year Shell showcased their groundbreaking iTwin platform adoption for deepwater project digital twins, all of which has occurred during 2020 to date. This extends the PlantSight cloud service that we have co-developed with Siemens to include conceptual stage engineering. Shell's commitment during this austere period for them reinforces that industrial and resources owner-operators recognize that going digital is necessary to effectuate the step changes in economic effectiveness on which their future depends.
Turning now from our corporate news to the tone of business, we do see a perceptible change from the second quarter of 2020. During the first half of the year, we saw progressively by month and in turn for regions across the world, a reduction compared to the same period in 2019 in the number of days that our applications were used, and which generally returned to parity by the end of the quarter. This is the slide from our IPO roadshow presentation, which summarized that. In this next slide from that roadshow presentation, we expressed our directional expectations in degrees reflected by a different color for each sector about the potential direction for the trend of application days. In a nutshell, during the third quarter of 2020, these directions have eventuated.
Of particular significance for impingement on our overall revenue and ARR growth in the industrial/resources sector, where I have just noted that owner-operators seem generally to be continuing their initiatives in going digital to improve the operations and maintenance performance. That's, as we say, the OpEx of their existing assets. There has been a detrimental impact for the workload of some of our users of the decline in capital projects activity, we say CapEx, and that has shown up for the first time, presumably after working off backlogs in the third quarter. Because these affected EPCs, engineering, procurement, construction contractors, are individually very large enterprises, they have tended to be early adopters of our E365 daily consumption-based enterprise licensing, which indeed helps them to manage just such cyclicality and so provides us a competitive advantage in their software budget allocation.
Commensurately, in Q3, under E365, this has reduced our revenues from these EPC accounts and annualized has also reduced our related ARR relative to earlier in 2020, and even more so relative to 2019. Since this relates to macroeconomic conditions that are expected to prevail for longer than the direct pandemic consequences, we want to particularly bring this to your attention.
On the other hand, we do not now see the same downturn in the majority of our business as in industrial resources CapEx. To help you more fully weigh and understand our interpretation of these crosscurrents, which seem to have never been harder to predict, and the impact of which we think depends on our business mix, both with respect to macro conditions by infrastructure sector and with respect to our different commercial contract sensitivity to short-term consumption, I will now provide many more observations about our software usage than we would normally find it useful to report. Thus, we do not undertake to update these measures in future.
Before breaking out each of the three infrastructure sectors that we measure, keep in mind that about a third of our revenue is from accounts who can't be reliably classified into any of these, generally because they are diversified engineering firms who variously work in all these sectors. It is our belief and experience that the sector weightings within all of their businesses tend to correspond in aggregate to BSY's overall proportions, and thus to reflect the same usage patterns that I will next share. To start with respect to the same period in 2019, our overall application days declined in 2020's third quarter by about 4%, after being essentially level year-over-year in 2020's second quarter, when late-arriving usage logs were factored in. Application days for our accounts within the commercial/facilities sector, that's about 8% of our total, were down most deeply from 2019 at about 9%.
For commercial/facilities CapEx accounts, engineering, architecture, and/or construction contractors, application days were down about 14%. Accounts in this sector do not tend to be on E365 contracts. Application days for our accounts within the industrial/resources sector, which is about 19% of our total application days, were down from 2019 about 8%, after being down from 2019 in 2020's second quarter by about 4%. For industrial resources CapEx accounts, EPCs, the largest of whom tend to be on E365 contracts, application days were down about 11%. Application days for our accounts within the public works/utilities sector were down from 2019 about 2%. For public works/utilities CapEx accounts, application days were up about 4% from 2019. For public works/utilities OpEx accounts, owner/operators, application days were down about 7%, representing more than half of our overall decline in application days from third quarter 2019.
We believe this is because many of these organizations reduced their workforce schedules to effectuate partial furloughs and suspended field work. Many German utilities, for instance, adopted half-time work in conjunction with pandemic lockdowns. In general, although obviously no one seems able to authoritatively predict future pandemic severity, we do not believe these policies will be extended or repeated, and we are considerably less concerned about an enduring overall downturn in public works/utilities. Application days for our products, rather than for our accounts, that are specific to public works/utilities, and this is about 24% of our total application days for these products, in the third quarter were up from 2019 by about 3%. While modeling and simulation applications still represent the majority of our revenues, our collaboration systems have consistently grown faster.
In the public works/utilities sector, users of our mainstay ProjectWise Design Integration in the third quarter of 2020 were up from 2019 by about 4%. During 2020, accelerated by work-from-home virtualization, overall users of ProjectWise through our Azure cloud services, including Design Integration Managed Services, as well as our native SaaS ProjectWise 365, have increased about 23%. Overall ProjectWise revenues for 2020 year-to-date through the third quarter are up from 2019 by about 17%. New business for AssetWise faces the challenge of completing enterprise-scale procurements during travel lockdowns. Usage of AssetWise in the third quarter of 2020 increased over the same period in 2019 by about 4%. Overall AssetWise revenues for 2020 year-to-date through the third quarter are up from 2019 by about 6%.
Finally, as to our new iTwin platform, ARR has grown throughout 2020, consistent with a reasonable chance to reach our goal of 8 figures by year-end. Significantly, I think, as to new business in general, through the third quarter of 2020 year-to-date, we have achieved 89% of our overall new business quota, which after the pandemic onset, we set and seasonalized to be the same as 2019. Within the new business quota, we double the weight for our cloud services, ProjectWise, AssetWise, and iTwins. To sum this up for 2020, during the third quarter, because of the E365 application days downturn led by the EPC accounts, which decremented our ARR and to that same extent, therefore, new business, we fell behind the year-to-date 90% of 2019 new business pace.
We hope to catch back up to that pace during the fourth quarter and thus the full year. There generally appears to be sufficient pipeline to chase. Obviously, this can't be taken for granted. What I can say is that we are quite confident in the relative resilience of infrastructure engineering and BSY and about our anticipated future. To speak now about going forward, as we announced last month, our going public process has helped us in recruiting world-class software executive talent. While I believe that we have self-developed an excellent pipeline of executives internally during our 36 years as a privately held company, we obviously were not able to internally develop public company experience.
I am pleased to say that we welcomed over this summer, first, Nicolas Cumins, as our new Chief Product Officer from SAP, where he had charge of SAP's marketing cloud businesses, and over time, also SuccessFactors and Concur, all of which we fully utilize. Nicolas also has entrepreneurial experience. He's based in Munich, Germany. The majority of our colleagues are in our product advancement group, reporting up to Nicolas, who reports to me. We welcomed Katriona Lord-Levins, our first Chief Success Officer, who led user and enterprise success initiatives at Autodesk over multiple decades with excellent results that we ruefully observed firsthand. Kat, based in San Francisco, is inaugurating our user success and enterprise success functions, reporting to me.
We welcomed Chris Bradshaw as our new Chief Marketing Officer, coming most recently from Blue Prism, but having been at Autodesk through 2017 in senior roles, including Chief Marketing Officer. Chris is based here on the East Coast and reports to Gus Bergsma, our Chief Revenue Officer. Our restructuring charge in third quarter 2020 for colleague severance is not to cut our costs, but rather the opposite. We are enabling Nicolas, Kat, and Chris to reprioritize new talent needs as we resolutely reinvest a large portion of our 2020 cost savings into our public company-era growth strategies, faster accretion in existing accounts, force multiplying inside sales to better reach small and medium-sized accounts, focusing on Asia, and joint cloud services with Microsoft and Siemens.
In addressing our final two growth strategies, programmatic acquisitions and digital integrators, we announced at our Year in Infrastructure conference that The Cohesive Companies, a digital integrator which we wholly own but manage independently within our Bentley Acceleration Initiatives, acquired the U.K.-based going digital advisory firm, PCSG. We are increasingly thinking of our Bentley Acceleration Initiatives as a second segment afforded by our increased resources as a public company, and today, we are excited to announce a further initiative. We are committing to invest over a period of years, $100 million in corporate venture capital through the new Bentley iTwin Ventures Fund, which in turn will invest in entrepreneurial and emerging companies participating in the infrastructure digital twins ecosystem being catalyzed by our iTwins platform.
The first such iTwin Ventures investment, representative of many more anticipated, has been in FutureOn AS, the Norwegian developer of sub-sea digital twins, including for Shell. That closes the loop on our prepared remarks, and we look forward to your questions.
Thank you, Greg. We'll start by taking a question from Joe Vruwink from Baird.
The commentary on industrial natural resources was really helpful. You've also seen some comments from some of your peers that maybe things started to improve a bit in the month of October, specifically with EPC customers in natural resources. I'm wondering if that was evident in your book of business at all.
Joe, thanks for the question. In the fourth quarter so far, everything has tended for the better. We talk about new business. New business for us is what we're selling. It's the accretion in ARR. It's not our whole ARR, it's just on the margin, how much it grows, and if it doesn't grow, that's a detraction from our new business. That's looking better than the third quarter looked at any point. Usage has turned in the right direction. I don't think that's true, however, for the EPCs, for the industrial resources firms. They use our software for really large projects. If you look at these EPCs and the ones that are public companies in their own reporting, they're down 15%-30% on the previous year.
Anyone on the call would be as well able as me to predict when capital projects come back in the process plant and energy sector. We don't see it yet, I'm afraid. Everything else more than makes up for that, thank goodness.
Okay, great. That's helpful. One other thing that I think is interesting, and it seems to be occurring maybe in this theme of divergence or crosswinds, but the backlog work with civil engineers seems to be holding up comparatively much better than civil contractors. It would suggest that the planning is still going on, project work is out there. There's still a good pipeline if you're a civil engineer. I suppose that speaks to your resilience, but how does that type of environment maybe speak to the type of budget you're likely to see in 4Q with your civil public works customers, and the degree of visibility that provides for your 2021 ARR performance?
Well, I think what you say is quite observable. It happens I was just on a panel during the past week with AEC Advisors. They do surveys of all of these firms, the world over. AEC for them is not construction, but rather consulting, as they say. These are the infrastructure engineering firms the world over, and they survey most of them. They believe they have most of the dollars, and the result of the survey for 2020 is that the firms of late expect to grow slightly in 2020. Now, when AEC Advisors weighted it then, not average or median, but by firm size, the growth came out to zero for the year.
Compared to what's gone on with GDP and the economy, it is a relatively resilient sector and, of course, is spending more on going digital, which was the subject of the panel that I was on. In this country, and remember, the United States is a little under half or a bit under half of our revenue, we look like we'll be welcoming a new administration. It's likely that we will join the rest of the world in focusing on infrastructure investment as a means of fiscal stimulus, because it's thought to be that which has the most enduring and best return on investment. Much of the world has already determined to do that, and is a part of the reasons we're encouraged about 2021, in particular. In the U.S., we think that factors in.
In fact, these firms that I'm talking about are expecting to grow in 2021 about to the extent they have grown annually prior to 2020. As I say, 2020 is a flat year, which looks pretty good for the population you're talking about.
Great. Thank you very much.
We'll next hear from Brian Essex from Goldman Sachs.
Brian, you're muted, I think. There we are.
Can you hear me?
Hey, Brian. Yes.
Oh, great. Thanks, and congrats on the results and emerging as a public company. I was wondering if maybe we could touch on, you've spoken previously about an initiative to penetrate some of the smaller customers on your platform. Obviously, the larger customers are pretty substantial in terms of their contribution. Any thoughts that you'd like to outline in terms of those plans for the other 33,000 or so smaller customers or mid-market customers on your platform, and the trajectory that you might expect from penetration of those accounts?
Well, it's very much our priority for 2021. It has everything to do with the reinvestments we're making now of our cost savings. We want to be going digital in the way that we develop e-commerce capabilities. We didn't have an e-commerce capability until very recently. Our new Chief Marketing Officer is very much dedicated to that, and our remix of resources will help us with direct engagement, and increase in marketing effort and effectiveness as a public company. We have a very significant appetite to improve our penetration there. Some of our initiatives are underway, more so will be in 2021. The engineers in the smaller firms need the same superior software as the engineers in the larger firms. It's just late in the game that it's getting our full attention, but it has that now.
I might just add, Brian, that as a proof of concept for us and a head start, one of our Acceleration Initiatives, Virtuosity, where we're targeting individual practitioners, which tend to be, of course, the smaller medium-sized enterprises, through an e-commerce platform and embedding expert services with them. Is actually performing as expected, and something that's going to serve to light the way for us as well, going after these small, medium-sized accounts.
Great. That is helpful. Maybe just to follow up. In terms of expectations of potentially, as we saw yesterday with news of the vaccine progress, when you think about the scenario where the economy might start to open up a little bit more, particularly with the respect that we have seen a little bit of pressure on utilization from some industries that have been pressured in this macro environment. Are there any kind of key economic or macro indicators that you typically look at to anticipate what the impact or the magnitude of reopening might have on utilization on your platform and how meaningful that might be? Kind of going into 2021.
I do want to emphasize that other than in industrial and resources and perhaps commercial and facilities, all that's relatively small for us. That's why we haven't dwelt on that. In public works and utilities, any decline in application days we think is kind of institutional about scheduling changing. I don't think application weeks or months would be down in some of these organizations that literally have responded to lockdowns with changing their work schedules. The most significant aspect of reopening are the infrastructure plans around the world. If I look at Asia in particular, we say focus on Asia because it's been such a reliable source of new business growth for us. This year so far, it's not been better than last year as it normally is. A lot of that depends on the fourth quarter.
The fourth quarter is always the strong quarter for Asia. Countries like China has just done a new five-year plan that emphasizes digital plant and digital cities and environmental initiatives and so forth. All of which are great for us and open up new opportunities. Australia, their new budget has a 10-year infrastructure plan for transport infrastructure where we're particularly strong. The same is true in most countries in the world. Anyone can handicap it, likely to be also true in the United States, we think. Focus on energy adaptation also is good for the work of infrastructure engineers. If I come down to another factor that could be helpful looking forward, it is that in our own country, to be less antagonistic about globalization will serve us well, for instance, in China.
These factors are coming together to, as you say, make us rather optimistic about 2021.
Great. Very helpful color. Thank you.
We'll next hear from Brad Sills from Bank of America.
Oh, great. Hey, guys. Thanks so much for taking my question here. Maybe just a follow-up to the comments you just made, Greg, on China. I think it might be helpful to talk about how Bentley's positioned. What's your presence in China? You mentioned this five-year infrastructure investment plan. How is Bentley positioned to participate in some of these projects? Thank you.
We're really well-positioned in China, fully invested there. It's only about 5% of our revenues now, but a lot higher portion of our potential. The question in China, challenge for us, well, there's an immediate challenge that there are not Azure cloud services. We're engineering around that and Chinese accounts need to do their own private cloud stack and so forth, which they can and have accomplished with their iTwins ambitions and so forth. We really have to scale up to that opportunity. The cities of China, for instance, have assigned industrial design institutes to be responsible for their BIM and digital cities strategies, and we often have experience with them already. There's just a lot of headroom in China. Over 2020 to date, it hadn't gotten easier. There are some export controls now that slow things down.
Generally, trade tensions have not been helpful. I think all changes for the better as far as that. We really are very well positioned in China in terms of talent and resources. There is a very good pipeline there, and we stand a good chance, we think, of reaching last year's new business level by the end of the fourth quarter. We're applying ourselves assiduously to do that.
Great. Thanks so much. Then, David, one for you, please, if I may. Gross margin came in quite a bit better than where we were modeled. Can you talk a little bit about the puts and takes there? What's driving the upside in the gross margin, at least to our model?
Sure. I assume you're talking about adjusted for the-
Yes
the non-recurring items that I've highlighted. Yeah. Part of it is Cohesive. The acquisition is a strong margin-performing services business for us. That's a bit of a lift. As I've said, the other piece of our cost of goods that goes into that gross margin is our cloud costs, our cloud delivery costs. We continue to find efficiencies in delivering our cloud services. That's contributing to some improvement there as well.
Great. Thanks so much, guys.
We'll next hear from Matt Hedberg from RBC.
Hey, guys. Good morning. Thanks for taking my questions. I wanted to ask about E365. It's obviously, as you alluded to, Greg, been a great competitive advantage for you guys over the years. When you think over the next several years, what's the right way to think about the pace of adoption of E365, in your base? What would that mix look like longer term? When a customer moves to E365, what typically happens to their spend?
Really, E365, you might say, "Oh, Greg, you've taken on this risk. Are you so sure about that?" Our predecessor ELS program also had a mark-to-market based on consumption, but was only once per year, so it was lagged. Now we don't have that lag, but it's so much preferable to the account for us to share the risk and cyclicality, and we're prepared to share that risk. It isn't a large risk in proportion to our huge book of annual resets, if you like. The reason we prefer the E365 format, despite taking on a bit more immediate volatility, is it includes the success plan component, where we really competitively distinguish by having more civil engineers and structural engineers and geotechnical engineers who can help in digital workflows that our accounts want and in taking full advantage of the software.
We embed them, and we charge for that in the application day charge. Typically, the spend does go up somewhat. From our standpoint, that's not the opportunity we're seeking. It's the accelerated accretion from application usage growing faster, when in effect, we can virtually embed our experts to help the software get better used. With our cloud services for our applications, we can see what functions are being used and not used, and be helpful in direct engagement, going digital ourselves to embed our Success Forces and better enable them. That's what is, for us, the motivation. We suppose it will take another several years, but we'll end up migrating the ELS book to E365.
A year like this doesn't deter us because in the combination of forces that creates economic challenges and so forth, we're pretty well able, as you see, to absorb this additional volatility on the margin, and we have an appetite to do more of it.
That's great.
Just to frame the scale of that for you, Matt. About a third of our ARR is these enterprise scale subscriptions. We're give or take about halfway through, in terms of the ELS to E365 migration.
Super helpful, David. Actually, just one more quick one for you on the margin side. Services revenue, I believe, has been about 8% of your year-to-date revenue. Can you talk about the long-term trend of services revenue in your overall mix, and do you think you're going to be able to leverage the channel a bit more for some of the lower margin services work?
Well, the better job we do with our ProjectWise and AssetWise offerings, the less services would be required in the scheme of things. The future of services for us is the digital integrator opportunity to literally help with the data in the case of an infrastructure digital twin. When you open up what's been dark data, so you can bring the ET along with the IT and the OT for this evergreen digital twin opportunity, you see there needs to be improvement in data quality. It's a good job for engineering firms. When you say channel, I encouraged in the panel that I described, I said the future for the engineering firms is to be the creator and the curator, because it's never done. You need to maintain the digital twin up to date and to be in the business of the analytics and the benchmarking and so forth.
Which we don't wish to be in those businesses. Those are opportunities for digital integrators, and the best channel for us would be the engineering firms themselves to bring the digital twin opportunity to the owner-operators. That would be the long-term way in which that occurs. As you know, we need to incubate some digital integrator experience. We have that under our Bentley Acceleration Initiatives, but we sort of compartmentalize that. We manage that differently than the software business. Yes, we do not wish to grow professional services within our core business. We wish that to be an opportunity for others, but especially for engineering firms who need to improve on their business model of selling their hours and wish very much to do so. As they virtualize this year, they have much more receptivity and motivation to do that than ever.
Thank you.
Great. We'll next hear from Matthew Broome from Mizuho.
Greg and David. During the quarter, you announced an expanded relationship with Microsoft. I guess more broadly, how did your strategic partnerships and ventures with the likes of Microsoft, Siemens, and Topcon perform during the quarter?
Well, with Microsoft and Siemens both, we sort of transformed our relationships in each case because we've, for the past few years, focused on technical collaboration, and now that has blossomed into these, what I call digital co-venture, these cloud services that are ready for market, and we need to focus on joint go to market where either of us can sell them. That's not just one initiative, but a bunch of initiatives which happen at the same time to be occurring with each of Microsoft and Siemens, transforming from merely technical collaboration to also go-to-market collaboration. With Topcon, our joint venture is Digital Construction Works, and that's not the best place to be at the very moment, and needs probably to diversify itself.
What we're working on, focusing on heavy civil construction, where there are particular opportunities for, we call it constructioneering, where a design build approach is considered the design and construction at the same time and innovate in that respect.
Yeah. The DCW Topcon joint venture is just right in the crosshairs of capital projects delays and implications. I think we picked up about a half a million dollar loss in the quarter relative to our equity method investment in that joint venture. The focus there has shifted from revenue generation in the face of some pretty severe headwinds towards demonstrating some referenceable successes and building out their capabilities.
Okay. That's definitely helpful. Thanks. I'm just interested if you could maybe provide a little more color on the sort of regional trends that you saw playing out during the quarter, especially in Europe?
Well, Europe has been a bunch of separate regions. It's tough to generalize. I will say for us, EMEA, we include Middle East. Middle East is still depressed in terms of application usage really hasn't come back as has most of the world after the lockdowns, which we think is now not regional, but a matter of the sector, the industrial resources sector, affecting that. We've done all right year to date in ARR in Europe. New business, I think, is regaining momentum. If I just turn my attention to Asia for a moment. In Asia, there are economies that depended most on face-to-face travel. Even for procurement processes, they weren't ready for being digital. In India and other parts of South Asia, there were not the cultures of working from home and laptops and so forth. All that has been surmounted now.
In general, I think all of infrastructure engineering has virtualized successfully, is better off for it. Everyone can work on projects anywhere with collaborators anywhere, and I think they're energized by it and will never give up that level of virtualization with the collaboration tools that they've learned and honed. Europe, it's tough to generalize, but has still some improvement to return to. Is there more to say on Europe?
No.
Okay, perfect. Thanks very much.
We'll next hear from Jason Celino from KeyBanc.
Hey, guys. Can you hear me all right?
Yes, we can, Jason.
Great. I really appreciate all the color on tone and the different businesses, but looks like renewal rates are still pretty strong. How should we think about the relationship between some of these usage trends and revenues for the non-E365 business?
I'm going to let David add more, but they're unrelated. When we count application days, that has a direct bearing on E365 revenue and therefore on ARR, and therefore on what we call new business. I realize new business is our own quota. It's not something we measure for you necessarily, but we're not far from last year's new business, and we can catch up even, we think. All of that is impinged upon by application days that's on E365. Otherwise, it makes no difference. For the great bulk of our renewals that are not based on application days, there isn't a direct impact and renewals are occurring and accounts are not expecting to spend less and I think are all expecting to spend more.
Yeah. I would throw into the category of application day usage that's already manifest in our ARR and our revenues. These are our term license business. The monthly and quarterly term licenses obviously would have been picked up in the results you see already. With the majority of our business being annually renewing, the implication is a matter of how prolonged is the impacts of the pandemic and if it carries on for another year, you'll start to see that in some of the annually renewing contracts.
Okay, great. Maybe a quick follow-up on some of the operating margin? If I look at adjusted operating margins of 34%, highest watermark in company history. Any particular thing about expense management? How sustainable are these margin levels?
Yeah. A good question. I'm not here to provide our 2021 margin guidance, but I would like to at least comment directionally on some of the trends you're seeing and where we're headed. As I said, 2020 is a fairly opaque, unusual year in terms of our profitability, because of some pretty significant cost savings initiatives. Even backing up from there and looking at 2019 as a normalized baseline to start thinking about margins, you would've expected us to see our normal 100 basis point or so embedded improvement just from efficiencies and scale. We have pretty significant cost savings in 2020, in the neighborhood, Jason, of $40 million. At least half of that is going to return to our cost structure in 2021.
The other thing I will point out for 2021 relative to 2020, is that effective October 1st, we're taking $30 million annually of compensation to executives that previously was cash-based, and will be stock-based going forward. We're not seeking or expecting margin expansion credit for that. I just want to make sure it's understood that in the fourth quarter, you'll see $7.5 million. For the fourth quarter 2020, moving from cash-based to stock-based, in 2021, a full year of $30 million, which is give or take 3.5 margin points. We'll add that back when we're reporting adjusted EBITDA just to be benchmarked and aligned to our peers. Of course, in 2021, you can also expect normal scale efficiencies of 100 basis points or so coming back to improve our margins.
We're committed to do that, and we demonstrated ability to do that regularly and annually going forward.
Okay, great. Appreciate it. Good going.
For our last question, we'll hear from Gal Munda from Berenberg.
Yeah. Hi, good morning. Thanks for taking my questions. The first one is just, I really like the Shell example you gave, Greg. I was wondering, if you could talk a little bit more about the usage. What happens between the more advanced owner operators like Shell versus an average customer that you have on the owner/operator side today, maybe specifically, how much of the uplift then there is when they did adopt other part of the technology, all the way to the digital twin versus just the modeling side?
Well, Shell, of course, is a private company. It is in the private sector. Wouldn't be so surprising if innovation happens faster in the private sector, as in the case of Shell than in the case of publicly owned infrastructure, with departments of transportation and government management. The great thing about selling to owner/operators in public works and utilities is they don't compete with one another, so they're glad to share advancements and so forth. The single thing that I'm most pleased by this year is at the Year in Infrastructure conference. Again, I hope you will all watch at least some of the sessions there. It is the fact that you have some of these owner/operators wanting to present and share with the affinity group of other owner/operators around the world. It's one of the reasons it's so important to watch the Chinese.
We sometimes say that in infrastructure, who does R&D? That in China, they do. They wish to adopt everything as quickly as possible in terms of digital twins, and it's great to see those examples. When we come back to the United States and our changes that we may expect here, we hope we may be able to influence an infrastructure spending bill so that it would encourage digital twin approaches to improve the economics for the whole life cycle. I think that's not out of the question. But generally, what holds us back is levels of ambition. Levels of ambition in going digital have discernibly improved. That's my take on our Year in Infrastructure this year and the owner/operators themselves sharing these success stories.
I know we're at the end of our time, and we've used more time than usual because of this notion that we had to include our introduction to the company and then this tone of business. In conclusion, I would like to thank our Bentley colleagues, the 4,000 whom are now all shareholders. I want to thank all of them for their sacrifices to start with, and then commitment and resourcefulness in this year so far. We all look forward to 2021, as I've said. First, let's deliver, which is in our sights, a terrific end to the year of resilience, 2020. Greg, was I supposed to say that now or are you going to sign us off?
Yep. Thank you everybody. Thanks, Greg. Thanks, David. We'll see you next quarter.