Cogent Communications Holdings Inc
NASDAQ:CCOI
Earnings Call Analysis
Q4-2023 Analysis
Cogent Communications Holdings Inc
The company celebrated the strategic acquisition of Sprint's business on May 1, 2023, marking a significant expansion in their network, customer base, and business scale. This milestone event raised the annual revenue rate above $1 billion due to a substantial intake of enterprise customer relationships and valuable network assets. The deal, completed for $1, included a fiber network appraised over $1 billion and facilities convertible to data centers. Impressively, this agreement also involves $700 million in payments from T-Mobile to cover operational costs attributed to enterprise customers, promising to enhance future cash flows. Amidst these developments, the company increased shareholder value by boosting the quarterly dividend from $0.995 to $0.965 per share, continuing a tradition of consistent dividend growth.
Financially, the company achieved a strong year, ending with quarterly revenues of $272.1 million and annual revenues of $940.9 million. The EBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) as adjusted stood at $110.5 million for the quarter and $352.5 million for the year. These figures translate to healthy EBITDA margins of 40.6% for the quarter and 37.5% on an annual basis. The company's debt ratios also experienced positive movements, with a gross debt to EBITDA ratio of 4.07 and a net debt ratio of 3.75, marking an improvement from previous metrics. In terms of operations, network traffic saw a 22% year-over-year increase. The firm aims to realize annualized cost savings of $220 million over three years, with hopes to even exceed this target—demonstrating their commitment to operational efficiency and profitability.
In a bid to diversify and enhance offerings, the company made strides in selling wavelength services to a broader customer base, which now includes Sprint's customers. Achievements here include sales in 65 locations and anticipated offerings in over 800 North American data centers by the end of 2024. Infrastructure growth following the Sprint acquisition is notable, with increases in both intercity and metropolitan fiber miles, and the conversion of facilities into data centers, thus strengthening the company's physical presence and service capabilities.
The company has recognized additional intangible assets through the Sprint acquisition, including IPv4 Internet addresses valued at $458 million, contributing to a bargain purchase gain of $1.4 billion. This intangible asset underscores the deal's significant value and future revenue potential. The corporate segment generated 46.5% of this quarter's revenues, with net-centric and enterprise customers contributing 34.2% and 19.2%, respectively. Acknowledging the fluid economic landscape, the company maintains a cautious stance on corporate revenues while noting a significant increase attributed to Sprint's inclusion.
Following the addition of Sprint's personnel, the company's sales force experienced a 20% year-over-year increase in size. Integration efforts are underway to train new enterprise reps on the company's sales processes, signifying an investment in human capital to leverage the expanded customer base fully. On-net revenues soared by 14.5%, reaching $518.6 million, driven by the sale of 100 and 400 gigabit connections, indicating an elevated average revenue per user (ARPU) and a shift toward higher bandwidth solutions.
In view of the comprehensive integration post-acquisition, the company projects a steady annual revenue growth of 5% to 7%. Furthermore, EBITDA margins are expected to incrementally improve by approximately 100 basis points each year, signaling confidence in the company's economic moat and efficient cost management practices. Strategic priorities include enhancing product offerings, such as wavelengths services, while concurrently optimizing the sales force and streamlining the vast infrastructure acquired through the Sprint transaction.
You don't have any saved screeners yet
Good morning, and welcome to the Cogent Communications Holdings Fourth Quarter and Full Year 2023 Earnings Conference Call. As a reminder, this conference call is being recorded and it will be available for replay at www.cogentco.com. A transcript of this conference call will be posted on Cogent's website when it becomes available. Cogent's summary of financial and operational results attached to its press release can be downloaded from the Cogent website.I would now like to turn the call over to Mr. Dave Schaeffer, Chairman and Chief Executive Officer of Cogent Communications Holdings.
Yeah, hi. Good morning. Welcome to our earnings call for the fourth quarter of 2023 and full year 2023. I'm Dave Schaeffer, Cogent's Chief Executive Officer. With me on this morning's call is Tad Weed, our Chief Financial Officer. Hopefully, you've had a chance to review our earnings press release. Our press release includes a number of historical metrics that we present on a consistent basis each and every quarter.Now for a quick summary of our results. We closed the acquisition of the Sprint business on May 1, 2023. This transaction significantly expanded our network, our customer base, and materially increased the scope and scale of our business. Our annualized revenue run rates are now in excess of $1 billion. We acquired a large number of enterprise customer relationships. These customers are typically larger than our Cogent legacy corporate customer base. We also acquired a significant network comprised of owned fiber and owned facilities, many of which are being converted to data centers. We acquired a network with an appraised value substantially above $1 billion for $1. We are repurposing the acquired fiber network to be optimized for the sale of wavelength services.We received a total of $700 million over time from T-Mobile to offset the operating losses of serving enterprise customers. $350 million of these payments will be made in the first year at $29.2 million per month, and then $350 million of payments will be spread out over the next 42 months of $8.3 million per month. We remain optimistic about the cash flow capabilities of our combined operations. Our recent results show that we have achieved immediate and substantial savings in multiple areas, many of which have exceeded our initial expectations. We anticipate additional cost savings from our current run rates.Our combined Cogent business had a very good quarter and a very good year. Our total revenues for the quarter were $272.1 million and $940.9 million for full year 2023.Our EBITDA as adjusted for the quarter was $110.5 million and for the full year was $352.5 million in 2023. Our EBITDA as adjusted margin was 40.6% for the quarter and 37.5% for full year 2023. We received 3 payments totaling $87.5 million from T-Mobile this quarter and a total of 7 payments in 2023 totaling $204.2 million.Our gross total debt trailing 12-month EBITDA is adjusted and our net debt ratio significantly improved in the quarter. Our gross debt to trailing 12 months EBITDA as adjusted ratio was 4.07 at year-end and our net debt ratio was 3.75, a substantial improvement from the 4.23x in the last quarter.Our network traffic increased sequentially by 7% and was up 22% year-over-year. We have a number of areas in which we expect to continue to execute on cost savings. We're in the process of realizing savings and synergies over a 3-year period that will result in an annualized savings of $220 million. We anticipate achieving additional SG&A savings and other cost and revenue synergies over this -- over the next several years, hopefully exceeding that $220 million target. Our recent progress in achieving these savings is very encouraging and we do intend to surpass the targets that we have laid out.Our sales force productivity last quarter was 3.6 units installed per rep and 3.3 units per full-time equivalent this quarter. Our sales rep productivity has been substantially impacted by the enterprise customer reps that joined us from the Sprint business. These new enterprise reps are continuing to receive training on Cogent sales processes and have not yet reached their full productivity. Now for the size and scope of our sales force, in connection with the Sprint acquisition, we hired a total of 942 employees. As of today, 742 of these employees remain employed with Cogent. During the quarter, our total sales rep count increased by 20 or approximately a 3% sequential increase. For full year 2023, our total sales reps increased by 109 or a 20% increase substantially ahead of our normal rate of sales force growth catching up for some of the slower growth that occurred throughout the pandemic years. We ended the year with 657 sales reps, of which 620 were counted as full-time equivalents.Now for some comments on our wavelength and optical transport services. In connection with our acquisition of the Sprint business, we are expanding our product offering to include wavelengths and optical transport over our newly acquired fiber optic network. We are selling these wavelength services to existing customers, as well as customers acquired from Sprint. These customers require dedicated optical transport connectivity without the capital and ongoing expenses associated with operating their own network. We have sold wavelengths to date in 65 locations, many of these have shorter provisioning cycles. We have connectivity and capability to sell wavelengths today in an additional 285 locations with longer provisioning cycles. By year-end 2024, we expect to be able to offer wavelength services in 800 North American carrier-neutral data center locations with substantially shorter provisioning times.Our footprint expanded materially with the acquisition of Sprint. We added 18,905 route miles of owned intercity fiber and 12,000 excuse me, 1,257 route miles of owned metropolitan fiber to our network. We also added 11,400 route miles of intercity IRU fiber and approximately 4,500 route miles of metropolitan IRU fiber to the Cogent network. We are in the process of rationalizing these acquired IRU fiber agreements as their contractual terms allow us to exit them. To date, we have also reconfigured 22 of the acquired Sprint facilities into data centers and we added these new data centers to the 1,558 carrier-neutral data centers that we operate and brought the total of Cogent operated data centers to 77, which today have 157 megawatts of power. We are in the process of converting an additional 23 Sprint facilities into Cogent data centers and optimizing and rationalizing our data center footprint.During the quarter, we returned $46.4 million to our shareholders with our regular quarterly dividend. We paid 4 quarterly dividends in 2023 totaling $181.7 million, or $3.76 per share. We expect the tax treatment for these dividends are generally treated as completely return of capital, so therefore 100% of those dividends will be treated on a tax-deferred basis. Our Board of Directors, which continues to evaluate our growth in cash flow and the capabilities of our team to execute against our opportunities, inclusive of the Sprint acquisition, increased our quarterly dividend yet again by $0.01 a share sequentially, raising our quarterly dividend from $0.995 per share per quarter to $0.965 per share per quarter. This increase represents the 46th consecutive sequential increase in our quarterly dividend and a 4.3% annual growth rate in that dividend.Now, for a couple of comments around our long-term targets. Now that Cogent is fully integrated and combined with the Sprint business, we anticipate our long-term average revenue growth to remain between 5% and 7% annually, and we expect our EBITDA margins as adjusted to increase by approximately 100 basis points annually. This will be impacted in the short-term by the step down in payments from T-Mobile. Our revenue and EBITDA guidance targets are intended to be multi-year goals and are not intended to be used as quarterly or annual specific targets. Our EBITDA as adjusted and leverage ratios are impacted by the $700 million payment stream that we receive from T-Mobile. Beginning in 20 -- May of 2024, these payments will step down from $29.2 million per month to $8.3 million a month and remain in place for the subsequent 42 months. The reductions will impact our future EBITDA as adjusted leverage ratios and will impact our ratios in the third quarter of 2024 on a trailing 12-month basis.Now, I'd like to turn it over to Tad to read our safe harbor language and provide some additional details, and then I will jump back on to address some additional operating metrics.
Thank you, Dave, and good morning, everyone.This earnings conference call includes forward-looking statements. These forward-looking statements are based upon our current intent, belief, and expectations. These forward-looking statements and all other statements that we made on this call that are not historical facts are subject to a number of risks and uncertainties and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. Cogent undertakes no obligation to update or revise our forward-looking statements.If we use non-GAAP financial measures during this call, you will find these reconciled to the corresponding GAAP measures in our earnings releases that are posted on our website at cogentco.com.So, some comments on the accounting for the Sprint acquisition which was very complex, frankly. In connection with our accounting for the acquisition, we recorded a total gain on bargain purchase for the year of $1.4 billion, or almost $27 per share for the year. Included in that $1.4 billion gain is the discounted present value of the $700 million IP Transit services agreement with T-Mobile. During the fourth quarter and in consultation with our auditors and valuation specialists, a Big 4 accounting firm, we recorded -- both Big 4 accounting firms, we recorded an additional intangible asset for $9.9 million of IPv4 Internet addresses that we acquired in the Sprint acquisition. These IPv4 addresses have an indefinite useful life and are not being amortized. This asset was recorded at $458 million or an average of about $46 per address. Because of the novel nature of this asset and the fact that the transaction has already resulted in a material bargain purchase gain prior to recording this asset, we recorded the asset after consideration of the appropriate valuation approach. The net after tax impact of recording the IPv4 Internet addresses and other valuation adjustments that we made this quarter resulted in a net additional gain on the bargain purchase of $254 million that we recorded in the fourth quarter.The acquired network, including owned real estate assets, fiber routes, right of way agreements, network equipment and the IPv4 Internet addresses have been appraised by a Big 4 accounting firm at a valuation -- total valuation of $1.4 billion. The total fair value of the net assets acquired, so net of liabilities, was $800 million and including the net present value of the consideration to be paid to us by T-Mobile of $600 million while -- on the discounted value and the $458 million of IPv4 Internet addresses. Again, the total acquisition resulted in a $1.4 billion bargain purchase gain. These amounts are subject to additional adjustments through 1 year from the closing date which will be May 1 of 2024.Some comments on corporate and net-centric revenue and customer connections. We analyze our revenues based upon network connection type, which is on-net, off-net, wavelength services and non-core services, and we analyze our revenues based upon customer type and we classify all of our customers into 3 types: net-centric, corporate, and enterprise. Our corporate business continues to be influenced by real estate activity in central business districts. We continue to remain cautious in our outlook for our corporate revenues given the uncertain economic environment and other challenges from the lingering pandemic effects. Our corporate business was 46.5% of our revenues this quarter and our quarterly corporate revenue increased year-over-year by 47.6% to a total of $126.6 million from the fourth quarter of last year and increased sequentially by 5.1%. For the full year 2023, corporate revenue increased by 29.5% to $443.7 million. We have 54,493 corporate customer connections on our network at year-end. This represented a sequential decrease of 1% and a year-over-year increase of 21.5%. For the quarter, the sequential impact of USF taxes recorded as revenues on our corporate revenues was a positive $5.9 million and a positive year-over-year quarterly impact of $16.3 million. For the full year, the positive USF impact was $34.8 million.Some comments on the net-centric business. Our net-centric business continues to benefit from the continued growth in video traffic, streaming and wavelength sales. Our net-centric business represented 34.2% of our revenues this quarter and declined sequentially by 1.9% to $93.1 million and grew by 40.7% on a year-over-year basis. For the full year 2023, our net-centric revenue increased by 33.7% to $343.6 million. We had 62,370 net-centric customer connections on our network at year-end that was a slight sequential increase of 0.1% and a year-over-year increase of 20.7%.Comments on the enterprise business. Our enterprise business represented 19.2% of our revenues for the quarter and was $52.3 million. We had 20,740 enterprise customer connections at the end of the year on our network. Our enterprise revenue decreased sequentially by $7.7 million or by 12.8%. For the full year 2023, our enterprise business revenue was 16.3% of our revenues. As a reminder, there was no enterprise revenue last year, or $153.6 million.Lastly, on the wavelength business, our new wavelength product represented 1.2% of our revenues this quarter, and was $3.3 million and we had a total of 667 connect -- wavelength connections on our network at year-end.Revenue and customer connections by network type. Need to make some comments also on the billing transition that we went through in the fourth quarter. In the fourth quarter, we fully integrated our Sprint customers into our billing platform. All Cogent customers worldwide are now billed from 1 Cogent billing system. This transition delayed some customer payments from December into January since the former Sprint customers needed to update their systems to remit payments to our lockbox from the T-Mobile lockbox. This increased our day sales to 37 at year-end, which was a temporary increase. Additionally, once we provisioned every Sprint order into our billing system, we reclassified $1 million of on-net revenue and $400,000 of off-net revenue from Q3 to non-core revenue. We also reclassified 1,373 on-net customer connections at the end of the third quarter to 157 off-net customer connections and 1,216 non-core customer connections. This was to conform to our classification methodology as we were using the T-Mobile billing system through October of 2023. These changes are reflected retroactively in our summary of financial and operational result tables that is included in our press release.On-net revenue. Our on-net revenue, including wavelength revenue, was $141.2 million for the quarter, that was a sequential increase of 6.9% and a year-over-year increase of 22.8%. For the full year 2023, our on-net revenue increased by 14.5% to $518.6 million. Our on-net customer connections were 88,733 at year-end. We serve our on-net customers in our 3,277 total on-net multi-tenant office and carrier-neutral data center buildings. We continue to succeed in selling larger 100 gigabit connections and 400 gigabit connections in carrier-neutral data centers and selling 10 gigabit connections in selected multi-tenant office buildings. Selling these larger connections has the impact of increasing our year-over-year and sequentially on-net ARPU.Our off-net revenue was $123.7 million for the quarter. That was a sequential decrease of 5.3% and a year-over-year increase of 235.4%. The sequential decline in our off-net revenue was partially impacted by our migration to certain off-net customers to on-net. For full year 2023, our off-net revenue increased by 169.2% to $393.5 million. Our off-net customer connections were 36,895 at year-end and we served these off-net customers in over 27,000 off-net buildings. These off-net buildings are primarily located in North America.Lastly, on non-core revenues. Our non-core revenue was $7.3 million for the quarter that was a sequential decrease of $5.6 million, or 43.5% due to our decision to end of life these non-core products. Non-core customer connections were 1,975 at year-end.Some statistics on pricing. Our average price per megabit for our installed base decreased sequentially by 7.1% to $0.28, but increased year-over-year by 4.9%. Our average price per megabit for our new customer contracts for the quarter was $0.10.ARPU. Our on-net ARPU increased sequentially and our off-net ARPU decreased. However, our year-over-year on-net and off-net ARPUs increased primarily from the impact of the Sprint business and also selling larger connections. Our on-net ARPU increased sequentially by 9.7% from $484 to $530. Year-over-year, our on-net ARPU increased by 14.4% from $464 last year. Our off-net ARPU decreased sequentially by 2.9% from $1,150 to $1,117. Year-over-year, our off-net ARPU increased by 22.2% from $914 last year.Our sequential churn rate for our on-net and off-net connections for the combined business improved. Our on-net unit monthly churn rate was 1.2% for the quarter, which was a material improvement from 1.8% last quarter. Our off-net unit monthly churn rate was 1.3% this quarter, an improvement from 1.5% last quarter.EBITDA and EBITDA margin. We reconcile our EBITDA to our cash flow from operations in each of our quarterly press releases. We incurred $17 million of Sprint's non-capital acquisition costs this quarter compared to $400,000 last quarter. Included in that $17 million of Sprint acquisition costs for the quarter are $16.2 million of severance costs that we paid but are fully reimbursed by T-Mobile and have been fully reimbursed. Under U.S. GAAP, these costs need to be reported as SG&A post-acquisition costs and correspondingly as a component of the bargain purchase gain, so no net P&L impact, and they are reflected as Sprint acquisition costs since they are directly tied to the acquisition classified that way on our P&L.EBITDA as adjusted and EBITDA as adjusted margin. Our EBITDA as adjusted includes adjustments for Sprint acquisition cost and cash payments received under the $700 million IP Transit services agreement with T-Mobile. We billed and collected $87.5 million under that agreement this quarter. We billed $233.3 million and collected $242.2 million under that agreement for the full year 2023. All amounts billed under the IP Transit services agreement have been paid to us on time. Our EBITDA as adjusted for Sprint acquisition costs and cash payments under the IP Transit services agreement was $110 million for the quarter and a 40.6% margin. Our EBITDA as adjusted was $352.5 million for the full year and a 37.5% margin.Comments on foreign currency. Our revenue earned outside of the United States is reported in U.S. dollars and was about 16% of our revenue this quarter and 18% for the year. About 10% of our revenues for the quarter were based in Europe and the remaining 6% outside of the U.S. were related to Canada, Mexico, Oceanic, South American and African operations. The average USD to euro rate so far this quarter is a $1.9 and the Canadian dollar average rate $0.75. And if those average rates remain at their current levels for the remainder of the first quarter of this year, we estimate that the FX conversion impact on our sequential quarterly revenues would be positive and about $0.5 million and the same impact on a year-over-year basis. We believe that our revenue and customer base is not very highly concentrated even with the Sprint acquisition. Including the impact of the customers acquired in the Sprint business, our top 25 customers represented 16% of our revenues this quarter and 15% for the year.CapEx. Our quarterly CapEx was $43.6 million this quarter and our CapEx was $129.6 million for the year. We are continuing our network integration of the former Sprint network and legacy Cogent network to one unified network and converting Sprint switch sites into Cogent data centers.On finance leases and payments. Our finance lease IRU obligations are for long-term dark fiber leases and typically have initial terms of 15 to 20 years or longer and often include multiple renewal options after the initial term. Our total IRU finance lease obligations were $484.5 million at quarter-end. We have a very diverse set of IRU suppliers and we have contracts with over 325 different dark fiber suppliers worldwide.Comments on cash and cash flow. At quarter-end, our cash and cash equivalents and restricted cash was $113.8 million. Our $38.7 million of restricted cash is directly tied to the estimated fair value of our interest rate swap agreement. Our operating cash flow results are materially impacted by the timing and amount of our payments under our transition services agreement with T-Mobile and the presentation of payments under the $700 million IP Transit services agreement. Payments under the IP Transit $700 million agreement under U.S. GAAP are considered cash receipts from investing activities and not classified as operating activities. Our operating cash flow was a use of $32.5 million for the quarter compared to $52.4 million of the use last quarter. Our operating cash flow this quarter was impacted by the billing conversion and our operating cash flow was $33.6 million for full year 2023. Our payments received under the IP Transit agreement are recorded as cash provided by investing activities and were $87.5 million last quarter, the same as this quarter, and for the year $204.2 million was collected.Our total gross debt at par, including finance IRU lease obligations was $1.5 billion at year-end and net debt was $1.4 billion. Our total gross debt to last 12 months EBITDA as adjusted and our net debt ratios both significantly improved this quarter. Our total gross debt to last 12 months EBITDA as adjusted was 4.07 at year-end and net debt was 3.75, an improvement to 4.23 at the end of Q3. This is compared to a gross debt in the last 12 months EBITDA as adjusted ratio of 4.79 at the end of Q3 and again, a net ratio of 4.23 last quarter. Our consolidated leverage ratio as calculated under our note indentures slightly different, reduced to 3.67 from 4.57 last quarter and our secured leverage ratios calculated under the note indentures reduced to 2.4 from 2.97 last quarter.Some comments related to our swap agreement. We are party to an interest rate swap agreement that modifies our fixed interest rate obligation associated with our $500 million 2026 notes to a variable interest rate obligation based upon the secured overnight financing rate, or SOFR, for the remaining term of those notes. We recorded the estimated fair value of the swap agreement each reporting period and incurred corresponding non-cash gains and losses due to the changes in market interest rates. Our interest expense and operating cash flow for the full year 2023 was impacted by $21.5 million of interest expense paid in May and November associated with the swap agreement and that was compared to $2.1 million last year. The fair value of our swap agreement decreased by $17.7 million from last quarter to $38.7 million. We are required to maintain a restricted cash balance with the counterparty equal to the liability.Our day sales outstanding, or DSO, as I mentioned earlier, was significantly impacted by the billing conversion. Our DSO for worldwide accounts receivable was 37 days versus 27 days last quarter. Our DSOs after year-end have reverted back to historical norms.Our bad debt expense was $1.9 million and 0.7% of our revenues for the quarter, that was also impacted by the billing conversion. Our bad debt expense was $8.6 million and 0.9% of our revenues for the year.Finally, I want to thank and recognize our worldwide billing and collection team members for managing this billing conversion from the legacy T-Mobile Sprint billing platform to our Cogent billing engine. This was a tremendous operational achievement and we completed this in only 6 months from the acquisition date. All customers were billed worldwide from the Cogent billing system starting in November of 2023.I will now turn the call back over to Dave.
Hey, thanks, Tad. I'd like to highlight a couple of the strengths of our network, our customer base and sales force. We continue to experience significant traffic growth in our net-centric business. We continue to be beneficiaries of increased over-the-top video and streaming, particularly in international markets. By quarter's end, we ended with 1,558 carrier-neutral data centers and 68 Cogent data centers directly connected to our network. That total of 1,626 data centers is more than any other carrier globally as measured by independent third-party research. The breadth of this coverage allows us to serve the net-centric market, better allowing our customers to optimize their networks for reduced latency. We expect to continue to widen this lead in the market as we project adding over an additional 100 carrier-neutral data centers to our network per year for the next several years. We also expect to continue to convert Sprint facilities into coaching data centers. 23 of these facilities are in process of being converted. To date, we have completed the conversion of 22 of the facilities into Cogent data centers.As of today, we are selling wavelength services in 65 carrier-neutral data centers. With expanded provisioning cycles, we can also sell wavelengths in an additional 285 carrier-neutral data centers or a total of 360 facilities across North America. We are generating $3.3 million of revenue from wavelength sales in the previous quarter with 667 discrete installed wavelengths. We have a significant funnel of wavelength orders in the pipeline. Today, we have, in a combination of orders signed, as well as in our sales funnel of over 2,300 orders. Our network traffic continues to increase. It increased 7% sequentially and 22% year-over-year. At quarter-end, we directly connected to 7,988 networks. This collection of ISPs, telephone companies, cable companies, mobile operators, and other carriers allow us to directly reach the vast majority of the world's broadband subscribers and mobile phone users. At quarter's end, we had a sales force of 271 net-centric reps focused on this market. That was in addition to the 374 reps that we have focused on our corporate segment and 12 sales reps focused on our enterprise market.The corporate trends that we are seeing are positive, but have still been impacted by the pandemic. Our corporate customers are continuing to integrate new applications which have become part of their normal workload, including the extended use of video conferencing. This usage requires high speed, high capacity connections, both inside and outside of their premises. Our enterprise customers continue to focus on dedicated Internet access and VPN services, inclusive of the older MPLS technology to manage their networks. We remain focused on improving our sales force's efficacy through training and managing out underperforming sales reps. Our sales force turnover rate did improve substantially in the quarter to 4.1% per rep per month for the quarter, down from a peak of 8.7 reps per month at the height of the pandemic and much better than our average historical number of 5.6%. We are continuing to train reps who join Cogent from the Sprint business.We remain optimistic about our unique position in serving the market, particularly around our corporate footprint in central business districts where we have over 1 billion square feet of rentable office space and 1,862 multi-tenant buildings on-net. We're excited about our large enterprise customer base as this provides us a new targeted market and our wavelength opportunity is just beginning to unfold as we continue to repurpose the Sprint network and optimize it for the delivery of wavelength services. As mentioned earlier, we can serve customers in 65 locations today, where we are delivering waves. We have another 285 locations that are enabled for service with longer provisioning windows. That significant backlog and funnel of approximately 2,300 wavelength opportunities gives us a great deal of confidence that as we continue to modify and enhance the Sprint network to provision wavelengths, we will be able to convert these on a much more expeditious schedule and by year-end hope to mirror the provisioning windows that we have experienced in our net-centric transit services.The key indicators of office activity, workplace reentry and leasing activity remain substantially below pre-pandemic levels. However, many tenants are returning to their offices, and leasing activity appears to beginning to improve. We are diligently working to continue to integrate all of these Sprint assets and customers into our systems, our processes, and one unified network. This will allow us to continue to improve our cash flow generation. Over the next 3 years, we anticipate an annual savings due to multiple synergies of over $220 million a year.With that, I'd like to open the call now for questions.
[Operator Instructions] Your first question comes from Anton Rennert with Cowen.
This is Anton filling in for Greg Williams at TD Cowen. I saw that CapEx came in a little bit higher at $43 million. How should we think about CapEx just going forward and the outlook there?
Yes, sure, Anton. As we outlined, the CapEx that we expect to spend on a going-forward basis should be about $100 million a year. We also indicated at the time we announced the acquisition of the Sprint network that there would be about a $50 million one-time set of expenditures. We are about 60% of the way through that extraordinary $50 million. That was the reason why our CapEx came in at approximately $130 million last year.In thinking about the capital required to run the combined business, there are really 3 categories. There is the maintenance capital required to run the legacy Cogent network and its associated IP and VPN business, which is about $35 million a year. There was approximately $30 million a year in continuing capital expenditures on the acquired Sprint network. We are continuing to spend that capital. However, we have repurposing those expenditures to primarily focus on the wavelength opportunity. And then third, we are expecting to be able to -- additionally be able to use capital to expand the footprint. We spend about $30 million a year in footprint expansion.The final point I'd like to make is, in thinking about our capital, you really need to look at the combination of what is reported as capital, as well as the principal payments on capital leases. And in fact, sequentially from the third quarter to the fourth quarter, those principal payments on capital leases declined materially from $41.3 million to $18.8 million. I think you should think about these as pretty good run rates going forward. So probably in the order of about $80 million a year for the next several years on principal payments on capital leases. And then on addition to that, about $100 million in CapEx.
Your next question comes from Alex Waters with Bank of America.
Maybe just first on wavelengths, can we maybe just talk a little bit about the rationale of no longer stripping out wavelengths in the press release?And then secondly on that, just heading into 2024, I think last quarter you noted that we should probably be around the $20 million range a quarter by midyear for wavelengths. Can we just talk about that as we get there?And then just on SG&A, can we just talk about the uptick quarter-over-quarter? And then how we should think about SG&A going into 2024?
Yes, sure. So the decision on just not including it in the press release was so we would have a more fulsome opportunity to discuss it in the prepared remarks as we did. We -- I think it is important to be able to disclose both the revenue run rate, which was $3.3 million up sequentially, and the unit count which was 667. We are still hampered by the number of sites that we can provision. The backlog has more than doubled sequentially in the quarter. So, on the third quarter earnings call, our backlog was approximately 1,000 orders in the sales funnel and provisioning funnel. That number is up to approximately 2,300. We actually anticipate, based on information from the sales force and their conversations with customers, that we're going to continue to see an acceleration in the order volume. We are frustrated by the amount of time it's taken to wave-enable sites. We are still confident by year-end that we will be able to have 800 sites that can provision waves with a kind of 2-week average provisioning window. We are definitely not there today.As a result, we will probably not be on a run rate by midyear of $20 million of installed business. I think we will have a funnel that will demonstrate that. But the sheer number of sites that need to be touched and the number of steps that have to be done to convert the former Sprint voice network into a wavelength optimized network is a very daunting task. We are progressing well. There are over 1,000 of our 2,000 employees who are almost completely full-time focused on this effort. We absolutely will meet the year-end targets, but I think by midyear the funnel will demonstrate that run rate, but it probably will not be provisioned due to these extended provisioning windows in those 285 sites.And I'll let Tad jump into the SG&A numbers.
Sure. So I think the best way to look at our cost run rate is to look at the combined COGS and SG&A together. We had some classification adjustments we needed to make in the quarter and for the year. Result of that is for a period of time, T-Mobile was paying our bills for us. So we were getting that information in having to classify it according with what came in. The takeaway is, for the third quarter, the combined cost of goods sold and SG&A rate was about $231 million. There was a benefit in that quarter of about $8.5 million for the change in accounting for a capital lease. So adjusted for that, it's about $240 million for the quarter. The combined for this quarter was about $249 million. Now, we had a couple of increases that are not going to reoccur. Our USF, as we mentioned on the call, increased by $6 million. We had a bad debt that we needed to record since it's based on the relationship of cash receipts to billing, and because of the billing delay that went up, and that was about $2 million. And then we have year-end audit adjustments and also bonuses to employees combined of about $2 million. So you're comparing an adjusted $240 million combined COGS and SG&A for Q3 to about $239 million this quarter. So it did slightly improve. I know that's a little complex, and I think the other way to look at SG&A on a going forward basis is about 27% of revenue. Hopefully that helps.
Your next question comes from Walter Piecyk with LightShed Partners.
That was going to be my question. I assume that there was some reversal of reversals, but USF revenue as well, right? So if the expense goes up and down...
That's correct. That is correct, Walt. $5.8 million sequentially.
So let's look at this a different way. Let's take the $110 million reported minus the $87.5 million, it gets you to $23 million, which was down 50% sequentially. Do whatever comparisons you want year-over-year. You basically just said that's the new run rate, maybe less $240 million versus $240 million, like less $9 million. So the run rate is $30 million of EBITDA for the legacy business. When you exclude the TSA payments, which we all know have a finite end.
So, Walt, your arithmetic is a bit flawed in this case because what you're doing is counting the expenses that we acquired in acquiring the Sprint enterprise base but then excluding from that the subsidy payments that T-Mobile contractually agreed to. While you are correct they are finite, so are those expenses and we are achieving substantial improvements by reducing headcount, by exiting on economic agreements, and by moving customers from off-net to on-net. As we had described, it would take us 3 years to do that, and we would achieve approximately $220 million in annualized savings. We actually are running ahead of that.The T-Mobile payments and their pacing were based on negotiations between the parties and the contractual schedules that were expenses that we knew we would exit but maybe could not exit immediately. So I actually think to calculate EBITDA you have to use both the expenses and the monies coming in, which...
That's not true. That's not flawed at all, because I fully appreciate the synergies that will be achieved in 3 years. But that's 3 years from now. You don't get the synergies today, and we're trying to figure out obviously a baseline for your EBITDA and then you will achieve synergies, which we'll give you full credit for over 3 years. And to say that you're ahead of schedule in the $220 million, but the baseline EBITDA is only $30 million, that's not necessarily a positive, right? You'd want to say we've got more synergies ahead of us to achieve off of that baseline as opposed to less. So I don't see how my math is flawed. I fully understand how synergies are achieved over time. I'm just saying that this is the baseline EBITDA today.
So first of all, we have begun achieving those synergies to date. Many of those synergies have contractual counterparty obligations that roll off between now and the end of 2026 as we have been very clear in explaining...
It's in our estimates -- which is in everyone's estimates. But that has nothing to do with today's EBITDA. Everyone knows that those expenses fall off.
Right. But to cover those expenses, we have a stream of payments that are contractually obligated to be paid by T-Mobile to us. In looking at EBITDA, you need to include those T-Mobile expenses. Tad pointed out, for example, $16.2 million in severance reimbursement that we got in the quarter is not in your EBITDA number, but it was a cash payment to us from T-Mobile. We're trying to be as transparent and granular as we can possibly be, but the...
I think it would be helpful then if you're trying to be transparent, to put all of this information in the press release rather than having people feverishly write down the data on -- I know you said earlier, you just said you want to have a robust prepared comment, but that's not helpful in the spirit of transparency as opposed to putting the numbers in print in the press release when the quarter is released.
I think we're pretty granular in the level of detail and consistent in putting it in the press release, Walt. Compared to other public companies, I'm going to differ with you, I think we are very granular. The only comment that I made is, I wanted the opportunity to describe the wavelength funnel and our frustration in that, we're only selling wavelengths in 65 locations today while we have orders in hundreds of other locations that we are rapidly enabling to be able to support those wavelengths.But again, to your EBITDA number and we can take this off-line. I'm happy to do a follow up with you. I think you can read the K, you can look at the detail. We're about as granular as possible. We are...
We read it all. Let me -- last question, Dave. Let's focus on revenue. Corporate revenue on the positive side, although again there's numbers that we don't get like in Sprint non-core. It looks like it at least is not declining anymore. Can you give us -- I can't even strip through all these numbers since there's stuff that's not reported. But what is the -- what do you think the sequential growth is in your legacy corporate revenue, excluding the excise taxes and everything else?
That are provided in the press release for the quarter, corporate revenue, $126.6 million...
That includes $20 million of excise taxes.
Hold on, let Tad finish, Walt.
You let me finish. Last quarter, $120.5 million. That's an increase of $6.2 million. In that increase is $5.8 million of USF. So there was a net increase of about $300,000 sequentially if you adjust for USF.
Well, and the reality is...
There's also, in that number -- also is in that number is Sprint corporate non-core, which is not broken out. And that number can go up and down.
So the total non-core declined from $12 million to $7 million. It declined by $5.5 million. It actually declined by more than that as we pointed out because when we converted customers from the Sprint billing platform that T-Mobile was operating ours, there was another $1.4 million of revenue that really was non-core, but not designated with product codes that would represent non-core under T-Mobile's billing. So the $5.5 million sequential decline in non-core is actually understated by $1.4 million.So at the end of the day, we're not moving stuff around, Walt. We're trying to get everything into a consistent way, so quarter-over-quarter, you can look at the non-core, it's not going to go back up, it's going to continue to go down. And the fact that it went from $12.5 million to $7.5 million with a $1.4 million add back, is to me very positive in terms of...
I didn't say it wasn't. I'm just trying to get to that number because I was saying that it looks like you inverted positive this quarter. I just wanted to affirm that because of these multiple moving pieces and getting to what was a very predictable number prior to all of this mess with Sprint. So I assume that that's positive now, which is an inversion, and I assume you're -- and so I assume you're expecting that to continue positive going forward. And if you want to talk about the dynamics of the corporate market, please do.
Well, you are correct, our corporate business net of USF did grow sequentially. It is continuing to grow. The pace at that growth is not at pre-pandemic levels, but we are seeing sequential improvements.
Your next question comes from Nick Del Deo with MoffettNathanson.
Number of questions, I guess, I'm still a little confused on the cost and EBITDA trajectory. Maybe to ask it a slightly different way. Last quarter your EBITDA, including the T-Mobile payments, excluding integration cost, was $131 million. This quarter, it was $110 million or $111 million. So, call it, a $20 million sequential decline in EBITDA. And, Tad, I know you called out the higher bad debt for a couple of million bucks year-end audit, and bonus for a couple of million bucks in Q4. I guess, anything else worth calling out to help explain that $20 million delta?
Yes. Last quarter we had the benefit of the lease accounting, which I said was $8.4 million. So that's a different to that. And then also the USF's impact.
Which was $5 million and then the bad debt. So those...
Then you've got year-end bad debt, audit, bonus accruals, that thing, those all combined are close to $5 million.
Okay. Just -- maybe I'll just drill down on 2 of that...
Really, the costs comparable, it's about $240 million versus $239 million. And that's COGS and SG&A combined. Now, revenue did decline, so that's really the pure as adjusted EBITDA.
$3 million less in revenue.
Because cash payments were the same. $87.5 million, $87.5 million. So the SG&A run rate, forget reclasses and other things, and year-end adjustments should be about 27% of revenues for modeling purpose going forward. Close to 30% at the -- for this quarter.
Okay, just to clarify 2 things. So the USF for $5 million, you called out. Again, isn't USF just a pass-through with no profit impact?
Yes, it is, but when you're reconciling those lines, you need to include it. I agree on the EBITDA perspective, it's a wash.
Okay. And then the lease adjustment, again, the thing that -- I thought that -- maybe I'm thinking something different, that you had the $12 million or $13 million adjustment where a lease went from OpEx to a finance lease that bumped up EBITDA. So I thought that would have occurred in Q3 and Q4 consistently.
No, because the previous expenses were reversed when the accounting was changed.
Nick, it does continue going forward. But you got a payment...
The payments this quarter hit interest and principal. And when we recorded the lease in the third quarter, you recorded the balance, and then you had to reverse the prior amounts that had been charged.
From May through September 30. So it was really 5 months of reversal versus the run rate.
Okay. Okay. That latter nuance I hadn't picked up on. All right. Well, thanks for clarifying all that. Maybe if I can ask 2 other questions, maybe. First, with the Sprint IPv4 addresses, what's the plan there? Do you sell them, do you lease them? Do you hang on to them for the business? And what might the financial impact be over time?
Okay. So, there's a fair amount of complexity here. So, when we acquired PSINet, we got a large number of addresses previous to Sprint, and we had some organic Cogent and some from other acquisitions. But pre-Sprint, we had approximately 28 million addresses that had 0 value on our balance sheet, yet they had real economic value. Those addresses are traded every day in public exchanges for about $55 an address. You can go online and look at the quotes as we talk. When we did the acquisition of PSINet, there was no value to addresses because they were still available for free. And second, the accounting rules were different, and we recorded things as negative goodwill, which is no longer how you record a gain. You now record it as bargain purchase gain as we did in Sprint's case.In the case of Sprint, we acquired 9.9 million incremental addresses, bring our total to about 37.8 million addresses that we own. When that happened, we actually initially were not going to focus on them. And because we have been generating leasing revenue from addresses since 2015, included in our corporate and net-centric numbers, it's primarily net-centric, it's roughly 85% net-centric, about almost 13% corporate, and a couple of percent in enterprise. We generate $40 million a year out of leasing those numbers out. We continue to lease out incremental inventory. Today, we are leasing about 11.4 million of the 37.8 million addresses that we currently have. Our average lease price per address is about $0.30 per address per month. We were somewhat unique in the market in having an inventory and being a service provider and leasing these addresses. We didn't focus a whole lot of attention on it. It was just included in all of our numbers, didn't even break it out as a separate product. It was just baked in.Now, we are going to break out the unit count and the number separately. And part of the change occurred about a year ago when Amazon, which had been a serial buyer of addresses, began to compete with Cogent and lease addresses. Amazon leases its addresses on an hourly basis through AWS but nobody really leases them by the hour. And in fact, they generate about $3.60 an address a month, or 12x Cogent's rate. So based on kind of now a 2-party leasing market and a transaction market, the accounting firm that did the appraisal came back and said, you guys have to recognize a bigger bargain purchase gain to account for these addresses. That's why we picked up another $254 million.The final point to this is, we are going to continue to evaluate opportunities to either sell addresses, which we have not done yet, or potentially securitize the cash flows from those addresses. We're generating about $40 million of EBITDA today off of our address leasing, and that continues to grow. We'll continue to evaluate. Is it better to lease or should we sell?
Just to summarize quickly on the inventory and the accounting. Just -- I know Dave covered it. So we have 38 million addresses in total. In 2002, 28 million were acquired with PSINet. We paid only $12 million in cash to PSINet. Back in those days, the accounting was, you record only the assets equivalent to your purchase price. So we recorded $12 million of intangibles basically. The rules now are you record the fair value of all assets acquired. With Sprint, we acquired 10 million, 9.9 million additional addresses. That is an acquired asset that needs to be valued. We had the Big 4 accounting firm include in their appraisal how much these addresses are worth. And it came out with a net total per address value of $46 per address. So they are reflected on our balance sheet in our 10-K, which we will file today at $458 million in total. The net gain for the quarter includes the $250 million that we increased to bargain purchase, includes that $458 million new address, the adjustment for the income tax impact of that. So, deferred taxes and then we had some other adjustments that we recorded.So, I hope that's a good summary as to where we are as Dave mentioned, we are leasing 11.4 million of the addresses and generating about $35 million of lease revenue currently per year. There's no costs associated with this revenue stream and they are an asset that's partly on our balance sheet and partly not.
Okay. A lot of great detail. I guess, just one quick clarification here. About how many do you think you need to run the business? So, in other words, how many are surplus that you could sell or lease out or securitize if you wanted to?
Virtually none, probably 100,000, you could run the network. You would use some to number your own network devices. Today, we have about 60,000 network devices in the network and customers can bring their own. Between 2015 and midyear 2022, we actually would only lease addresses to people that bought bandwidth from us. Anticipating Amazon's entry into the market, we relaxed those rules and we saw a significant spike up in leasing activity by leasing addresses to people that don't buy bandwidth, as well as those who do continue to buy bandwidth. We could sell those addresses above that roughly 100,000 or run the company. Many other service providers have very limited pools of addresses.
Okay. And then the last question I wanted to hit on. I'm sorry for this being so long. I guess, I'm not sure what you can say or would want to say about the peering dispute with NTT, but I'm interested in what you can share and whether you think it's going to have any customer revenue impacts in the coming quarters.
Happy to talk about it. It was actually something initiated by Cogent. So, we have had a peering relationship with NTT since 2001. They are the fourth largest network in the world. As we continue to gain market share in Asia, NTT refused to give us connectivity in Asia. We remained connected in the U.S. and in Europe. We had multiple conversations with NTT technical and management individuals and they basically said they did not welcome Cogent's entry into the Asia market and refuse to add ports in Asia as our peering required. In retaliation, after multiple attempts to get them to connect even outside of the home market of Japan, we're willing to take Singapore, Australia, Taipei, Hong Kong, a number of other Asian locations being sensitive to the protection of their home market, they continue to refuse. So as a result, we de-peered them, but only in Europe. They do have a European business and they were forcing Asian traffic to trombone to the U.S. Basically, there's still connectivity, it was just customers had to go from Japan to LA and back. We kind of implemented a similar strategy in Europe. Their European customers have to come to D.C. or New York and back to Europe. I know it sounds almost childish and it's a tit for tat, but unfortunately, there are still bad actors in the world who don't embrace net neutrality.Now, I'll expand one point further, you know we have a similar situation with Deutsche Telecom which has been public and they continue to expand capacity in Asia but refuse to expand in Europe. So they'll expand connectivity in the U.S. or in Asia, but they want to protect their German market. So while they may meet the 4 corners of a net neutrality order in their country, they are absolutely violating the spirit. And it's actually somewhat encouraging that the FCC may finally codify this in a way to stop people from playing these games.
Your next question comes from Tim Horan with Oppenheimer.
So, Dave, if AWS is 12x the price, I mean, why wouldn't you increase your prices like 6x? I mean, where a customer is going to go, do you think you would lose those customers? And if I do the math, if you have a 6x price increase, that's obviously $40 million goes to $240 million of free cash flow. What's preventing that?
Yes. So, first of all, we didn't have AWS as a benchmark with a pricing umbrella in 2015 when we started leasing addresses. And we have never changed our pricing on addresses while our bandwidth pricing has declined at 23% a year during that same period.Second point is, you are absolutely correct in your statement that customers probably are pretty sticky. The expense is small and the labor involved in renumbering is pretty significant. Amazon doesn't disclose its numbers from this particular product, so I can't tell exactly how successful they are. What I do know is that, their inventory of addresses which have been acquired, they were not organic to Amazon, is about comparable in size to Cogent. Actually, both Microsoft and Amazon have been serial buyers for the past decade, and each have address inventories about equivalent to Cogent.We are evaluating should we do something that is quite honestly foreign to our thought process and raise prices. We're conscious of that and we are going to evaluate over the next few months what are the best ways to maximize the value of these addresses and whether that includes securitization, that includes raising prices, or it includes the ability to sell addresses. We're evaluating all of those opportunities and we understand we were lucky that we just have a big asset that turned out to be valuable. Like I explained to 1 investor, it's kind of like bitcoin with a purpose. So it's sometimes better to be lucky than smart.
Well, I guess, Dave, related to this, what everyone's a bit frustrated about is the complexity in accounting. Can you just give us a sense of what the adjusted EBITDA should look like in '24? Is the $110 million a good run rate? I mean, I know there's a million moving parts, and if you can't talk about that, maybe the $220 million in expense savings. How much -- how does that kind of pace out over the next 3 years? Or any more color would be helpful.
Yes, I'll try to touch on both of them. So the first point is, we have been very clear that there is going to be a step down in the cash payments that we receive from T-Mobile 1 year from closing, so May of '24, that will impact our EBITDA as reported. Secondly, we are continuing to achieve those cost synergies. Those cost synergies, unfortunately, are limited by some of the contractual obligations that we assumed when we bought this business from T-Mobile. And again, these were all things that were considered in the negotiation, in the purchase price. Even back to the IP addresses, T-Mobile sold 2 million addresses prior to closing. It was fully disclosed to us and they recorded $120 million gain from that sale. Quite honestly, they could have chosen to keep more addresses and sell them, and then the subsidy check they would have had to write would have been larger. So it's not like these are surprises, they were all thought about in the negotiations.In terms of EBITDA, you should think about the current run rate is about correct for the next couple of quarters. Then it will step down when those payments come down from T-Mobile and then gradually increase as those cost savings are achieved. What we have said publicly, and we are absolutely reaffirming is that, within 5 years of closing, the company will do at least $1.5 billion of revenue and have at least $500 million of EBITDA. The $133 million that we did last quarter and $110 million this quarter kind of says we're already on pace for that. The reality is, EBITDA will go down and then come back up. And I understand it's complicated accounting, and I also understand it's hard to build a model that goes up and down. We're trying to be as transparent as possible without giving exact guidance. Tad?
Yes, the payments are fixed per month, so we know it's $29.2 million per month for a year and then $8.3 million for the remaining term. So -- and the step down will occur in May. I would say that the SG&A cost run rate that we had in the fourth quarter, which was about 30% will be less in the first quarter and we should think about an SG&A as percentage of revenues, I mentioned earlier of about 27%.
And then network operating expenses. What percent should that be roughly?
That will be relatively flat.
May come down slightly as well.
We'll have some experience, savings, additional savings from circuits and from facilities.
So it should come down slightly from the 64% this quarter?
Yes. Now, it's going to go up from additional on-net sales, but net-net, it should decline. Yes. I should have said off-net sales.
Off-net.
Your next question comes from Michael Rollins with Citi.
A few questions. Just a quick one on the IP addresses. Can you remind us what the addressable market for that is, and just your share of that market?
Yes, sure. So when the Internet was conceived, there were a numbering scheme and they chose IPv4, which is 2 to the 64th power. That is approximately 4.3 billion addresses. Those addresses were initially allocated by the Department of Commerce, then allocated by IANA as a contractor for the Department of Commerce, and then subsequently those allocations were delegated to 5 regional registries around the world. The U.S. government held back addresses for its own purposes, primarily DoD. So there has been about 3.5 billion usable addresses. The registries began to run out of addresses in 2011. A market for those addresses developed and prices to buy an address at that time were about $4. In the intervening decade, or 13 years, that price has gone from $4 to $55. All of the registries ran out of addresses by 2018 and there are no more addresses to get.Now, there has been a movement to migrate to IPv6, which is 2 to the 128th power number of addresses. So basically the square of 4.3 billion, a very, very large number. But there have been many challenge in doing that, including literally trillions of dollars of capital that would be necessary to replace equipment to support that. So IPv6 still has a relatively small presence on the Internet, about 7% of traffic. So the market is anybody who needs those addresses and wants to reach the whole Internet. There are multiple workaround schemes, but those schemes are not easily implemented and are not as easily managed as just renting or buying IPv4. So I think for the foreseeable future, the world will run on v4. There will be limited supply and as a result, people will either lease or buy these to fulfill their needs.
And then just a question on the business. The sales productivity look like it dipped on kind of a flat FTE count. Can you unpack what's happening in terms of sales productivity, where there might be pluses in performance, and where there might be some minuses in performance?
Yes. I mean, I would say the biggest drag on productivity, and I'm not trying to call out specific names, but have been employees that came over from Sprint. As part of our contractual obligation, we had to guarantee their quota for a full year from closing. That's very different than the way an organically hired Cogent salesperson would be comped. And their productivity has been a significant drag. Now, there's only a dozen of them left focused on enterprise accounts, but we have transferred those reps into corporate and net-centric roles, but their productivity remains depressed.The second factor that has drawn down productivity has been the rapid expansion of the sales force coming out of the pandemic. Normally pre-pandemic, we were growing at about 7% a year and absorbing those people and training them. And reality is, while they have 3 months to become a full-time equivalent, it takes really a year to be fully productive. Last year, we grew the sales force 20% year-over-year, so almost 3x faster than we normally grow the sales force. So that's been a second drag on those numbers. I think we will see some additional turnover and hopefully, we'll see productivity start to improve.
[Operator Instructions] Your next question comes from Bora Lee with RBC Capital Markets.
One bit of housekeeping first. There was a step down in G&A during the quarter relative to 3Q while CapEx increased. Was that related to some of the reclass activities and is that fourth quarter number of the jumping off point go forward? Or is there something else we should be thinking about?
More of the reclass occurred in the fourth quarter versus the third quarter. Really the impact on the run rates in the third quarter predominantly had to do with the impact on COGS from the lease accounting.
Okay. So the G&A step down in the fourth quarter was related to [Technical Difficulty]
Well, SG&A in the fourth quarter has an increase from the third quarter and I tried to reconcile all that.
G&A.
Yes, SG&A.
SG&A. So, the right way to think about Bora is just about 27% of revenues going forward. I think these moving pieces of reclass are hopefully behind us and that we've got everything into Cogent systems. They're all been audited by Ernst & Young and comply with all of the GAAP requirements and critical accounting matters for lease accounting. So, I think we're in pretty good shape.
We are.
Okay. And as you're clearing out the Sprint spaces. Sorry, Tad, were you saying something?
No, go ahead.
Go ahead, Bora.
Great. As you're clearing out the Sprint spaces, is there an opportunity to sell some of that old equipment? And is that meaningful or just sort of a task that needs to be done?
It's definitely a task that needs to be done. It does not have a meaningful salvage value. Most of this equipment has been not manufactured for 15-plus years. There is a de minimis amount of scrap value for copper in the equipment. There is some equipment that has been sold to third-party brokers, but there's also a work effort associated with taking those 22,500 racks of dead equipment out. Net-net, this is not going to be a significant either cost or a significant revenue opportunity for us. The better opportunity is going to be as we depopulate these facilities of this dead equipment and convert them to Cogent data centers, those 68 facilities and the 157 megawatts that we have now that are less than 30% utilized become a significant opportunity, particularly as there is a short-term crunch for power and space for data centers driven by AI. We are in the process of looking in multiple ways to fill that space up more quickly.
And, I guess, lastly for me, recognizing that it's still early days, do you have any quantitative or qualitative color on the extent to which there has been actual cross-selling or cross interest across the legacy Cogent and Sprint customer bases?
So, I will say, I'll start with wavelengths. The majority of the wavelengths in that funnel and the wavelengths that we have sold are to customers that Cogent already had a relationship with. There are a handful of cases where there are a customer that Cogent had not worked with previously, but the vast majority of those 2,300 orders in the funnel and those that have been installed, the 667, have come from people that Cogent had a relationship with.The second thing is, the Sprint enterprise base has been receptive to our on-net footprint, our global reach, and our ability to modernize their VPN technologies. We actually saw very modest but a very slight uptick sequentially in the number of enterprise connections. And remember, this is with a much smaller sales force focusing on those customers. So I think there is going to be the ability to help modernize some of those enterprise customers. So I think there will be cross-selling opportunities in both directions.
Your final question comes from Brandon Nispel with KeyBanc Capital Markets.
It's Evan on for Brandon. The backlog you guys talked about for wavelengths, you were saying it's growing and now pacing the provisioning you're able to do on your facilities. Do you think you'll be able to get through that 2,300 order backlog by the end of the year? Or are you finding any customers finding alternative solutions because of the backlog?
So, Evan, I think 2 different things will happen. We will provision most of those orders, but there will be some customers who cannot wait. We are trying to be very transparent with customers, and it's a site-by-site discussion on what that provisioning window will look like. We know that with the network reconfigurations that we have going on, we'll have more than double the number of sites and a standardized provisioning window by year-end. But in the intervening time, if a customer needs to go somewhere else and we can't provision, we are going to let them out of that obligation. I mean, it's -- if we want to do business with them going forward, we need to understand that this is a Cogent problem and not the customer's problem.Conversely, as we continue to build credibility with customers and we get more sites enabled and shorter provisioning windows, we actually anticipate the pace of that funnel building actually accelerating. In the last quarter, it took us basically 5 months from closing to build a funnel of 1,000. In the last quarter, we got that up to 2,300. And yes, there have been some fallout, but the net number grew, and I think that will continue to grow. As I stated earlier, over the long run, I think it's not healthy to talk about funnels but install revenue. But until we get the network configured correctly and get enough sites where we can provision in a expeditious manner of a couple of weeks, we have to give both customers and investors an understanding of what the backlog looks like.
This will conclude our question-and-answer session. I will now turn the call back over to Dave Schaeffer for closing remarks.
Hey, thank you very much. I know, again, it was a long call, but there are a lot of pieces of information. We are definitely trying to be wholesome and transparent in what we are reporting. Thank you all very much, and we'll talk soon. Take care. Bye-bye.
This concludes today's conference. Thank you for your participation. You may now disconnect.