# The air-traffic-control computers the FAA mostly threw away: the Advanced Automation System Of about $2.6 billion the FAA spent modernizing air-traffic-control computers, roughly $1.5 billion was written off as wasted. Here is the full record. Author: J.A. Watte Published: July 16, 2026 Source: https://jwatte.com/blog/faa-advanced-automation-system/ --- Air-traffic controllers who guide airliners across the continental United States do not watch the sky. They watch computer displays. In the early 1980s the equipment behind those displays was already aging, and the Federal Aviation Administration set out to replace it with a single ambitious system for the nation's en route centers, the facilities that manage high-altitude traffic between airports. The program was called the Advanced Automation System, or AAS. By the time the FAA restructured it in 1994, the agency had spent about $2.6 billion, and according to the U.S. General Accounting Office, roughly $1.5 billion of that was wasted. That is the headline number, and it is worth pinning down before anything else, because the AAS story is easy to garble. The $1.5 billion is the portion of money already spent that the government wrote off as not usable for follow-on work. It is not the same as the $2.6 billion total spent, and it is not the same as the roughly $7.6 billion the FAA at one point projected the finished program might cost. Those are three different figures measuring three different things, and keeping them separate is the whole discipline of reading this case honestly. ## What the program was The FAA announced the Advanced Automation Program in 1981 and moved into development with a plan to modernize the computers and controller workstations at its en route centers. This was safety-critical, real-time software for the busiest airspace in the world. It had to fail safely, run continuously, and be usable by controllers who could not stop working while it was installed. In 1983 the FAA estimated it would cost about $2.5 billion to develop AAS, with completion scheduled for 1996, according to GAO report T-RCED/AIMD-98-85. After a design competition lasting roughly four years, IBM Federal Systems won the development contract in 1988. At that point the FAA estimated the project at $4.8 billion, with completion pushed to 1998, again per GAO. The design was what critics later called a grand-design approach: build the whole thing more or less at once, across several linked segments, and deliver a sweeping replacement rather than a series of smaller upgrades. Those segments included the Initial Sector Suite System, the Terminal Advanced Automation System, the Tower Control Computer Complex, the Area Control Computer Complex, and the Peripheral Adapter Module Replacement Item. ## What happened The program drifted. Requirements kept changing, software development ran behind, and the cost estimates climbed. By December 1993, amid roughly $1.2 billion in overruns, the FAA ordered a review of the IBM-affiliated project, according to a contemporaneous United Press International report. That $1.2 billion overrun figure is a pre-restructuring number from 1993 press coverage, and it should not be confused with the later $1.5 billion write-off, which is a separate accounting from the 1994 restructuring. The reckoning came in 1994. By then the FAA estimated AAS could cost as much as $7.6 billion, with completion pushed as late as 2003, per GAO. That is an estimate of projected total cost, not money spent. GAO's own summary text of the same report rounds the figure to as much as $7 billion, and contemporaneous press often cited a range of about $5 to $7 billion, so a citation of $7 billion is not a contradiction. The specific detailed figure in the GAO report body is $7.6 billion, and that is the one used here with its source attached. Against the program's 1988 baseline milestones, GAO found that components had slipped from a minimum of 3.5 years to as much as 8 years. Faced with that, the FAA restructured the program rather than continue as planned. ## The restructuring, segment by segment The 1994 restructuring did not cancel everything, and the details matter because the surviving pieces are the reason the story has a second half. GAO frames the restructuring as affecting four of the five AAS segments. - The Initial Sector Suite System was scaled back and renamed the Display System Replacement, or DSR. - The Terminal Advanced Automation System was cancelled and its mission was carried forward by a new program, the Standard Terminal Automation Replacement System, known as STARS. - The Area Control Computer Complex was cancelled in 1994, with the replacement of the existing Host computer deferred to roughly 2004 to 2005. - The Tower Control Computer Complex was scaled back in 1994 and then cancelled in 1997. It was not cancelled outright in the 1994 restructuring, and it is worth stating precisely because the two dates are easy to merge. - The fifth segment, the Peripheral Adapter Module Replacement Item, or PAMRI, was unaffected by the restructuring. It was fully operational across the en route centers by 1993 and was never in trouble. So the accurate framing is that four of five segments were affected, with PAMRI as the untroubled survivor, DSR and STARS as the segments that survived by being restructured, and the Area Control and Tower Control complexes as the pieces that were cancelled. ## The money, kept in its buckets Here is the arithmetic that the case turns on, with each figure attributed and dated. - 1983: the FAA estimated about $2.5 billion to develop AAS, completion 1996. This is an estimate. - 1988: at the IBM contract award, the FAA estimated $4.8 billion, completion 1998. This is an estimate. - 1994 restructuring: the FAA estimated the program could cost as much as $7.6 billion, completion as late as 2003. This is a projected estimate that was never spent as a total. - Through May 1994: the FAA had actually spent about $2.6 billion on AAS, per the estimate of the agency's investment analysis group cited by GAO. This is money spent. - Of that $2.6 billion spent, about $1.1 billion in laboratory facilities and computer hardware and software was salvageable. The balance, about $1.5 billion, was wasted because the equipment and work did not contribute to follow-on projects. The Department of Transportation Office of Inspector General audit AV-1998-113 corroborates the split of about $2.6 billion spent and roughly $1.5 billion lost. So the correct headline waste figure is about $1.5 billion. It is not the full $2.6 billion, because $1.1 billion of that carried forward. And it is not the $7.6 billion, which was an estimate of a finished program that was never built as designed. ## The contractor chain One more detail is easy to get wrong: who was actually running the program at each stage. IBM Federal Systems was the original prime contractor from the 1988 award. In 1994 IBM sold its federal-systems business to Loral, and the government, IBM, and Loral entered a novation agreement under which Loral assumed IBM's rights and claims under the AAS contract. In 1996 Lockheed Martin acquired Loral's division, and Lockheed Martin executed the salvaged DSR work. That means it is inaccurate to call it the IBM program at the completion stage, or the Loral program at the 1988 award. The prime changed hands twice by acquisition and novation while the underlying contract continued. ## The failure critique, stated plainly By the measure GAO applied, AAS is a textbook government IT failure. A program estimated at $2.5 billion in 1983 was projected at as much as $7.6 billion by the time it was restructured in 1994, and of the roughly $2.6 billion actually spent, about $1.5 billion was written off. GAO attributed the collapse to an overly ambitious development plan, inadequate FAA oversight of contractor software development, and continually changing system requirements. Key segments slipped 3.5 to 8 years against their 1988 baselines. The Area Control Computer Complex was cancelled in 1994, and the Tower Control Computer Complex was scaled back and then cancelled in 1997. AAS became a recurring example in critiques of FAA modernization and federal IT management, and it is part of why FAA and DOT modernization drew sustained oversight scrutiny. None of that is softened by the second half of the story. The waste was real, it was documented by two independent government auditors, and it was large. ## The mission defense, stated just as plainly The other half is also real. Replacing 20-to-30-year-old, safety-critical air-traffic-control computers for the world's busiest airspace is genuinely hard. The software had to be real-time and fail-safe, and it had to be usable by controllers across the en route centers handling tens of millions of flights a year, all while requirements evolved as the mission did. The 1994 restructuring was a course correction, not just a cancellation. The FAA moved from an all-at-once grand design to incremental delivery of less-costly alternatives, and that salvaged work paid off. The scaled-back Display System Replacement, the renamed ISSS segment, was awarded in April 1995. According to a Lockheed Martin press release, DSR went operational at the Seattle Center on December 15, 1998, and was installed at all 20 en route centers by May 31, 2000, ahead of schedule and within the FAA's established budget. That ahead-of-schedule and on-budget characterization comes from the contractor, and it should be read as contractor-sourced; GAO tracked DSR implementation independently in report T-RCED-00-19. DSR gave controllers modern color displays and a new software architecture. The terminal-modernization mission continued through STARS. And about $1.1 billion of the money spent was salvageable rather than lost. So both verdicts sit side by side, and neither cancels the other. AAS wasted about $1.5 billion on a grand design that could not deliver, and it also funded laboratory work and hardware that carried into a restructured program that did reach every en route center. The honest lesson is not that the mission was pointless. It is that the way the mission was structured, as a single sweeping replacement with shifting requirements and thin oversight of software development, is what produced the write-off, and that the pivot to incremental delivery is what eventually shipped. ## Fact-check notes and sources - 1983 development estimate of about $2.5 billion, completion 1996: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm), which states the FAA estimated the cost to develop AAS at $2.5 billion with completion scheduled for 1996. This is a development-cost estimate, not full lifecycle. - 1988 IBM contract award after a roughly four-year competition, estimate $4.8 billion, completion 1998: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm). - December 1993, roughly $1.2 billion in overruns and an FAA-ordered review of the IBM-affiliated project: [UPI archive, Dec 13, 1993](https://www.upi.com/Archives/1993/12/13/Amid-12-b-of-overruns-FAA-orders-review-of-IBM-affiliated-project/9222755758800/). This is a contemporaneous pre-restructuring figure and is the weakest-sourced claim here; the archive page currently returns an access error, and the figure rests on the indexed headline. It is distinct from the later $1.5 billion write-off. - 1994 restructuring estimate of as much as $7.6 billion, completion as late as 2003: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm). GAO's own summary text rounds this to as much as $7 billion; the detailed body states $7.6 billion. This is a projected estimate, never a spent amount. - About $2.6 billion spent through May 1994: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm), citing the FAA's investment analysis group. - Of that spend, about $1.1 billion salvageable and about $1.5 billion wasted: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm), corroborated by the [DOT OIG audit AV-1998-113](https://www.oig.dot.gov/sites/default/files/av1998113.pdf). The $1.5 billion is the written-off portion, not the total spent and not the projected estimate. - Schedule slippage of 3.5 years to as much as 8 years against 1988 baselines: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm). - Segment fates (ISSS scaled back and renamed DSR; Terminal AAS cancelled and carried forward by STARS; Area Control Computer Complex cancelled in 1994; Tower Control Computer Complex scaled back in 1994 and cancelled in 1997; PAMRI unaffected and operational by 1993): [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm). Contemporaneous testimony on the restructuring appears in [GAO T-RCED-94-188](https://www.gao.gov/products/t-rced-94-188). - Contractor chain, IBM to Loral by novation in 1994, Lockheed Martin acquiring Loral's division in 1996: [GAO T-RCED/AIMD-98-85](https://www.govinfo.gov/content/pkg/GAOREPORTS-T-RCED-AIMD-98-85/html/GAOREPORTS-T-RCED-AIMD-98-85.htm), with the novation-agreement term documented in [DOT OIG AV-1998-113](https://www.oig.dot.gov/sites/default/files/av1998113.pdf). - DSR contract awarded April 1995, operational at Seattle Center December 15, 1998, installed at all 20 en route centers by May 31, 2000, ahead of schedule and on budget: [Lockheed Martin press release, July 2000](https://news.lockheedmartin.com/2000-07-14-Lockheed-Martin-Completes-Final-Air-Route-Traffic-Control-Center-Installation-Ahead-of-Schedule). The ahead-of-schedule and on-budget framing is contractor-sourced; GAO independently tracked DSR status in [GAO T-RCED-00-19](https://www.gao.gov/products/t-rced-00-19). ## Related reading - [DIMHRS military HR and payroll failure](/blog/dimhrs-military-hr-payroll-failure/): another grand-design government IT program whose scope outran its delivery. - [Air Force ECSS ERP failure](/blog/air-force-ecss-erp-failure/): a large enterprise-software effort cancelled after heavy spending, with a comparable write-off question. - [FBI Virtual Case File and Sentinel](/blog/fbi-virtual-case-file-sentinel/): a first attempt that collapsed and a restructured second attempt that shipped, an arc that rhymes with AAS. - [The GAO High-Risk List and improper payments](/blog/gao-high-risk-list-improper-payments/): the oversight lens, and why programs like AAS drew sustained GAO scrutiny. - [The public-money programs index](/blog/public-money-programs-index/): the full set of where-the-money-goes explainers in this series. *This post is informational and journalistic, not legal or financial advice. It describes public programs and documented events; mentions of third parties are nominative fair use and no affiliation is implied.* --- Canonical HTML: https://jwatte.com/blog/faa-advanced-automation-system/ RSS: https://jwatte.com/feed.xml JSON Feed: https://jwatte.com/feed.json Hero image: https://jwatte.com/images/faa-advanced-automation-system.webp