Katerra: The Construction Company That Called Itself Tech

A rendering on a projector in a Menlo Park conference room, mid-2016. A modular apartment building in cross-laminated timber, five stories tall, factory-built in panels, delivered to site by truck, assembled in six weeks by a small on-site crew. The rendering shows the whole thing at 30 percent lower cost per square foot than a traditional stick-frame apartment building at the same size in the same city. The pitch is that construction is the last major industry Silicon Valley has not yet disrupted, that the industry runs on 18th-century labor practices and 19th-century supply chains, that the productivity has not moved in fifty years, and that a vertically integrated company owning architecture, engineering, manufacturing, and on-site assembly could take 20 to 30 percent out of the cost curve of building a mid-rise apartment complex. The people in the room have the credentials to believe the pitch. Michael Marks ran Flextronics for 13 years and grew it from $93 million in revenue to $16 billion. Fritz Wolff runs a large private real estate family office. Jim Davidson co-founded Silver Lake. Three men with more than $100 billion of pooled operating and investment history between them are looking at a rendering and seeing the future of construction.

Six years later Katerra would file for bankruptcy and leave $1.29 billion in unpaid bills to the contractors who had actually done the building.

On June 6, 2021, Katerra Inc. and 32 affiliated debtors filed voluntary petitions for relief under Chapter 11 of the United States Bankruptcy Code in the United States Bankruptcy Court for the Southern District of Texas, Houston Division. The lead case was Case No. 21-31861, jointly administered under Judge David R. Jones. The filings listed cumulative reported losses of $2.78 billion across 2018, 2019, and 2020 alone. Approximately 730 US employees had already been laid off by early 2021 out of a US workforce of 1,300. Eighty-two projects had already been shut down before the petition. SoftBank had invested more than $2 billion. Michael Marks had already been ousted as CEO the prior year. Paal Kibsgaard, the former Schlumberger CEO who had replaced Marks in June 2020, was overseeing an orderly wind-down.

This is the story of how three private-equity operators in Menlo Park set out in 2015 to build the Flextronics of construction. Of how SoftBank's Vision Fund gave them $865 million in January 2018 at a valuation over $3 billion. Of how they acquired more than 20 subsidiaries in three years, opened factories in Tracy, California and Spokane, Washington, hired 8,000 employees globally, won a contract to build 14,000 housing units for the Saudi Arabian Housing Authority. And of how the vertical-integration thesis, correct in general electronics manufacturing, ran into a construction industry whose economics, workforce practices, and site-by-site variation the thesis did not fit. The lender collapse of March 2021 was the trigger. The doctrine had been reading the ending since Katerra passed $500 million in annualized losses without ever posting a profitable operating quarter.

This is the rise and fall of Katerra.

The verdict. Katerra did not die from Greensill Capital's insolvency in March 2021. Katerra died from the strategic decision made between 2015 and 2018 to build a vertically integrated construction company at Silicon Valley speed, funded by Silicon Valley capital, on the theory that construction was electronics manufacturing at a different scale. The theory was wrong on three specific counts. First, electronics manufacturing runs on identical units at industrial scale, and construction runs on custom units at site-specific scale. Second, electronics manufacturing labor is offshored to low-cost jurisdictions, and construction labor is unionized and geographically fixed. Third, electronics manufacturing has a supply chain designed around global consolidation, and construction has a supply chain designed around local relationships that resist national consolidation. Katerra spent $2 billion of SoftBank's money trying to force construction into an electronics-manufacturing operating model. The Fixed Cost Capacity biomarker was terminal by 2019. The March 2021 Greensill collapse was not the fracture. It was the moment the capital tap turned off on a business that had never posted a profitable quarter. Every dollar of the $2 billion was funding the difference between the operating loss and the promised operating profit. When the funding stopped, the operating loss showed up on the balance sheet as immediate insolvency.

The Rise

Michael Marks did not start Katerra to disrupt construction. He started Katerra because Fritz Wolff, his longtime friend and business partner, told him construction needed disruption and Marks believed the credential he had built at Flextronics translated.

Marks was 65 years old in 2015. He had spent 13 years as CEO of Flextronics, taking the company from an $8.5 million contract manufacturer on the verge of bankruptcy in 1991 to a $16 billion global electronics manufacturer by 2005. Under Marks, Flextronics had built factories in Mexico, China, Hungary, Malaysia, and Brazil. It had manufactured Xbox consoles for Microsoft, printers for HP, and mobile handsets for Sony Ericsson. It had operated on gross margins of six to eight percent and net margins of two to three percent, which is what a vertically integrated contract manufacturer running on scale efficiency looks like when it works. Marks was recognized as one of the industry's most skilled operators. Fortune had called him a manufacturing hero. He had briefly served as interim CEO of Tesla in 2007. He had then joined KKR as a senior advisor for two years. In 2007 he co-founded Riverwood Capital, a Menlo Park-based private equity firm focused on rapidly growing companies. By 2015 he was 66 years old, wealthy, credentialed, and looking for a next act at the scale of the first one.

Fritz Wolff was executive chairman of The Wolff Co., a Scottsdale, Arizona-based multifamily real estate investment company his family had run for generations. Wolff and his family had been developing and operating apartment communities for decades. He knew the construction cost curve for multifamily development intimately. He knew that a mid-rise apartment building in 2015 cost about the same per square foot to build as it had in 2005, adjusted for inflation, while the underlying materials and technology had improved substantially over that decade. Where did the productivity gain go? In Wolff's view, it did not go anywhere. It never showed up. Construction had missed the productivity revolution that had transformed every other American industry between 1970 and 2010.

Jim Davidson, the third co-founder, brought private-equity credibility. Davidson had co-founded Silver Lake in 1999 with Glenn Hutchins, David Roux, and Roger McNamee. Silver Lake by 2015 had grown to become one of the largest technology-focused private equity firms in the world, managing tens of billions of dollars across multiple funds. Davidson had done technology and industrial deals for decades. He believed the same operator-plus-capital thesis Marks and Wolff believed. If you brought Flextronics discipline to construction, you would get Flextronics results.

The three of them founded Katerra in Menlo Park in 2015. The founding thesis was straightforward. Construction was fragmented. Architecture firms designed buildings and handed them off to engineering firms. Engineering firms handed them off to general contractors. General contractors hired sub-contractors. Sub-contractors managed labor. Materials came from suppliers. Everything was project-based. Nothing was standardized. Every project was a one-off. Every project ran over budget and over schedule. The productivity of the average American construction job in 2015 was not measurably better than the productivity of the same job in 1990. A vertically integrated company that combined architecture, engineering, off-site factory manufacturing of building components, and on-site assembly under one corporate roof could compress the timeline, reduce the cost, and produce buildings faster, cheaper, and better than the fragmented industry it competed with.

The thesis was elegant. The credentials were impressive. The capital was available. Between 2015 and early 2018, Katerra raised roughly $1 billion from a mix of institutional investors, real estate developers, and technology venture funds. Marks served as CEO. Wolff served as chairman. Davidson served on the board. They hired an A-list leadership team from architecture, construction, engineering, and technology backgrounds. They acquired subsidiaries at pace. Michael Green Architecture, the leading architectural firm in mass timber construction, was acquired in 2018. Equilibrium, a structural engineering firm specializing in tall wood buildings, was acquired the same year. More than twenty other companies would be absorbed over the following three years, including firms in prefabrication, HVAC, and specialty engineering.

In January 2018, SoftBank Vision Fund led an $865 million investment in Katerra at a valuation above $3 billion. It was one of the largest single investments in construction technology ever made. The Vision Fund would ultimately provide more than $2 billion of Katerra's total capital across the next three years. Marks appeared on the cover of trade publications. McKinsey interviewed him about the future of construction. The Wall Street Journal profiled the company. Katerra was, in the trade press of 2018 and 2019, the definitional example of construction technology done at scale.

And it was already, in the general ledger, on fire.

The Fracture

To understand what killed Katerra, you have to understand what a vertically integrated construction company actually looks like when the vertical integration hits the construction industry.

The thesis Marks brought from Flextronics was that vertical integration reduces cost by eliminating the margin stack. In electronics manufacturing, a client design goes through a chain of vendors, each of which takes a margin. A vertically integrated contract manufacturer captures that margin by doing everything under one roof. This works in electronics because the units being manufactured are identical, the production runs are at industrial scale, and the labor can be geographically consolidated in low-cost jurisdictions. Flextronics ran million-unit production runs of identical products at a Chinese factory with predictable labor costs and predictable material costs.

Katerra tried to apply the same operating model to construction. And construction did not cooperate.

The first failure of the analogy is unit identity. Every building Katerra built was different. A mid-rise apartment building in Phoenix is not the same as a mid-rise apartment building in Denver. Zoning is different. Building codes are different. Seismic requirements are different. Fire suppression is different. The soil conditions are different. The local labor market is different. The delivery logistics are different. What Katerra was calling "modular construction" required custom engineering for each project even when the physical components were standardized. The engineering time did not compress the way the manufacturing thesis assumed. The design-to-delivery timeline stayed roughly the same as traditional construction. The promised productivity gain did not appear.

The second failure was labor. Flextronics ran on offshored labor. Katerra could not offshore construction labor. Every building sits somewhere. That somewhere has a labor market and, in most major American markets, a unionized labor market with jurisdictional rules that govern which crafts perform which work. Katerra opened factories in Tracy, California and Spokane, Washington to produce building components. Those factories had their own labor forces, largely non-union, running on Katerra's payroll. But the components had to be delivered to sites where the on-site assembly work was done by local crews, many of them union. Katerra effectively created a labor arbitrage between its own factory workers and the local construction unions, and the arbitrage generated persistent friction on every project. Union crews refused to assemble Katerra components without union labor credit. Sites where Katerra used its own crews faced pickets and grievances. The projected labor savings were consumed by the disputes.

The third failure was the supply chain. Katerra had tried to build a national supply chain of standardized building components. Windows. Cabinets. HVAC systems. Cross-laminated timber panels. Doors. Framing kits. The theory was that national consolidation of these components would produce Flextronics-style purchasing power. In practice, construction supply chains are intensely local. A window from a Katerra factory in Tracy shipped to a site in Miami costs substantially more in freight than a locally-sourced window would cost, and the freight cost consumed the volume discount. Sites in states adjacent to Katerra factories saw some cost savings. Sites more than 500 miles from a factory saw net cost increases. The economics never scaled the way electronics component sourcing had scaled at Flextronics.

While these three structural failures ran on every project, Katerra's cost base grew. The factories in Tracy and Spokane required substantial capital investment and ongoing operating cost. The architecture and engineering firms Katerra acquired came with corporate overhead. The A-list leadership team from Silicon Valley came with Silicon Valley compensation packages. The Menlo Park headquarters ran on Silicon Valley office rent. Marc Liebman, who joined the leadership team, would later testify in bankruptcy court that Katerra had also offered "significant discounts to certain legacy customers." In hindsight, he stated, "those discounts were value-destructive." The company was underpricing its own product to win reference projects, then losing money on those projects because the underlying construction economics did not deliver the promised cost curve, then eating those losses in the P&L.

By 2019, according to filings, Katerra had passed $500 million in annualized losses without ever having posted a profitable operating quarter. The cost base was north of $1 billion a year. The revenue was climbing but never fast enough to close the gap. The Business Biomarker Index composite score, if Katerra had ever run one, would have been terminal by early 2019. The company was scaling revenue without operating profit, which the doctrine reads as scaling losses. Every additional dollar of revenue was matched by more than a dollar of cost. Growth was consuming capital, not creating it.

SoftBank continued to fund the difference. That is what a growth-stage Silicon Valley investor does when portfolio companies burn cash. The theory is that scale will eventually produce the operating profit the current losses do not show. The theory is sometimes correct in software businesses where marginal cost approaches zero. The theory is systematically wrong in physical goods manufacturing at construction scale, where marginal cost is stubborn and volume produces at best incremental savings against a fixed cost base that continues to grow.

In December 2019, Katerra announced its first major restructuring. It would close its Phoenix, Arizona factory. It would lay off approximately 200 employees. It would consolidate manufacturing at Tracy, California, where the factory was more automated and costs were lower. The trade press reported this as a rationalization move. The doctrine reads it as the first admission that the original operating model did not scale.

In May 2020, in the middle of the pandemic, Katerra announced Michael Marks was stepping down as CEO. Paal Kibsgaard, former CEO of Schlumberger and Katerra's COO, would take over. The announcement was paired with a $200 million capital raise. The trade press reported this as a natural leadership transition. The doctrine reads it as the point at which Marks recognized the operating model could not be fixed by the founder who had built it. Handing the wheel to Kibsgaard, whose background was in oilfield services rather than construction, meant Marks was no longer betting his personal credential on the operating recovery.

In November 2020, Katerra came close to a recapitalization deal in which existing stakeholders including SoftBank plus new investors would inject $380 million in exchange for 90 percent of the company. The deal broke down.

In December 2020, SoftBank alone provided a $200 million bailout, of which $175 million was cash and $25 million was forgiveness of an existing loan. In the same month SoftBank forgave a $440 million loan Katerra owed to Greensill Capital, which was another SoftBank portfolio company. That $440 million loan forgiveness would later become the subject of a lawsuit by Credit Suisse, whose clients had bought that debt from Greensill and expected to be repaid.

By early 2021 Katerra had four options and was working all of them. Sell assets. Raise new capital. Find a strategic acquirer. Explore a partial or full bankruptcy filing. The company had hired Houlihan Lokey in September 2020 to explore strategic alternatives. It was pursuing a proposed $147 million investment from the Public Investment Fund of Saudi Arabia in exchange for a 49 percent stake in Katerra Saudi Arabia. It was talking to third-party lenders including Kevin Genda's Blue Torch Capital. None of it was closing.

Then Greensill collapsed.

The Greensill Trigger

In early March 2021, Greensill Capital, the London-based supply-chain finance firm founded by Lex Greensill, became insolvent. Greensill had been a principal lender to Katerra. It had also been another portfolio company of SoftBank's Vision Fund, and its collapse rippled through the Vision Fund's other portfolio companies in ways that took several months to fully manifest. Credit Suisse, which had bundled Greensill's loan receivables into supply-chain finance funds sold to institutional investors, was left with billions of dollars in exposed positions. The Wall Street Journal reported that SoftBank had provided funds to Greensill expecting the money to flow through to Credit Suisse's investors, but the money had not made it.

For Katerra, the direct consequence of Greensill's insolvency was that its principal lending relationship disappeared overnight. Any receivables Katerra had been factoring through Greensill were immediately frozen. Any project financing that had been arranged through Greensill supply-chain products was suddenly at risk. And, more consequentially, the broader lending market saw Katerra as damaged goods and became reluctant to extend replacement credit.

The banks and insurance companies that had been willing to work with Katerra before Greensill's collapse became reluctant to work with Katerra after. Bonding companies, which underwrite the surety bonds that construction projects require, became more cautious. Katerra's clients, watching from outside, began preparing contingency plans. Some clients started replacing Katerra as general contractor on active projects. Others withheld payments pending clarity on Katerra's continued solvency. The customer confidence that had sustained the company through eighteen months of losses evaporated in eight weeks.

The Public Investment Fund of Saudi Arabia deal, which had been expected to close in early 2021 with $147 million of investment, did not close. Katerra had to use its own $23 million of cash to meet the credit facility obligations that the Saudi money was supposed to have funded. The 14,000-unit Saudi housing contract that Katerra had won was suddenly in doubt.

In mid-May 2021, Katerra hired Matthew Niemann of Houlihan Lokey as restructuring consultant. Niemann approached the original founders. Marks, Wolff, and Davidson had all left operational roles by this point. None of them wanted to inject additional capital. Niemann approached other stakeholders. He approached third-party lenders. Every rescue option was declined. He came close to a deal with Kevin Genda's Blue Torch Capital, but Blue Torch ultimately decided against it. Niemann later described Katerra's cash position as "perilous" and said the company would be "dead" without the immediate release of $15 million from a bankruptcy financing loan.

SoftBank told Katerra it "could not reasonably invest additional funds." SoftBank agreed to provide $35 million to help sell the assets and wind down US operations. Katerra had requested $50 million.

On Sunday, June 6, 2021, Katerra Inc. and 32 affiliated debtors filed voluntary petitions for Chapter 11 relief in the United States Bankruptcy Court for the Southern District of Texas, Houston Division, jointly administered as Case No. 21-31861. The cases were assigned to Judge David R. Jones. The next afternoon, in a virtual hearing, Judge Jones provisionally approved the $35 million debtor-in-possession financing from SoftBank and released $15 million of it for immediate operating needs. SoftBank's own counsel, Alfredo Perez of Weil Gotshal & Manges, told the court "we are very much a reluctant DIP lender here. It's not the position that we wanted to find ourselves in."

Over the following weeks and months, Katerra's assets were sold in tranches. Volumetric Building Companies acquired the Tracy factory assets. Michael Green Architecture and Equilibrium were transferred back to their original principals. Other subsidiaries were spun back to their founders. The Wolff Co., the family real estate firm Fritz Wolff had co-founded with Katerra, received a first-lien security interest in the Spokane cross-laminated timber factory as part of the pre-bankruptcy recapitalization structure. Contractors who had been owed money by Katerra filed proofs of claim totaling approximately $1.29 billion. Most would recover cents on the dollar.

The 14,000-unit Saudi housing contract was left in ambiguous status. The Saudi Arabian Housing Authority was left to find alternative delivery arrangements.

The Doctrine Overlay

Which capacity broke. Fixed Cost Capacity broke first, and Working Capital Capacity broke downstream of it. Katerra's Fixed Cost Capacity by 2019 included two operating factories in Tracy and Spokane, a substantial Menlo Park corporate headquarters, more than twenty acquired subsidiaries with their own overhead structures, an A-list leadership team on Silicon Valley compensation, and a Saudi Arabia office building out a 14,000-unit contract. That fixed obligation ran to more than $1 billion a year. Working Capital Capacity was structurally weak because the construction operating cycle is long. From bid to project completion to final payment is often 18 to 30 months on mid-rise multifamily projects, and Katerra was funding materials, labor, and factory operating costs across that entire window. The doctrine says Working Capital Capacity has to be sized to the operating cycle. Katerra sized its Working Capital Capacity to the availability of SoftBank funding, not to the operating cycle of its actual business. When SoftBank funding stopped, the Working Capital Capacity biomarker read terminal within weeks.

Which layer of the cake collapsed. Layer 1 Revenue was climbing but never fast enough to close the operating loss. Layer 3 Gross Margin was structurally negative on most projects because of the "value-destructive" discounting Marc Liebman later described. Layer 4 Overhead was oversized for the revenue base by roughly 3x industry standard for construction of Katerra's scale. Layer 5 Debt Service was manageable in isolation, becoming terminal only in the March 2021 credit-freeze phase. The layer that actually collapsed first was Layer 3 Gross Margin, and it collapsed on the value-destructive discounts, and every subsequent decision was downstream of a negative-margin operating base that no amount of scale was going to convert into positive-margin operating base. The Layer Cake reads bottom-up. Katerra was serving Layer 5 through Layer 1 backward, using capital raised in Layer 5 to fund losses in Layer 1 without ever building a Layer 3 that produced positive Layer 2 residual. The stack was inverted from day one.

Which sub-layer of Minimum Mandatory Profit got starved. Every sub-layer. Katerra had no MMP by design. It was a growth-stage company burning venture capital on the theory that scale would eventually produce operating profit. That theory is the specific Silicon Valley doctrine that most SMB owners are trained by media coverage to admire. The Aldebert Doctrine does not admire it. The Aldebert Doctrine says every operating business, at every stage, must be sized to its own MMP. When a business is scaled beyond its MMP capacity by external capital, the business is operationally insolvent regardless of how large the external capital position is. Katerra was operationally insolvent every quarter of its existence. The insolvency was masked by SoftBank funding. When the masking stopped, the insolvency became visible as bankruptcy within 90 days.

Where the diagnostic would have flashed. Not in March 2021. The Greensill collapse was not the moment the diagnostic would have flashed. The diagnostic would have flashed in Q4 2018, roughly nine months after the SoftBank Vision Fund investment. Return to Owner would have read Katerra's operating numbers against its cost base and produced a single sentence. Your gross margin is negative on 60 percent of your active projects, your annualized fixed cost is 3x the industry standard for construction companies of your revenue scale, and your operating loss is running at approximately $500 million a year against reported revenue of approximately $1.2 billion. Your capital position at present funding cadence supports approximately 24 to 36 months of continued operation. Your Fixed Cost Capacity biomarker is in the Failure band. The Business Biomarker Index composite score is in the Failure band. No amount of additional capital investment will convert a negative-margin operating base to a positive-margin operating base without a structural change in your unit economics. The structural change requires either a fundamental reduction in your fixed cost base or a fundamental change in your gross margin per project. If neither structural change is made within the next 12 to 18 months, the business will file for bankruptcy protection within approximately 30 months. That flash was missed. Or rather, it was seen by the operators inside the company and dismissed by the capital providers outside the company who continued funding the difference. By the time the capital providers stopped, the operating window had closed. The March 2021 Greensill collapse was not the cause. It was the specific event that removed the mask.

The Flextronics analogy that killed the company. Michael Marks brought a genuinely powerful operating credential to Katerra. Under Marks, Flextronics scaled a contract electronics manufacturing business from $8.5 million to $16 billion in a decade. The scaling worked because electronics manufacturing has specific structural features that support vertical integration at industrial scale. Katerra assumed those features would carry over to construction. They did not. The doctrine reads Marks's error as the specific failure of pattern-matching across industries whose operating structures look superficially similar but are fundamentally different. Every subsequent decision Katerra made was downstream of the founding pattern-match. Acquire more subsidiaries. Build more factories. Hire more Silicon Valley talent. Discount project prices to win reference customers. Take more SoftBank money. Each of these decisions made sense inside the Flextronics analogy. None of them made sense inside the actual construction industry. The doctrine has one non-negotiable rule for any operator entering a new industry on the strength of credentials from another industry. Prove out the unit economics on a small number of pilot projects before scaling the fixed cost base. Katerra did the opposite. It scaled the fixed cost base to $1 billion a year before proving out unit economics on a single profitable project.

The pattern that connects Katerra and WeWork. This is the second Autopsy in the archive to name a Silicon Valley venture capital thesis colliding with a physical-goods industry as the specific mechanism of collapse. WeWork was a commercial real estate business that Adam Neumann and SoftBank successfully marketed as a technology company for a decade before the S-1 filing in August 2019 revealed the operating reality. Katerra was a construction company that Marks, Wolff, and Davidson successfully marketed as a technology company for six years before the June 2021 Chapter 11 filing revealed the operating reality. Different industries, same doctrine surface. When a physical-goods business is capitalized and operated on Silicon Valley scaling assumptions, the business burns capital faster than the operating model can generate returns, and eventually the capital tap turns off. The doctrine reading in this Autopsy names the specific pattern. If your business is called a technology company but its actual product is a physical thing that has to be manufactured, delivered, installed, or built, the technology framing is marketing. The underlying business obeys the operating economics of the physical-goods industry it lives in. Katerra was a construction company. WeWork was a commercial real estate company. Both of them died on operating economics their Silicon Valley framing was designed to obscure.

The Intervention

There were four specific moments where the doctrine could have caught this. Each required somebody with authority to name what the Fixed Cost Capacity breach actually was.

The first was 2017, before the SoftBank Vision Fund investment closed. Katerra at that point was operating on $500 million of previously raised capital and building out its first factory. If Marks or Wolff had committed to running the first factory to full utilization on a profitable set of pilot projects before accepting the $865 million SoftBank check, the underlying unit economics would have been forced into visibility within 18 months. Either the factory would have produced positive gross margin at scale, in which case the Vision Fund investment would have been rational, or the factory would have produced negative gross margin at scale, in which case the entire vertical-integration thesis would have been falsified and the Vision Fund investment either restructured or declined. Marks and Wolff did not run this test. They took the SoftBank money and used it to accelerate acquisition and factory build-out, on the theory that scale itself would produce positive unit economics eventually. It did not.

The second was Q4 2018, roughly nine months after the Vision Fund investment. By this point the operating losses were clearly visible on the internal financial statements. The board had the information needed to conclude that unit economics were not scaling to positive. A responsible board would have paused acquisition activity, paused factory expansion, and forced a 12-month operating cycle at existing scale to prove out gross margin. If gross margin remained negative through the 12-month operating cycle, the board would have needed to either restructure the operating model dramatically or wind down the business while capital reserves were still sufficient to pay contractors and employees. Neither happened. The board continued approving acquisitions and factory expansions on the theory that additional scale would resolve the unit economics.

The third was May 2020, at the CEO transition. Marks stepping down and Kibsgaard taking over could have been the moment to run a clean restructuring. Kibsgaard, coming from Schlumberger, had experience running large industrial operations at operational profitability. A restructuring at CEO transition, before any December 2020 rescue funding, would have preserved substantially more optionality than the eventual June 2021 filing did. Instead, Kibsgaard took over paired with a $200 million capital raise that extended the runway for another 12 months. The extension bought time. It did not fix the underlying operating model.

The fourth was November 2020, when the proposed $380 million recapitalization deal was under negotiation. If the deal had closed with new investors taking 90 percent and management taking 10 percent, and if the new investors had insisted on immediate restructuring conditions attached to the capital, Katerra might have survived as a smaller, focused business. The deal did not close. SoftBank instead provided a smaller $200 million alone, without the restructuring conditions the broader deal would have carried. The Greensill collapse followed three months later.

The Lesson For SMB Owners

Katerra is not a story about big-company hubris that does not apply to SMB owners. Katerra is the story of what happens when an operator with credentials in one industry tries to scale a business in a different industry on the strength of those credentials, backed by capital that removes the discipline of proving out unit economics before growing the fixed cost base.

This happens at SMB scale constantly, in a specific pattern the doctrine sees every year. An operator with a successful track record in one industry, often a related industry, buys or starts a business in a new industry on the theory that the operating discipline transfers. A commercial HVAC contractor buys a plumbing service company on the theory that home services all work the same way. A residential remodeler expands into commercial construction on the theory that the underlying skills transfer. A trucking fleet operator buys a warehousing and logistics business on the theory that both are "supply chain." Each of these expansions is defensible on paper. Some of them succeed. The ones that fail almost always fail on the same mechanism as Katerra. The operator scales the fixed cost base of the new business on the assumption that the operating discipline from the original business will produce the same gross margin. It does not. The new industry has different unit economics, different labor markets, different customer relationships, different supply chain, different regulatory constraints. The operator is 12 to 24 months into the expansion before recognizing that the promised gross margin is not appearing. By then the fixed cost base of the expansion has already committed. The operator now has a personal guarantee on the expansion debt, a personnel roster that has to be maintained, a facility lease that has to be paid, and no gross margin to cover any of it.

The specific SMB version of the Katerra pattern is when the successful operator gets access to capital that removes the discipline the original business built. A residential general contractor who ran a healthy $2 million-a-year residential business for fifteen years buys a struggling small commercial GC on the strength of a bank loan, expands the crew from 8 people to 22 people to service the commercial book, and 18 months later realizes commercial construction margins do not support the expanded crew. The original residential business had a Working Capital Capacity biomarker in the healthy band. The expanded business has a Working Capital Capacity biomarker in the Failure band. The operator is now personally guaranteed on an SBA loan and a crew payroll he cannot make from residential margins alone. Everything the operator had built over fifteen years is now collateral for an expansion he executed in eighteen months.

The doctrine has one non-negotiable rule for any expansion into a new industry, regardless of how related the industry looks on paper. Prove out unit economics on a small number of pilot engagements before scaling the fixed cost base. Not by taking on one small project as an experiment while the main business subsidizes the exploration. By running three to five real, priced, at-scale engagements in the new industry with full accounting for labor cost, material cost, overhead absorption, and project completion timelines. If the pilot engagements produce positive gross margin at industry-standard pricing, the expansion is validated. If the pilot engagements produce negative gross margin, the expansion is a Fixed Cost Capacity breach in progress. Either accept the validation, restructure the expansion, or walk away.

The second lesson is about capital that removes discipline. Katerra had $2 billion of SoftBank money available. That was the problem. If Katerra had operated on $50 million of capital, the operators would have been forced to prove out unit economics on a small number of pilot projects because the alternative was running out of cash. The $2 billion removed the forcing function. The forcing function is not an inconvenience of small-scale operating. The forcing function is the mechanism by which unit economics get proved out. When capital removes the forcing function, operators build businesses that cannot survive without continuous capital infusion. Every SMB owner who has ever accepted a growth investment from a family office, a private equity firm, or a friendly bank should understand this. Capital that removes forcing functions is not a gift. It is a debt payable in fixed cost commitments the operating business has to eventually service.

The move this week. If you have expanded into a new industry, service line, or geography in the last three years, pull the last four completed engagements in the new segment. Compute gross margin on each after full labor absorption, material cost, overhead allocation, and project management time. Compare that gross margin to the gross margin in your original business at similar revenue scale. If the new segment gross margin is more than 30 percent below the original segment gross margin, the expansion is not delivering the unit economics required to support the fixed cost base it has committed. Options at that point are to reduce the fixed cost base in the new segment back into range with the actual gross margin the segment produces, to raise pricing in the new segment enough to close the gross margin gap, to accept the new segment as a strategic loss leader with a specific dollar limit on how much loss you will tolerate before shutting it down, or to wind down the new segment while the original business still has capacity to absorb the wind-down costs. If you have not expanded but are considering it, run the same unit economics stress test on the target segment before committing any fixed cost. Katerra's board did not run the stress test in 2017. Katerra's board should have.

Run Return to Owner on your actual numbers. Read your Fixed Cost Capacity biomarker against your segment-level gross margins. And ask the question Katerra's board never asked in 2017 or 2018. If the promised operating synergies from my industry expansion do not appear at the promised timeline, does my current fixed cost base still support the original business, or does the expansion take down both.

Postscript

Michael Marks was ousted as Katerra CEO in June 2020 and left the board shortly after. In 2021, the same year Katerra filed for Chapter 11, Marks quietly founded a new company in Florida called ONX Homes. ONX Homes is a technology-enabled construction company focused on prefabricated single-family and multi-family housing. Marks serves as chairman. The company has raised additional venture capital. Whether the doctrine lessons from Katerra will apply to ONX Homes remains to be seen. The mechanism is the same industry with the same structural features that killed Katerra. The question is whether the second attempt has learned what the first attempt did not.

Fritz Wolff continues to run The Wolff Co. as executive chairman. The Wolff Co. received a first-lien security interest in the Spokane cross-laminated timber factory as part of the pre-bankruptcy recapitalization, giving the Wolff family a priority claim on those assets during the bankruptcy sale process. The optics of a co-founder receiving priority creditor status on the way into the filing were noted at the time.

Jim Davidson continues to co-lead Silver Lake, one of the largest technology-focused private equity firms in the world. Silver Lake as a firm was not directly exposed to the Katerra collapse in any material way. Davidson's personal reputation absorbed a portion of the Katerra failure.

Paal Kibsgaard, the former Schlumberger CEO who took over as Katerra CEO in June 2020, departed with the bankruptcy. He has held various private-company advisory roles since. In 2023, Kibsgaard was named as a defendant in an adversary proceeding in the Katerra bankruptcy relating to the December 2020 recapitalization transactions. The litigation continues.

Deloitte & Touche LLP, Katerra's auditor, was named as a defendant in a separate adversary proceeding filed on December 31, 2022. That litigation also continues.

SoftBank's Vision Fund reported the Katerra loss as one of its largest single write-downs. The Vision Fund had also written down substantial positions in WeWork, Uber, and other portfolio companies during the same period. Masayoshi Son, the head of SoftBank, has publicly stated on multiple occasions that certain Vision Fund investments, without naming them specifically, represented the wrong bets at the wrong price. Katerra is widely understood to have been one of those bets.

The contractors who were owed $1.29 billion by Katerra at the time of the filing recovered a fraction of face value through the bankruptcy sale process. Many of the smaller contractors on Katerra's project list went out of business themselves as a result of the unpaid claims. The doctrine cannot recover those small businesses. The doctrine can only name the mechanism by which a $2 billion technology-enabled construction company left an $1.29 billion crater across the American construction industry, so the next owner considering a Silicon Valley expansion into a physical-goods business, or the next SMB owner considering a related-industry expansion of their own, knows what to watch for before the fixed cost base of the expansion commits.

The 14,000-unit Saudi Arabian Housing Authority contract was ultimately reassigned to other providers. The units, or some fraction of them, are being built by other companies using conventional construction methods. Katerra's mass timber and modular construction thesis, for the Saudi market as for the American market, ended before it could be tested at the scale the founders had envisioned.

The Tracy, California factory was acquired by Volumetric Building Companies out of the bankruptcy sale and continues to operate under new ownership. The Spokane, Washington cross-laminated timber facility continues to operate under Wolff Co. control. The intellectual property and design assets of Michael Green Architecture and Equilibrium reverted to their original principals and continue operating as independent firms. The vertical integration thesis, in that specific sense, ended in June 2021. The individual components live on in disaggregated form.

Silicon Valley continues to invest in construction technology. Several successor companies have raised meaningful venture capital pursuing variations of the Katerra thesis. None of them have yet scaled to Katerra's peak revenue. Whether any of them will eventually solve the operating economics of applying vertical integration to construction remains an open question. The doctrine reading in this Autopsy would predict that the answer is no, unless the successor companies begin from a fundamentally different assumption about unit economics than Katerra did. Katerra assumed unit economics would emerge from scale. The doctrine says unit economics have to be proven before scale, not through it. Every successor company that starts from Katerra's assumption is likely to end at Katerra's ending.

The doctrine cannot bring back the $1.29 billion Katerra owed to contractors or the small businesses that failed on the unpaid claims. The doctrine can only name the specific mechanism that killed a $3 billion-valued construction company in six years, so the next operator who inherits a credential from one industry and considers scaling into a different industry, or the next SMB owner considering a related-industry expansion, knows what to watch for the day the promised synergies fail to appear.