The Pitch and the Plumbing
Why technical due diligence has never been more urgently relevant, and why most of it is being done badly.
There is a number from SpaceX's June 2026 IPO that ought to make every investment committee pause.
SpaceX listed on 12 June 2026 at $135 per share, opened at $150, and now trades at roughly a $2.5 trillion market capitalisation, making it briefly larger than Meta (Seeking Alpha, 2026; Mexico Business News, 2026). The valuation arc is genuinely remarkable: $137 billion in January 2023, $350 billion in December 2024, $400 billion in July 2025, $800 billion in December 2025 through an insider share sale, and roughly $2.5 trillion three weeks after the listing. That is an 18x increase in equity value across 41 months for a company best known for rockets and satellite internet.
The interesting question is not whether the numbers are correct. They are. The interesting question is what the market is now paying for.
Pravin Pradeep, senior consultant at Frost & Sullivan, put it bluntly in a Via Satellite roundtable in early June: "The most surprising element is how strongly the IPO is framed as an AI story, even though it is being presented under the SpaceX umbrella. The company's proven businesses today are launch services and Starlink, but a significant portion of the valuation narrative is tied to more forward-looking opportunities including xAI, orbital compute, enterprise AI, and even space-based data centres" (Via Satellite, 2026). The S-1 prospectus presents a total addressable market estimate of $28.5 trillion for SpaceX's AI infrastructure strategy. That figure, as the same coverage notes, was produced by SpaceX itself and "has not been independently verified by third-party analysts" (Tech Jacks Solutions, 2026).
This is the central, uncomfortable truth of investing in 2026. Even the most sophisticated buyers in the world, the institutional investors and sovereign wealth funds taking down SpaceX paper at a $2.5 trillion valuation, are being asked to underwrite TAM figures that the company has produced about itself. The same is true at every scale below the trillion-dollar mark. And it is the reason technical due diligence, the unglamorous discipline of verifying what a company's technology can actually do, has become the most important defence the modern investor has.
The fraud at the other end of the spectrum
If SpaceX is the trillion-dollar version of the AI narrative problem, Builder.ai is the unicorn-scale version of the same phenomenon, and it ended very differently.
Builder.ai, the London based startup founded in 2016 as Engineer.ai, raised more than $450 million from Microsoft, SoftBank, the Qatar Investment Authority and the IFC on the promise of an AI platform that made software development "as easy as ordering pizza" (People Matters, 2025). At its peak it was valued at $1.5 billion. Its flagship AI assistant, called Natasha, was presented as the technology that orchestrated the entire development process.
In February 2025, founder Sachin Dev Duggal was replaced as CEO. An internal audit by his successor revealed the company had claimed $220 million in 2024 revenue against actual figures closer to $50 to $55 million, an alleged 300% inflation (Pro Pakistani, 2025). On 15 May 2025, creditor Viola Credit seized $37 million from Builder.ai's accounts after discovering covenant breaches tied to the overstated sales figures. Six days later, the company filed for insolvency across five countries (eDiscovery Today, 2025). Around 1,000 employees lost their jobs.
The board had operated for ten months without a CFO. The auditors had ties to the founder. Investigations into round-tripping schemes between Builder.ai and another company called VerSe Innovation suggested that both firms had invoiced each other for similar amounts without rendering services, artificially inflating revenue (eDiscovery Today, 2025). US prosecutors had ordered Builder.ai to hand over internal data in the weeks before its collapse, and the FBI demanded customer lists and accounting policies (Sifted, 2026). Duggal himself, who retained the title "Chief Wizard" even after his removal as CEO, has been named as a "key beneficiary" in an Indian Enforcement Directorate money-laundering investigation.
The most uncomfortable fact about the Builder.ai collapse is that the warning signs were public for six years before the failure. A Wall Street Journal investigation in August 2019 had already documented that the company's AI was largely a facade, with the actual development work being done by approximately 700 engineers in Delhi (Pragmatic Engineer, 2025). Six years of investor enthusiasm, $450 million of capital and a $1.5 billion valuation followed that investigation. The diligence that should have caught what the Wall Street Journal had already published either was not done or did not surface what was already in the public record.
What tech audits now has to do
For most of the past twenty years, technical due diligence has been a relatively predictable workstream in a broader M&A or PE process. It examined the technology stack, the cybersecurity posture, the data privacy compliance, the integration costs and the technical debt. The most common red flag in mid-market deals was legacy system dependency: a business running customer data on a 15-year-old ERP carried integration costs that belonged in the valuation model rather than the risk register (v7 Labs, 2026).
That model is no longer sufficient. The proliferation of AI-related claims in pitch materials means that tech DD now has to perform a second, harder function: verifying what a target's AI capability actually is.
Gartner's June 2025 research on agentic AI vendors is the most concrete data point available on the scale of the problem. Of the thousands of vendors marketing themselves as agentic AI providers, Gartner estimates that only around 130 are genuinely so. The remainder are engaged in what Gartner calls "agent washing", the rebranding of existing chatbots, robotic process automation and AI assistants as autonomous agents without the underlying technical capability (Gartner, 2025). At an industry-wide scale, this means that more than 95% of what is being sold as agentic AI is not the thing being purchased.
PwC's 2026 analysis of how AI is reshaping software valuations puts the point in commercial terms. The traditional metrics that defined SaaS quality, net revenue retention chief among them, can now mask seat contraction beneath expansion revenue from AI add-ons. An AI agent that can do the work of three analysts means the customer does not need three seats, which means seat-based pricing is under structural pressure across the entire enterprise software sector. Investors, PwC argues, must now disaggregate retention metrics by cohort, by product module, and by AI-impacted versus non-impacted revenue streams (PwC, 2026).
This is a fundamentally different DD job than the one being done by most providers. It requires the ability to assess whether the target's claimed AI capability is real or theatre, whether the data foundation supports the claimed capability, whether the unit economics survive contact with agentic AI displacement, and whether the revenue figures presented in the CIM have been generated by genuine commercial activity or by round-tripping schemes designed to look like genuine commercial activity.
What good actually looks like
The interventions that close the modern tech DD gap are well-understood by practitioners and almost entirely absent from the standard PE diligence playbook.
Capability verification before commercial diligence. The first question is no longer "does the technology scale" but "is the technology that is being claimed actually the technology that is running". This means code-level verification, examination of the actual model pipelines and data structures, and independent assessment of whether the AI capability described in the pitch deck corresponds to the AI capability deployed in production (Dealsuite, 2026).
Data foundation assessment. The Foundation Problem, covered previously in this column, applies with particular force to AI targets. A target whose AI products run on undocumented, ungoverned, low-quality data is selling a model that will fail to scale regardless of how impressive the demo appears. Independent assessment of data lineage, quality and governance against ISO/IEC 25012 and ISO 8000 standards is now a precondition for any AI-anchored valuation, not an optional extra.
Revenue model stress testing. If the target's revenue depends on seat-based pricing in a category where AI agents are eliminating the seats, the forward revenue model is not what the pitch deck claims. If the target's revenue comes from AI add-ons layered on a declining base business, gross revenue retention (GRR) is a more honest metric than net revenue retention (NRR), and investors need both, by cohort and by AI exposure (PwC, 2026).
Governance and audit verification. Builder.ai operated for ten months without a CFO. Its auditors had ties to the founder. Its CEO retained the title "Chief Wizard" after his removal. Each of these facts was discoverable through routine corporate governance review, and any one of them should have been disqualifying for a $250 million Series D in 2023. The fact that they were not tells you everything about how the AI narrative is overriding the diligence process.
Independent expert calls beyond management. McKinsey's 2026 research on Generative AI in M&A documents that AI-assisted diligence can compress timelines materially, with CIM extraction reduced from 10 to 40 hours to under one hour and Q of E completed 46% faster (McKinsey, 2026). That speed advantage is genuinely useful. But the same research notes that AI is a structural upgrade to how diligence operates, not a replacement for the human judgement that asks whether the technology actually does what is claimed. The Wall Street Journal had already done that work on Builder.ai in 2019. The institutional investors who funded the subsequent six years either did not commission the same work or did not read it.
Where Neurotic comes in
The proliferation of AI claims in pitch materials, the structural compression of valuations toward AI narratives, and the inadequacy of traditional DD methodologies against this environment, have combined to make technical due diligence the single most important risk-management discipline in the 2026 deal market.
Neurotic's technical due diligence service is built specifically for this work. We assess capability versus claim, data foundation versus narrative, revenue model versus disruption exposure, and governance versus founder personality cult. We are independent of any AI platform or technology vendor, which means the recommendations are driven by what is actually in the target rather than by what someone is trying to sell you. If you are evaluating a tech-led acquisition, a direct investment, or a co-invest opportunity, the right next step is an independent technical assessment that reflects what 2026 actually requires.
The £450 million lost in Builder.ai went somewhere. Some of it went to the founders, some of it went to operations, and a significant portion went to investors who exited at higher valuations than the company was worth. The next Builder.ai is already raising money. The diligence that catches it is the work that gets done before the wire transfer, not after the bankruptcy filing.
**Talk to us for an Audit today → [email protected]
References
Dealsuite (2026) Unlocking Real AI Value: A Due Diligence Guide for PE Investors. Available at: https://www.dealsuite.com/en/blogs/unlocking-real-ai-value-a-due-diligence-guide-for-pe-investors [Accessed 22 June 2026].
eDiscovery Today (2025) The Biggest AI Fraud in Startup History, 5 June 2025. Available at: https://ediscoverytoday.com/2025/06/05/the-biggest-ai-fraud-in-startup-history-artificial-intelligence-trends/ [Accessed 22 June 2026].
Gartner (2025) Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027, press release, 25 June 2025. Available at: https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027 [Accessed 22 June 2026].
McKinsey & Company (2026) Generative AI in M&A, January 2026. Cited in v7 Labs (2026), see below.
Mexico Business News (2026) SpaceX Valuation Tops Amazon Amid AI Infrastructure Push, 19 June 2026. Available at: https://mexicobusiness.news/aerospace/news/spacex-valuation-tops-amazon-amid-ai-infrastructure-push [Accessed 22 June 2026].
People Matters (2025) AI fraud: 700 Indian engineers did the work while Builder.ai claimed it was AI, 5 June 2025. Available at: https://www.peoplematters.in/news/funding-investment/ai-fraud-700-indian-engineers-did-the-work-while-builderai-claimed-it-was-ai-45865 [Accessed 22 June 2026].
Pragmatic Engineer (2025) Builder.ai did not "fake AI with 700 engineers", 12 June 2025. Available at: https://blog.pragmaticengineer.com/builder-ai-did-not-fake-ai/ [Accessed 22 June 2026].
Pro Pakistani (2025) Indian AI Startup Worth Billions Collapses Over Completely Fake "AI Backend", 29 May 2025. Available at: https://propakistani.pk/2025/05/29/indian-ai-startup-worth-billions-turns-out-to-be-biggest-scam-ever/ [Accessed 22 June 2026].
PwC (2026) How AI is reshaping software valuations in M&A. Available at: https://www.pwc.com/us/en/services/consulting/deals/library/ai-software-valuations-ma-private-equity.html [Accessed 22 June 2026].
Sacra (2026) SpaceX revenue, valuation & funding. Available at: https://sacra.com/c/spacex/ [Accessed 22 June 2026].
Seeking Alpha (2026) SpaceX: How AI Ruined A Perfect Business, 18 June 2026. Available at: https://seekingalpha.com/article/4916046-spacex-how-ai-ruined-a-perfect-business [Accessed 22 June 2026].
Sifted (2026) Builder.ai founder named 'key beneficiary' in money-laundering probe, 23 April 2026. Available at: https://sifted.eu/articles/sachin-duggal-money-laundering [Accessed 22 June 2026].
Tech Jacks Solutions (2026) AI Infrastructure News: SpaceX S-1 Shows Starlink Profits Funding xAI Losses at Scale. Available at: https://techjacksolutions.com/ai-brief/ai-infrastructure-news-spacex-s-1-shows-starlink-profits-fun/ [Accessed 22 June 2026].
v7 Labs (2026) Private Equity Due Diligence Process: AI vs Traditional. Available at: https://www.v7labs.com/blog/ai-vs-traditional-pe-due-diligence [Accessed 22 June 2026].
Via Satellite (2026) Assessing SpaceX Finances, Addressable Market, and the AI Pitch Ahead of IPO, 3 June 2026. Available at: https://www.satellitetoday.com/finance/2026/06/03/assessing-spacex-finances-addressable-market-and-the-ai-pitch-ahead-of-ipo/ [Accessed 22 June 2026].
