In October 2011, Hewlett-Packard paid $11.1 billion for Autonomy, a UK enterprise software firm, betting it could transform HP from a hardware company into a software one. Eighteen months later, HP wrote down $8.8 billion of that value and accused Autonomy's founder of accounting fraud.
 |
| Photo by Brett Sayles from Pexels |
The deal became one of the costliest acquisitions in tech history — not because the technology was bad, but because HP never independently verified what it was buying, and had no integration plan waiting at close.
The first five parts of this series made the case for buying over building. This part makes the other case: acquisitions fail constantly, and they tend to fail in the same handful of predictable ways. If speed is the reason companies buy, the failure patterns below are the price of that speed when nobody checks it.
The Hall of Fame of Failed Deals
A few deals show up in almost every post-mortem of tech M&A, and each one teaches a different lesson:
- HP – Autonomy ($11.1B, 2011): HP skipped independent verification of Autonomy's revenue mix and paid the price when accounting irregularities surfaced 18 months later
- Microsoft – Nokia ($7.2B, 2013): Microsoft bought a handset maker to fix a strategic anxiety about mobile share, not because customers wanted the combination; it wrote off $7.6 billion and cut 18,000 jobs within two years
- AOL – Time Warner ($165B, 2000): The largest merger of its era ran on a story about media-internet convergence with no operational plan for how two incompatible cultures would actually run one company
- eBay – Skype ($2.6B, 2005): eBay paid a premium on the assumption that auction users wanted voice calls; they didn't, and eBay sold Skype at a loss a few years later
- Adobe – Figma ($20B, 2022): Not a post-close failure but a pre-close one — EU and UK regulators signaled they'd block it on competition grounds, and Adobe walked away, paying a $1 billion termination fee
Between them, these five deals alone destroyed or forfeited well over $30 billion in value. The pattern repeats because the underlying mistakes are rarely technical. They're structural.
Why Deals Fail: The Four Repeat Offenders
Strip away the specifics, and most failed acquisitions collapse into four causes:
- Overpayment without independent verification — HP took Autonomy's numbers at face value; Bank of America did the same with Countrywide's mortgage book
- Strategic mismatch dressed up as vision — Microsoft needed a mobile story, not necessarily Nokia specifically; Google's purchase of Motorola followed a similar logic
- Cultural collision with no integration plan — AOL and Time Warner never resolved who was actually in charge of what, or how; HP and Autonomy had the same problem across the Atlantic
- Regulatory exposure priced in too late — Adobe's Figma deal was economically sound and legally fatal; antitrust review has only gotten more aggressive since
Notice what's missing from this list: bad technology. Almost none of these companies failed because the thing they bought didn't work. They failed because nobody stress-tested the deal thesis before the wire transfer went out.
Why Some Acquisitions Actually Work
The deals that hold up share a different set of habits: a narrow, specific thesis rather than a story about "convergence"; diligence built to find reasons to walk away rather than confirm a decision already made; a retention plan for key people locked in before close, not after; and room in the integration plan to run the acquired unit semi-independently, at least at first.
Facebook's 2014 purchase of Oculus is the usual counterexample. Zuckerberg didn't need Oculus to patch an urgent hole in Facebook's business the way Microsoft needed a mobile story, and he let Oculus keep its own culture and roadmap for years rather than force an immediate merge. The current wave of AI acqui-hires runs on a version of the same instinct: the point is rarely the product on day one, it's whether the team stays and keeps building. Anthropic's Vercept deal shows both sides of this at once — most of the founding team joined, and the technology got folded into Claude's computer-use work, but not every co-founder made the move, and the transition wasn't friction-free in public.
Failed vs Successful: The Pattern
| Factor | Deals That Fail | Deals That Hold Up |
|---|---|---|
| Deal thesis | Vague ("convergence," "the future of X") | Specific operational outcome |
| Diligence | Confirms a decision already made | Actively looks for reasons to walk |
| Talent plan | Assumed, addressed after close | Negotiated and locked in before close |
| Integration pace | Immediate, full merge | Staged, semi-independent at first |
| Regulatory read | Priced in late or ignored | Modeled into deal terms from the start |
None of this is exotic. It's the same discipline good product management demands — a clear hypothesis, a plan to disprove it, and a rollout that doesn't bet everything on day one. Most acquiring companies know this. Most skip it anyway, because a live bidding process rewards speed and conviction, not caution.
What This Means for the Current AI Acquisition Wave
Every deal covered earlier in this series — Cursor, Vercept, Forethought, and the rest — is still too new to know which bucket it lands in. But three failure patterns are worth watching as the AI acquisition wave keeps accelerating: talent-driven deals live or die on retention, not technology, since the entire thesis depends on whether an acquired research team stays past year one; valuations are running ahead of verified revenue, priced against growth curves rather than trailing financials, which is the exact setup that preceded HP-Autonomy; and antitrust scrutiny of AI consolidation is only getting sharper, as Adobe-Figma showed regulators are willing to block deals late, after both sides have already committed.
The Takeaway
Buying is faster than building. It is not automatically safer. Every acquisition covered in this series looked smart at signing — whether it stays smart depends entirely on the unglamorous work that happens after the press release, the part almost none of these companies want to talk about. Speed wins the deal. Discipline decides whether it was worth doing.
Previously in the series: NVIDIA and the Race for AI Infrastructure. Next: Why Startups Choose Acquisition Over IPO.
Erwin Castro
Founder & Editor • The CODEW
Erwin Castro is the founder and editor of The CODEW, covering technology mergers and acquisitions, startup exits, artificial intelligence, enterprise software, and Build vs Buy strategy. With more than a decade of journalism experience, he has contributed to Sportskeeda, IBTimes, University Herald, US Blasting News, and Seeking Alpha. His work focuses on explaining the business strategy behind technology deals and their impact on the global technology industry.
Reviewed by Erwin Castro
on
Monday, July 13, 2026
Rating:
No comments: