Tudovu Field Notes

Engineering

When Someone Else's Security Failure Becomes Your Investment Problem

One compromised vendor can put an entire portfolio at risk. Why investors need to look beyond the companies they fund and examine the technology those companies depend on.

By Tudovu · · 6 min read

img

On November 16, 2026, hackers broke into a website testing tool called Optimizely. Zoom uses it. So does HSBC, the BBC, CNN, Equifax, IKEA, Salesforce, GitLab, and 170 other organizations. Researchers at Report URI found malicious code hidden inside Optimizely's scripts on 178 companies' websites at once. The code swapped out cryptocurrency wallet addresses, skimmed payment card numbers, stole browser cookies and screenshots, and on iPhones and iPads it served known iOS exploit kits, though researchers have not shown that any device was successfully compromised. Attackers pulled in about $7,000 in stolen cryptocurrency within a short time. By November 17, Optimizely said it had patched the vulnerability and removed the injected code from affected projects, and the attacker's domains were being shut down. The company says the attacker exploited a flaw in its Experimentation product and planted the code in an optional field meant for a customer's own JavaScript. It still hasn't said how many customers were affected.

None of those 178 companies were hacked directly. Optimizely was, and everyone who used it got pulled into the blast radius anyway.

The same trick hit Rockstar Games earlier this year. Hackers never touched Rockstar's own systems. They broke into Anodot, a small analytics vendor Rockstar used to track data. One weak vendor opened a door into one of the biggest game studios in the world.

The pattern repeats because it works

Attackers don't need to beat a company's defenses anymore. They just need to find the weakest vendor in that company's supply chain and walk through the door it leaves open.

The numbers back this up. Comparitech counted 2,627 ransomware attacks in the third quarter of 2026. That's the highest quarterly total on record. By October, Breachsense found that 2026 had already passed all of 2025's ransomware victims combined. Two months were still left on the calendar.

Speed is changing too. In Japan, hackers took down a data center subsidiary tied to SoftBank in seven minutes. Four hundred ninety-five client companies were exposed before anyone could react. Seven minutes isn't enough time to call a security team, let alone fix anything.

Most of the exposed companies in that incident were small. Thin teams. Thin budgets. No one whose full-time job was security. That's not a coincidence. Secureframe's 2026 benchmark report found that 68% of companies have one or fewer full-time cybersecurity employees. For most startups, security is whatever gets done after the product ships. That often means nothing gets done at all.

Attackers know this. They don't need to break into a well-defended bank. They can break into a small vendor that bank already trusts and ride that trust straight through the front door.

What this means for venture due diligence

If you back startups, this changes what diligence should look like. A growth story doesn't mean much if a breach at a thirty-person vendor can freeze a company's biggest contract. It matters even less if that breach pulls your whole portfolio into someone else's blast radius. Growth speed matters less than whether a startup can survive being the easiest target in someone else's supply chain, or the weak link that drags a bigger partner down with it.

Secureframe's research shows what already happens to companies that skip this question. Sixty-one percent say compliance certification was required to win or renew a contract. Forty-seven percent say missing certifications delayed a sales cycle. Thirty-eight percent lost revenue or a competitive bid outright because they couldn't show the security proof a buyer wanted. These aren't forecasts. They're costs landing on companies right now.

Thirty-three percent of companies already feel pressure from investors or partners to show security maturity. That number should be higher. If an investor can't see how a company finds and fixes security gaps, that investor is betting blind on part of the business.

The cost that never shows up on a balance sheet

An unplanned security hire costs money. So does a rushed consulting contract or a remediation sprint. Each one pulls time and money away from the roadmap, and that cost rarely gets its own line item. Instead, it shows up as a delayed launch. Or a distracted engineering team. Or a funding round that drags on because diligence turned up more questions than expected.

Here's one way to picture the size of it. Pull two engineers off the roadmap for six weeks to fix a security finding. That's 480 hours of lost product work. The median U.S. software developer salary is $135,980 a year. At that rate, 480 hours costs about $31,380 in base salary alone, before benefits, equity, or whatever feature didn't ship on time. This illustrative number shows the scale of the problem. It doesn't come from one company's real books, though the underlying math repeats across nearly every portfolio.

Now flip it around. A $250,000 enterprise deal stalls for 90 days. The reason: a buyer's security team sent back a questionnaire, and the startup wasn't ready to answer it. If that company burns $150,000 a month, it spent $450,000 keeping the lights on while a $250,000 deal sat in limbo. Again, this is a hypothetical, built to show the shape of the problem. Any operating partner who has watched a deal stall over a security questionnaire will recognize it right away.

Why the backlog is the real problem

Most startups caught off guard by a breach had a different problem: they treated security like a side project instead of a normal part of engineering work. A scanner flags a misconfigured bucket or an outdated dependency. Someone writes it down. Then it sits in a backlog. The person who found the problem is rarely the person with time to fix it.

That gap between finding a problem and fixing it is where the real cost lives. Finding an issue is only the first step. Someone still has to write the fix, test it, and prove it happened.

A question worth sitting with

Optimizely patched the vulnerability and pulled the malicious code within hours. Report URI's researchers could trace the stolen crypto, but they still don't know how many credit cards or cookies were taken before the cleanup. Some of the affected companies say they found no evidence that customer data was stolen. Others are still checking.

Multiply that uncertainty across 178 companies. Most of them probably didn't know Optimizely was a single point of failure for their customers' data until this week. Every company in a venture portfolio depends on vendors it never vetted closely, and audits it never ran.

The Rockstar breach proved that a thirty-person analytics vendor can compromise a studio worth billions. The Optimizely breach proved that one shared tool can turn 178 unrelated companies into a single attack surface overnight. The next version of this story is already running somewhere, inside a vendor nobody in your portfolio has thought to ask about yet. The question is whether anyone finds out before the attackers do.


Source: Eduard Kovacs, "Optimizely Vulnerability Exploited to Serve Malware on Major Websites," SecurityWeek, November 17, 2026. https://www.securityweek.com/optimizely-vulnerability-exploited-to-serve-malware-on-major-websites/