TLDR
- Outages like AWS’s demonstrate that enterprise resilience must advance in lockstep with AI adoption.
- BDO data shows that most executives are accelerating AI plans, yet nearly half admit their infrastructure isn’t ready – widening the gap between ambition and operational reality.
- Culture, incentives and communication barriers still slow investment in resilience, leaving companies exposed until a breach or outage forces action.
Ric Opal was in Hamburg, Germany in late October when AWS went down, disrupting a wide swath of the internet and knocking thousands of websites offline. By some estimates, the outage cost $75 million an hour. Stretch that across a full workday and the tally climbs past half a billion dollars.
“I thought to myself, ‘it’s only going to get worse’,” said BDO’s global digital leader, in an interview with The AI Innovator. As companies embrace AI throughout their operations − including AI agents managing other AI agents − “our reliance on technology, technology infrastructure and the security of that infrastructure is actually growing,” Opal said.
Across the C-suite, companies are increasingly tying their performance goals to AI deployment. “I have talked to leaders whose goal sheet says X percent of work will be performed by artificial intelligence,” Opal said. “If you want to drive revenue and profitability … then you have to hold people accountable to a metric that has teeth.”
But the AWS outage underscores a broader reality: Enterprise AI ambitions are only as durable as the systems underneath. As organizations race to deploy AI agents, teammates and automated decision-makers, resilience – the ability to withstand disruption even as reliance on automated systems rises – becomes the real differentiator.
New BDO data points to the gap between AI aspirations and readiness: 57% of executives are fast-tracking AI adoption, but 42% acknowledge their underlying infrastructure isn’t ready. As companies automate more tasks and dependence on cloud platforms deepen, the attack surface expands.
No company, however, can engineer total protection. “You can’t protect against everything. It’s just fundamentally impossible,” Opal said. The real task is deciding “where are my best bets for resilience? Where are my best bets for risk?” − and aligning those decisions to business goals.
How AI strengthens resilience
Opal said AI can directly bolster resilience by automating the tedious work of documenting and assessing the current state of a company’s systems – a foundational step many companies struggle to take.
“You have to understand where you are … to understand how you get to where you want to go,” he said. Established frameworks from NIST and others can guide the process. “Can we use AI to automate the collection of baseline information relative to a framework? The answer … is absolutely yes.” But he cautioned that companies must also ensure the data collected is free from hallucination or bias.
Opal stressed the importance of bottom-up adoption. Workers closest to the day-to-day processes know where the relevant data resides. Once leaders have an accurate baseline, they can target investments toward systems that will blunt the impact of future cloud outages. The goal is to “weather that storm in a bit more tolerable way with less financial impact,” he said.
Why companies do not prioritize resilience
For many organizations, the barrier isn’t ignorance − it’s incentives. “It’s easy to know what you should do. (But) it may be hard to fundamentally do that thing,” Opal said. Companies reward profit and revenue growth. Risk mitigation typically lags behind.
That leaves many enterprises discounting the urgency of resilience until a breach or outage forces action. “Those who are breached will invest in security,” he said. “Those who are not breached think it won’t happen to them.”
The rapid growth of AI-enabled cyber attacks heightens the stakes. “Outages will increase because of the use of AI in the realm of cyber by nation-state or others, and it is frankly going to call into question (the importance of) critical infrastructure in a material way,” Opal said. “You cannot deny it.”
Internal mistakes can be just as costly. Even mature AI-heavy firms can falter when a human mistake slips through. “It’s never going to be perfect,” Opal said. Still, AI-driven tooling is improving fast. “If you look at where we were 12 to 18 months ago, it’s improved radically.”
Culture, structure and communication gaps
Culture, corporate politics and organizational structure often present hurdles to stronger risk management, Opal said. One reason is the communication gap. Risk teams and business leaders “speak two different languages. One speaks the language of return, the other speaks in binary,” he said. Companies need people who can “rationalize the conversation” so that cybersecurity teams can link risks directly to business outcomes.
He recommended cross-functional revenue and risk “councils” that enable teams to meet, talk and understand each other’s priorities. “You’ve got to have a bit of cross-functional teaming” to ask questions, compare notes and “decipher what each other is saying,” Opal said.
Benchmarking is also essential. “I would want to benchmark myself, and then against my peer group,” he said. He would also talk to his peers and even rivals, noting that these conversations can help reveal blind spots internal teams often miss.
Exciting yet scary, too
Looking ahead, Opal said he is both hopeful and anxious about AI’s impact on society. “I am some days afraid for my children – what will their jobs be and how will they make a living? Some days, I’m excited for what the possibilities are and the good that can come.”
“You cannot deny the power of artificial intelligence,” he continued. “Once we couple that with quantum, I’m hopeful that cures to diseases and many other (beneficial) things are going to happen. And that’s fantastic. I’m also confident that bad things are going to happen.”
But he’s clear-eyed about the urgency. Companies still early in their AI planning must act soon. “If you’re not already moving down this path, or have a plan, or an excellent strategy, the writing is, fundamentally, … on the wall,” Opal said.
This future is “exciting and it’s scary,” Opal concluded. “I hope that we, as the human collective, find a good path forward to use it for all the right things as much as we can.”
