The launch of Amazon Web Services (AWS) in 2006 accelerated not only public cloud adoption but also the awareness it created accelerated the adoption of new services, with Infrastructure As a Service (IaaS), Platform as a Service (PaaS), Software as a Service (SaaS) and more recently Function as a Service (FaaS) as well as machine learning operations offerings. The low-cost, highly flexible model offered by the public cloud was too tempting to resist and very quickly organisations of all sizes were moving a variety of workloads to the cloud. However, “cloud regret” soon set in, with a variety of challenges arising that hadn’t been foreseen when opting to migrate, causing headaches across organisations.
As the famous saying goes: “Those who cannot learn from history are doomed to repeat it” – George Santayana. Nearly two decades later, artificial intelligence (AI) is going to pose similar challenges for organisations looking to rush to benefit from increased efficiency without considering the long-term implications of vendor-lock-in, such as cost management, as cloud services costs can be difficult to understand, control and account for, as well as data sovereignty and lack of safety nets, especially with the fast-moving pace of recent AI developments.
With AI experiencing unprecedented growth, and subsequent adoption by organisations, it’s no wonder that it is on track to become a $1.3 trillion market (equivalent to £1 trillion) in less than a decade. With implementations set to continue increasing, learning from the top 3 challenges from the public cloud adoption could prevent later AI regret.
Challenge 1: Putting all of Your AI Eggs in One Basket
Many technology leaders will remember the pain of being tied to expensive contacts during the initial rush to benefit from the cloud, with an increasing number of businesses reliant on a single cloud provider, leaving them vulnerable to update issues, outages and reliability problems. 80% of cloud-migrated organisations faced vendor lock-in issues, according to Gartner.
Solving problems and delivering value will require creating services which leverage different AI technologies, constraining your solutions will impact your competitive advantage. Cloud lock-in made it difficult for organisations to switch to another cloud provider, or revert back to on-premises solutions. In the same way, what is considered best in class for AI today may not be tomorrow. Having the flexibility to shift between AI vendors is crucial as the market is moving at a rapid pace. Organisations that were hoping to leverage the earliest AI technology might find themselves tied to models that are now obsolete, leaving them in the dust of companies with long-term AI implementation strategies.
It isn’t just technology that changes, your needs do too, and being married to just one vendor is likely to hold you back, and, like the cloud, see you paying more.
Challenge 2: Where’s Your Data Gone?
When businesses first moved to the public cloud, there were few regulations governing data storage and privacy. However, regulations always eventually catch up with innovation and within data storage we saw a number of frameworks, including the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA) emerged; as a result, companies had to react quickly to stricter compliance requirements.
Similarly, as AI adoption continues to increase, regulations will likely follow suit, particularly around data processing. Much like the public cloud transition, businesses looking to successfully implement AI will need to account for region-specific compliance standards and take necessary steps to protect their data.
Challenge 3: Running Before You Walk
The hype around new technologies can make organisations eager to adopt them as fast as possible, and AI is no different. The excitement surrounding AI, much like the early rush to the public cloud, could push organisations to adopt it hastily, often without proper groundwork. AI adoption should follow a similar trajectory. 25% of tech leaders have reported investing in AI too quickly, which has led to several teething problems becoming viral in recent months.
Organisations looking to invest in the benefits of AI should look to integrate AI-based applications piece by piece, from well-versed established providers. While AI promises efficiency and transformative potential, it’s crucial to start small, applying AI to less critical functions before scaling up. This iterative approach allows for troubleshooting and learning without putting core business operations at risk.
Learning from the Cloud
As organisations navigate the rapidly advancing AI landscape, the lessons learned from the early days of public cloud adoption are invaluable. The challenges of vendor lock-in, data sovereignty, and rushed implementation are just as relevant today as they were almost two decades ago. To avoid repeating past mistakes, technology leaders must approach AI implementation from a strategic perspective: starting small, ensuring flexibility in vendor relationships, and staying ahead of evolving regulations. By doing so, businesses can leverage AI’s transformative potential while safeguarding against costly missteps and ensuring sustainable, long-term success.
