Trending Articles
Trending Topics
Trending
In a world of ever-evolving technologies, staying ahead of the threats of tomorrow takes more than good instincts: it takes strategy, speed and experience. Fortunately, Andre Reitenbach has all three. Since co-founding Gcore over a decade ago, he’s spent his time wisely. Building a company that isn’t just designed for performance, but for protection.
As Andre notes, “cyber threats are constantly evolving, in part led by bad actors exploiting AI’s capabilities to amplify their attacks”. One of his key concerns is in phishing attempts. Once easy to spot, they are now “virtually indistinguishable from personal correspondence,” thanks to generative AI.
Amongst the most aggressive threats at present are DDoS attacks, as pointed out by our security expert Davey Winder yesterday. Andre says such attacks are “increasing in volume year-on-year” and now becoming more powerful, “wielding up to 1.7 terabits per second to flood servers with malicious traffic”.
In the face of this, Andrew emphasises that the IT industry will need to invest in a combination of “rigorous employee training and advanced security tools such as Web Application and API Protection,” to properly combat these AI-driven threats.
Fortunately, AI has plus sides too. “Virtually every industry will be impacted by AI, albeit with varying speeds of adoption,” said Andre, pointing to examples such as entertainment, product development, prototyping and communications.
Innovation in technology will always come with new challenges, especially in the world of AI, and anticipating these innovations is the only way to stay ahead. That’s why, to start, we asked Andre what emerging technologies we can expect in the short term.
AI will have the widest-ranging impact of any emerging technology. Virtually every industry will be impacted by AI, albeit with varying speeds of adoption. One area that AI has already begun to revolutionise is the entertainment industry: we’ve seen streaming services use AI to enable enhanced content delivery, including personalised recommendations, real-time subtitles and dynamic content moderation. What’s more, as generative AI capabilities continue to advance, we may be able to experience something approximating the Star Trek holodeck in the metaverse. Businesses will also apply this technology to areas outside of entertainment, such as product development, prototyping and communications.
Further developments will include physical AI, like robots that can perform complex functions in the physical world. Use cases for physical AI will range from operating in warehouses to highly intricate tasks such as surgery. AI agents are another emerging trend – these are a form of AI that can make decisions autonomously when performing tasks on behalf of a user. As a result, the difference between briefing a human employee and briefing an AI agent will be practically indistinguishable. The autonomous reasoning, problem-solving and decision-making capabilities of AI agents have the potential to transform the way we work across all industries.
The evolution in cloud computing will in part be driven by a changing regulatory landscape. As companies began training AI models in recent years, regulators introduced a series of policies that set ethical standards for the use of data in AI training. The purpose of these regulations is to compel businesses to control where data is processed for security. For instance, businesses cannot play it ‘fast and loose’ with training AI on healthcare information; they must have direct control over where data is processed with clear geographical boundaries. Traditional cloud computing cannot offer this level of control, which will increasingly lead businesses to adopt a sovereign cloud approach based on either dedicated private data centres or edge computing.
Private clouds are computing environments dedicated to a single organisation. This approach ensures compliance by localising the cloud environment within a specific jurisdiction. However, few businesses can afford to invest in building and maintaining their own data centres. Edge computing provides an alternative that brings processing power and data storage closer to the devices or locations where they’re needed. Unlike private clouds, edge computing is almost always provided by a third party. This is an affordable option for businesses to stay compliant by dynamically routing requests of specific jurisdictions to the nearest edge location. All of this is taken care of behind the scenes by the provider, making this the simplest and most affordable approach for data compliance moving forward.
Cyber threats are constantly evolving, in part led by bad actors exploiting AI’s capabilities to amplify their attacks. Cybersecurity must therefore adapt to keep pace with escalating threats. For example, AI can now be used for advanced threat detection and automated monitoring. This takes the pressure off security teams by outsourcing repetitive tasks. AI-driven security is capable of interrogating vast amounts of data in real time, far beyond the scope of human analysts. This limits the possibilities for attackers to gain access and reduces the likelihood of a security breach. AI can also speed up incident responses with predefined actions on detecting a threat. Time is a critical factor in the event of a cyber-attack, and automated responses such as isolating affected systems can be a lifeline that contains the threat and lowers downtime costs.
AI-driven cyber threats are not necessarily new in their strategy, but their scope and sophistication have dramatically increased. Phishing, for example, is an old strategy used by cybercriminals to exploit human error. In the past, human-generated attempts at phishing had certain tells that could be detected by a reasonably vigilant user, such as typos or unnatural wording. Now, however, generative AI models can produce flawless phishing emails that are virtually indistinguishable from personal correspondence. Even video content can be manipulated to trick users through deepfake technology. In the face of these alarming innovations on the part of cybercriminals, the IT industry will need to invest in a combination of rigorous employee training and advanced security tools such as Web Application and API Protection (WAAP) to identify and neutralise AI-driven threats in real-time.
Intellectual property is another concern that needs to be addressed head-on. AI models are being trained on vast stores of data, sometimes indiscriminately and in violation of copyright law. Notable artists such as Paul McCartney are speaking out against copyright infringements driven by AI. Regulations such as the EU AI Act are coming into force to protect intellectual property, which means that businesses must ensure that their models are being trained only on data that belongs to the public domain.
AI models have justifiably received a lot of attention in recent years, but the methods used for training them have been comparatively underexplored. AI training requires immense resources, computational power and expertise. These demands have created a barrier to entry that restricts AI training to a small handful of tech giants. However, smaller businesses can work around the limitations of their resources by leveraging inference technology. Inference allows businesses to adapt pretrained AI models rather than starting from first principles. This will have a democratising effect on AI training as businesses will not need to bear the burden of creating entire models from the ground up.
Inference will be increasingly central to AI applications in the years to come. One of the trends driving this is the growing demand for immediate insights in sectors such as healthcare, finance and retail. More and more businesses have come to rely on AI systems that can analyse data and generate responses in fractions of a second, such as customer service chatbots. Since inference models have already undergone intensive training, they can quickly adapt to new inputs.
The development of pretrained models will depend on a combination of hardware and infrastructure. Advanced GPUs will provide the necessary processing power and speed to enable real-time AI applications. This can work in tandem with edge computing, which brings processing data closer to its source to minimise latency.
Another overlooked opportunity is the growing potential to make AI cheaper and more accessible. Algorithms used for model training are constantly improving. As the algorithms become increasingly efficient, they will be less demanding in cost and energy usage. More businesses will benefit from AI as it becomes less resource-intensive and more affordable. This process will be further amplified by the growth of open-source AI, as we’ve seen with the DeepSeek phenomenon. As these trends develop, AI will be democratised as a commodity or mass product, rather than something exclusive to tech giants.
Our Everywhere Inference solution will play a transformative role in AI application development. Everywhere Inference can be deployed effortlessly on the cloud, on-prem, or a combination of both, making it an accessible tool for startups and enterprises alike. Everywhere Inference was designed with a forward-looking view toward changes in data compliance and data security. The system offers smart routing capabilities that enable users to direct workloads to their preferred region, helping them stay compliant with local regulations. The solution also allows businesses to securely isolate sensitive information on-premise, such as a hospital building.
Aside from its flexibility, Everywhere Inference also boasts the support of Gcore’s network, which covers over 180 points of presence worldwide. This ensures real-time processing and seamless performance across geographical boundaries. With a combination of high performance and flexibility, Everywhere Inference eliminates the need for customers to spend additional time managing costs, skills, or infrastructure for global AI deployment. We believe the edge is key to delivering top performance and exceptional user experiences.
Cybercrime is undoubtedly one of the biggest challenges for IT professionals in the next decade. Distributed denial-of-service (DDoS) attacks in particular have been increasing in volume year-on-year, with a total of 445,000 incidents recorded in first half of 2024 alone. These attacks are also becoming more powerful, wielding up to 1.7 terabits per second to flood servers with malicious traffic. To put this in perspective, one terabyte per second equates to more than 212,000 high-definition video streams in terms of data flow. Although most DDoS attacks (over 85%) last less than ten minutes, their sheer frequency and intensity can still lead to major operational disruptions. Notably, the longest attack recorded in the first half of 2024 spanned 16 hours, highlighting the critical importance of resilient and adaptive mitigation strategies.
Over the next decade, AI inference will shift towards more efficient, scalable and cost-effective infrastructure. Specialised accelerators and domain-specific chips will reduce reliance on general-purpose GPUs, while edge AI adoption will drive lower-latency processing. At Gcore, we see emerging architectures – such as photonic computing and quantum-inspired AI models -reshaping how enterprises deploy AI at scale, optimising both performance and energy efficiency across global workloads.
