IBM Among AI, Digital Sovereignty, and Quantum: The Route for the Coming Years

Three missions and two threads that connect them. This is how Nico Losito, General Manager of IBM Italy since April 1st, summarized the company’s strategy in front of the press in Milan. The first two, artificial intelligence and hybrid cloud, have accompanied IBM's work for at least four or five years. The third, quantum computing, has just emerged from research labs to face its first commercial applications. Two conditions remain indispensable for Losito: security and skills.

Losito starts with the numbers. IBM depicts the adoption of AI with research conducted on 6,000 CEOs worldwide, revealing an ambivalent picture. Sixty-nine percent state that AI has already entered core processes, but only ten percent use it to generate new revenue, while most initiatives still focus on productivity and savings. The wave is large and fast, yet adoption in businesses is proceeding more slowly than expected, especially in small and medium-sized Italian enterprises, where a certain skepticism persists. Regarding the risk of a bubble, however, the General Manager has no doubts: "Today, the promise of AI is being realized. I find it hard to believe that everything could deflate; there are solid economic fundamentals and tangible returns on investment." Rather than a bubble, he adds, what we are witnessing is financial volatility while technology continues to enter companies.

The Client Zero Method and IBM Bob

The first proving ground for IBM is IBM itself. With the IBM Client Zero program, the company has utilized its watsonx platform to redesign internal processes, starting every time from four questions: does this process generate revenue, profit, customer satisfaction, or cash? When the answer is always no, the process does not benefit from an AI upgrade and is eliminated. "Our first customer is ourselves," summarized Losito. Each year, this mechanism collects tens of thousands of proposals from teams in various countries, selects them, and moves them into production; 165,000 proposals were processed in the last year alone, resulting in savings reaching $4.5 billion over three years, about 7% of global revenue.

Following the same principle, IBM Bob was created, which entered beta in November and is now purchasable. Bob is not the usual code-writing assistant; it works across the entire software development lifecycle, from design to monitoring to governance and compliance. Bob generates specialized agents for testing, documentation, or refactoring, and internally accommodates different models, including those from Anthropic. It also monitors resource consumption and costs, a thorny issue when the intensive use of models inflates bills. Internally adopted by 80,000 IBM developers, it has reduced development times by 70% and improved quality and testing by 85%.

Digital Sovereignty as Freedom

AI and infrastructure travel together, and many initiatives run aground right upstream amid fragmented data, isolated systems, and clouds multiplying without a common operating model. IBM's answer is an open hybrid cloud architecture. Red Hat OpenShift ties the two worlds together, making models interoperable, from IBM's Granite to those from Google, Meta, and Anthropic, and computing resources, from IBM Z mainframes to LinuxONE systems and the clouds of other operators. Recently, the z17 and LinuxONE 5 ranges have expanded with rack mount configurations designed for the space and cost constraints of data centers, featuring already integrated post-quantum encryption.

The theme that has become a priority in recent months in conversations with clients is digital sovereignty. For IBM, it does not just pertain to where data resides, but is a matter of freedom on four fronts: data, operations, technology, and AI. At the center is IBM Sovereign Core, the software platform presented on May 4, which the company contrasts with the approach of hyperscalers. "They add a level of sovereignty, but it's like staying in a five-star hotel where the hotelier always has the keys," explained Losito. In contrast, Sovereign Core is a product that the client installs on their infrastructure, currently x86 and in the future also Z and Power. The control plane, identities, keys, and audit proofs remain in their hands, while compliance transforms into a continuous process with automatic checks based on NIST and CIS standards, with about 150 regulatory frameworks already loaded. A catalog of services finally allows for the management of separate tenants, each with its own rules depending on jurisdiction.

In the demonstration that followed the meeting, an IBM technician showed how an administrator can create tenants, dynamically distribute resources, and check in real-time if data ends up where it shouldn't.

Quantum, Security, and Skills

On quantum computing, IBM claims a roadmap that it has adhered to for years. It brought the first quantum computer to the cloud in 2016 and has since deployed over 90 systems, more than 30 of which exceed 100 qubits. The latest, IBM Nighthawk with 120 qubits, performs circuits about 30% more complex than the previous version. The stated goal is to demonstrate quantum advantage by 2026, that is, a measurable advantage over traditional computers, and by 2029 to achieve fault-tolerant systems capable of correcting errors, IBM Starling. These two dates are significant and indicate a shift from past communication focused solely on the number of qubits. Setting the year for quantum advantage and indicating 2029 for full error tolerance, in a field accustomed to shifting horizons, is a commitment far from guaranteed. Supporting this journey is an investment of over $10 billion over five years. Around the hardware revolves an ecosystem of over 340 companies, universities, and startups, along with a community of more than 600,000 users, with the open-source framework Qiskit used by 70% of developers.

In this context, Losito placed much emphasis on the value of creating an ecosystem even in Italy, where an IBM quantum machine still does not exist, despite some initiatives being started, such as the competition from the Campania region for an installation in Salerno, with local facilities already ready to host it. His point, rather than a complaint, is a hope for a more determined systemic approach capable of supporting the implementation of the national strategy on quantum computing and positioning Italy to seize its competitive advantages.

Quantum also brings with it a security problem that Losito considers underestimated. The reasoning is that of "I collect the data today, believing that in two years someone will produce the key to open that vault." Whoever holds a secret intended to last twenty or fifty years must already move towards post-quantum algorithms because migration typically requires five to ten years on average. And this is a field in which IBM claims a direct contribution, with three of the four algorithms standardized by NIST originating from its research, two of which were developed at the Zurich center.

Security is also the countermeasure to another acceleration, that with which AI multiplies the discovery of vulnerabilities, as has been made evident by Anthropic's Mythos model. IBM has participated from the beginning in the Glasswing project and has launched Project Lightwell with Red Hat, a $5 billion effort to secure open-source software. Additionally, IBM Concert has been implemented, an agent-based platform that does not replace existing tools but coordinates them, relates signals, and assigns priorities. The cited numbers reflect this direction. Deutsche Telekom has reduced response times to critical vulnerabilities from 80 to 8 hours, while SIXT has cut detection and resolution times by 67%.

The final chapter is that of skills, which Losito places at the bottom but not in terms of importance. Traditional organizations, with their rigid roles, risk becoming a brake; he suggests a reorganization around mandates rather than roles, investing in AI literacy at every level and creating what IBM calls a new competency unit, composed of digital agents alongside typically human capabilities like judgment, emergency management, and problem-solving. This is where the company's positioning takes shape: not a supplier promising sovereignty and security by contract, but an actor aiming to deliver the keys to the customer, in terms of infrastructure as well as models, leaving AI with the task of empowering people rather than replacing them.