At Intel Vision 2025, newly appointed CEO Lip-Bu Tan outlined a bold reset for the company, emphasising a return to engineering excellence and a customer-first mindset. In his keynote, Tan revealed how Intel will navigate the AI era with a software-led strategy, custom silicon, and a revitalised foundry model.
Fourteen days is not long. It is barely enough time to find the coffee machine, let alone set the strategic direction of one of the world’s most iconic technology companies. However, for Lip-Bu Tan, Intel’s new CEO, two weeks has been enough to understand that the company is standing at a crossroads. With trust to rebuild, talent to regain, and a new era of computing taking shape, Tan’s early observations point to a fundamental reshaping of what Intel is and how it operates.
Engineering-first or not at all
Tan’s opening priority is unambiguous: restore Intel as an engineering-first company. That statement is not a soundbite. It is a line in the sand. After years of organisational drift, missed deadlines and product underperformance, Tan is placing engineers and architects back at the centre of the company’s identity. He believes innovation has become trapped under layers of bureaucracy and that culture must be dismantled.
“We will move fast, break through bureaucracy and empower small, focused teams to drive innovation,” Tan says. “Bureaucracy kills creativity. I have seen small teams move quickly and take on the biggest incumbents. We must replicate that energy and urgency at Intel.” Tan’s approach recalls a start-up mentality, flat hierarchies, empowered individuals, and a direct line between engineering brilliance and market impact.
This also means reversing the loss of key talent. Years of attrition have weakened Intel’s bench strength. Tan’s response is to create environments where engineers want to stay, thrive, and build. The focus is not only on technical excellence but also on rediscovering a culture of urgency, clarity, and accountability.
Architectures for the AI era
AI is no longer a speculative investment or a siloed capability. It is the defining architecture of modern computing, and that reality is reshaping Intel’s priorities. “Artificial intelligence is at the heart of this transformation,” Tan adds. AI is not new, but the scale of data available today is enabling entirely new approaches to learning, forecasting, and innovation.”
Traditional compute models are being outpaced by the demands of generative AI, agentic systems, and high-throughput inference. From content creation to robotics, new applications demand rethought architectures, novel materials, and new ways of matching software to silicon. Tan is not shying away from the complexity. Instead, he places AI at the centre of Intel’s future relevance.
“AI is forcing a total architectural shift in computing,” Tan continues. “We are moving beyond traditional models toward hyper-scaled, data-intensive architectures. This requires massive computational power and new infrastructure.” That infrastructure includes investments in silicon carbide, indium phosphide, and photonics and a longer-term view towards quantum computing. These shifts demand more than new chips; they require a new relationship between hardware and software.
Software first, not an afterthought
That new relationship is being codified into a shift in design philosophy. For decades, Intel’s model was to engineer the hardware first, with software built afterwards to match it. But in an AI-native world, that model no longer serves. The company is flipping its development process. Customer problems now define the starting point. “We must flip the process,” Tan confirms. “We need to begin with the problem the customer is trying to solve, define the required workloads and design from that starting point.”
Tan describes this as a “software-first mindset,” but it is better understood as systems-level thinking. Hardware and software will no longer be built sequentially but co-designed as integrated systems. That shift is not cosmetic. It shortens development cycles, improves power efficiency, and enables higher performance for compute-intensive and agent-based workloads. It also opens the door to customised silicon designed in collaboration with customers.
“Over the weekend, I met with three major data and AI platform leaders eager to collaborate on defining the next generation of x86 solutions,” Tan explains. “These conversations are already leading to deep dives with our engineering team to begin reimagining our portfolio.” It is a quiet pivot to a service mindset, where Intel becomes a supplier and a partner in workload-specific innovation.
The foundry as strategic muscle
Foundry is not a sideline. It is core to Tan’s strategic plan. As global chip demand accelerates, geopolitical concerns and the need for resilient supply chains have propelled foundry capabilities to the centre of national and commercial strategy. Intel’s advantage, Tan argues, is that it is the only American company designing and manufacturing advanced chips.
“Foundry is a service business built on trust,” Tan says. “Each customer has unique design methodologies, preferred IP vendors, and EDA tools. We will not force changes on them. We will adapt and optimise based on their preferences and requirements.” The focus on customer adaptability is both practical and reputational. Foundry success is measured in yield, consistency, and the ability to meet diverse customer needs without friction.
Tan’s experience at Cadence, where he built deep familiarity with design tools and customer collaboration, is evident. Weekly meetings with engineering leaders are underway to track process technology and foundry execution. The milestone ahead is Intel 18A, with Panther Lake planned for launch later this year. The ambition is clear: high-volume production, improved yield, and a service model built around external tape-outs and anchor customers.
The promise and the pressure of performance
For all the strategic clarity, Intel still faces operational urgency. Product gaps in the data center, a need to regain competitive edge in client computing, and power efficiency challenges in AI workloads are immediate hurdles. Tan is placing clear stakes in the ground. “Performance, power efficiency, and on-time, first-pass quality are non-negotiable for me,” he stresses.
That is not rhetoric. It is a declaration of how Intel must measure itself, not by ambitions but by delivery. Customers are increasingly focused on the total cost of ownership. Intel’s roadmap will only matter if it translates into real-world benefits for those deploying AI at scale. That includes improved latency, reduced energy use, and systems supporting edge and cloud workloads.
The broader context also matters. Tan’s involvement with the MIT advisory board and his attention to power breakthroughs, including nuclear fusion, points to a CEO thinking in long arcs. But his focus remains on the now: delivering on promises, removing layers of hierarchy, and building momentum one product, one customer, one engineer at a time.
Intel’s transformation will not be instant. But Tan now has a leader who understands that the old ways of building compute no longer work in an era of intelligent systems. What comes next will not be shaped by brand legacy or organisational size but by the quality of questions Intel asks—and the speed with which it learns to answer them.
