Intel's AI-first pivot: what executives should track after CES
Intel is reworking its strategy around AI, backed by a historic U.S. investment and public support from President Donald Trump. Under new CEO Lip-Bu Tan, the company is pushing beyond PCs and into edge devices and robotics - areas it believes can reset its trajectory against Nvidia, AMD, and Qualcomm.
The centerpiece is the Core Ultra Series 3. Intel wants it in nearly every new laptop this year, while also embedding it in devices that sit between the PC and the cloud. As Jim Johnson, head of Intel's client computing group, put it at CES: the number of edge devices is "almost infinite."
What's new: Core Ultra Series 3
Intel is attacking two buyer priorities: better battery life and faster on-device AI for real workflows. Think coding assistants and video calls that auto-improve audio and video without offloading to the cloud. The chip is slated to power 200+ new PC designs in 2025.
The bet is simple: move everyday AI tasks onto the device to cut lag, improve privacy, and reduce cloud costs. For large enterprises, that's a direct lever on total cost of ownership and user experience - if performance holds up at scale.
Beyond PCs: the edge and robots
Intel is targeting the gray space between endpoints and data centers: robots, industrial systems, and specialized devices that need low-latency inference. Oversonic Robotics plans to shift from Nvidia to Intel's Core Ultra 3 for on-device processing. Intel says the switch lowers costs and speeds responses because robots won't need to round-trip to the cloud for inference. Training still happens on Nvidia.
If this pattern repeats across sectors - healthcare, logistics, manufacturing - it creates a second growth lane for Intel beyond PCs. The open question is how quickly real deployments scale.
The competitive reality
- AMD introduced laptop chips at CES that can run larger AI models locally, aiming to reduce cloud reliance.
- Qualcomm, a smaller PC player, is pushing new laptop silicon with multi-day battery claims and AI-centric design.
- Nvidia still dominates data-center AI and is doubling down on robotics with new models and showcases across industries.
Intel remains the top PC chip supplier with more than 71% market share in 2024, per IDC. But it's under pressure: Apple exited Intel chips for Mac in 2020, AMD is competitive in premium laptops, and Intel cut 15% of staff last year. The turnaround must be product-led, not just narrative-led.
Leadership and operating cadence
Tan has set a direct, customer-first cadence. Johnson says he's been asked to escalate wins and problems in real time. For enterprise buyers, that signals tighter feedback loops and potentially faster product fixes - a key gap in prior cycles.
Policy tailwind and investor sentiment
The U.S. government took roughly a 10% stake in Intel last year and is publicly backing domestic chip manufacturing. That support can steady long-cycle capex decisions and supplier confidence. Wall Street has noticed: Intel shares climbed about 84% in 2025 and are up roughly 98% year-over-year, though they're still down more than 18% over five years.
What this means for your roadmap
- Device strategy: Test Core Ultra 3 systems for on-device AI tasks like meeting enhancement, code generation, and secure summarization. Compare real-world battery life, thermals, and sustained performance against AMD and Qualcomm alternatives.
- Hybrid AI architecture: Split workloads. Keep sensitive, latency-critical inference on devices; send large, bursty jobs to the cloud. Model selection and token budgets should reflect this split.
- TCO and privacy: Quantify cloud egress savings and privacy gains from local inference. Factor in model update cadence, driver stability, and IT operational overhead.
- Risk and vendor balance: Avoid single-vendor lock-in across training (Nvidia-heavy today) and inference (PCs, edge). Keep at least two viable paths for key workloads.
- Robotics: If you're piloting humanoids or mobile robots, start with constrained, high-ROI tasks. Validate safety, compliance, and field service before scale. Analyst views suggest real deployments are still sparse.
- Procurement timing: If you refresh fleets mid-year, run lab benchmarks by workload, not just synthetic tests. Prioritize user-perceived performance (cold starts, context windows, battery under load).
Signals to watch in 2025
- OEM adoption and attach rates for Core Ultra 3 in business-class laptops.
- Independent battery and on-device AI benchmarks across comparable AMD and Qualcomm systems.
- Enterprise case studies for edge deployments (robots, industrial clients) with audited cost and latency data.
- Driver stability and AI framework compatibility over the first two quarters post-launch.
- Nvidia's pace in data-center AI and robotics, and any shift in training-to-inference economics.
90-day action plan
- Run a controlled pilot of 100-300 users on Core Ultra 3 systems with AI-heavy workflows (engineering, sales, support). Measure time-to-task and battery deltas against current devices.
- Stand up an edge inference testbed. Compare on-device vs. cloud cost, latency, and privacy for your top five AI use cases.
- Update your procurement scorecard to weigh AI-on-device metrics alongside CPU/GPU specs.
- Define a two-vendor strategy for both training and inference to keep leverage and supply resilience.
- Align security and compliance for local AI processing, including data retention and audit trails.
Bottom line
Intel is pushing hard to win the AI edge while defending its PC base. The plan makes sense: better battery life, credible on-device AI, and a path into robots and specialized clients. Execution will decide the outcome - and your pilots over the next 90 days will tell you whether Intel's new stack fits your cost, privacy, and performance targets.
Further learning for leadership teams: explore role-based AI upskilling to operationalize on-device AI across functions. Courses by job and popular certifications can help your teams move from testing to production with fewer false starts.
Your membership also unlocks:
