Artificial intelligence is reshaping industries faster than most strategic planning cycles can accommodate. Yet the breathless headlines about AI capabilities often obscure a more nuanced reality: the trajectory of AI development is shaped by tangible constraints—particularly the availability and cost of computational resources.
For strategy leaders, consultants, and executives, understanding these dynamics isn't optional. The organisations that navigate AI's evolution most effectively will be those that move beyond surface-level trend awareness to grasp the underlying forces shaping what's possible, when, and at what cost.
This guide provides a clear framework for understanding AI trajectories and compute constraints. You'll learn how to assess AI trends strategically, identify where constraints create opportunities or risks, and integrate these insights into your planning process. I've built Portage to help strategists work with exactly these kinds of complex, evolving signals—let me share what that work has taught me.
AI trajectories refer to the direction, pace, and pattern of artificial intelligence development. This encompasses advances in model capabilities (what AI can do), accessibility (who can use it and at what cost), and integration (how AI embeds into products, services, and workflows).
Compute constraints are the practical limitations on AI development imposed by the availability, cost, and efficiency of computational resources. Training large AI models requires enormous amounts of processing power, energy, and specialised hardware. These constraints directly influence which organisations can develop advanced AI, how quickly capabilities advance, and where innovation bottlenecks emerge.
Many strategic conversations about AI focus solely on capabilities—what the technology can do. But capabilities without context leads to flawed planning. A model that theoretically achieves superhuman performance means little if the compute required to run it costs more than the value it creates.
Think of it this way: AI trajectories describe where the technology is heading; compute constraints determine the road conditions along the way. Some paths are clear motorways. Others are unpaved tracks that only well-resourced travellers can navigate.
For strategists, this means asking not just "What will AI be able to do?" but "Who will be able to do it, when, and at what cost?"
The interplay between AI trajectories and compute constraints creates distinct strategic implications that warrant attention.
Investment timing becomes critical. Organisations that invest too early in AI capabilities may find themselves locked into approaches that become obsolete or uncompetitive as the landscape shifts. Those that wait too long may find themselves unable to access the compute resources or talent needed to catch up.
Competitive dynamics are reshaping. Compute constraints favour organisations with capital, infrastructure, and energy access. This is creating new forms of competitive advantage—and disadvantage—that cut across traditional industry boundaries.
Supply chain dependencies emerge. The concentration of AI chip manufacturing, rare earth mineral sourcing, and data centre capacity creates geopolitical and operational risks that strategists must account for.
Business model assumptions require testing. Many AI-enabled products and services depend on compute costs continuing to fall. What happens to your strategy if that trend slows or reverses?
Without a structured approach to monitoring these dynamics, strategic plans risk being built on assumptions that may not hold.
Integrating AI trajectory analysis into your strategic process requires moving beyond headline-level awareness. Here's a practical framework:
Start by mapping where AI intersects with your organisation's operations, products, and competitive environment. This includes:
Direct applications: Where you're using or considering AI Competitor positioning: How rivals are deploying AI Supply chain impacts: Where AI affects your partners or inputs Customer expectations: How AI is changing what buyers expect
Example: A professional services firm might identify exposure across client delivery (automation potential), talent acquisition (competing with tech firms for AI talent), and client expectations (pressure to deliver AI-enhanced insights).
Not all AI developments matter equally to every organisation. Focus your monitoring on trajectories that directly affect your identified exposure points:
Capability trajectories: Advances in reasoning, generation, perception, or decision-making relevant to your domain Cost trajectories: Changes in the economics of deploying AI for your use cases Accessibility trajectories: Shifts in who can access advanced capabilities (proprietary vs. open source, API access, etc.) Regulatory trajectories: Emerging governance frameworks that may enable or constrain AI adoption
Tip: Create a simple matrix mapping your exposure points to specific trajectories worth monitoring.
For each relevant trajectory, assess the compute constraints that shape its development:
Training constraints: What resources are required to advance the underlying models? Inference constraints: What resources are needed to deploy the technology at scale? Infrastructure dependencies: Where are the bottlenecks (chips, energy, data centres)? Concentration risks: Who controls the critical resources?
Example: If you're tracking advances in large language models, note that training frontier models requires billions of dollars in compute investment, concentrating development among a handful of organisations. However, inference costs for existing models are falling, broadening access to current capabilities.
Compute constraints aren't static. Build scenarios that explore how different constraint evolutions would affect your strategic position:
Constraint loosening: What if compute costs fall faster than expected? What opportunities open? Constraint tightening: What if chip shortages, energy costs, or regulatory restrictions slow progress? What risks emerge? Uneven distribution: What if certain regions or actors gain preferential compute access? How does this affect your competitive landscape?
AI trajectories evolve continuously. Establish a systematic approach to monitoring:
Track key indicators (chip production forecasts, energy prices for data centres, major model releases) Set trigger points that would prompt strategic reassessment Review your exposure mapping quarterly to capture emerging intersections
Apply your trajectory and constraint analysis to specific strategic decisions:
How would accelerated AI capability affect this investment's value? What assumptions about AI cost curves underpin this business case? Where are we dependent on compute access remaining stable?
A mid-sized manufacturer is evaluating significant investment in AI-powered quality control systems. Applying trajectory analysis, they identify:
Relevant trajectory: Computer vision capabilities for defect detection Constraint consideration: Current systems require substantial on-premise compute for real-time analysis Scenario insight: If edge computing advances (constraint loosening), costs fall and implementation becomes simpler. If energy costs rise (constraint tightening), operational economics worsen.
Strategic response: Rather than a large upfront investment, they pilot with cloud-based systems while monitoring edge computing developments. They build flexibility into vendor contracts to pivot as constraints evolve.
A strategy consultancy is considering launching an AI-enhanced strategy service. Their analysis reveals:
Relevant trajectory: Large language model reasoning capabilities Constraint consideration: Accessing frontier model capabilities requires either significant API spend or partnership with major AI providers Scenario insight: If open-source models close the capability gap (constraint loosening), differentiation becomes harder. If API costs rise (constraint tightening), margins compress.
Strategic response: They invest in proprietary methodologies that combine AI tools with human expertise, creating value independent of which specific models dominate. They avoid dependencies on single AI providers.
A regional health system is planning its five-year technology strategy. Trajectory analysis highlights:
Relevant trajectory: AI diagnostic capabilities in medical imaging Constraint consideration: Regulatory approval timelines and data privacy requirements create sector-specific constraints beyond compute Scenario insight: Even as technical capabilities advance rapidly, deployment constraints remain significant
Strategic response: They focus near-term investment on AI applications with clearer regulatory pathways (administrative automation) while building internal capability and data infrastructure to adopt clinical AI as constraints evolve.
Monitor leading indicators, not just announcements. Chip manufacturing capacity, data centre construction, and energy infrastructure investments signal future constraints before they manifest in AI capabilities.
Distinguish capability from deployment. What's possible in a research lab often takes years to become practical in enterprise settings. Build this lag into your trajectory analysis.
Watch the open-source ecosystem. The gap between proprietary and open-source AI capabilities directly affects competitive dynamics. Narrowing gaps democratise access; widening gaps concentrate power.
Consider second-order effects. Compute constraints don't just affect AI directly—they influence talent availability, geographic distribution of innovation, and competitive dynamics across adjacent sectors.
Avoid single-point forecasts. The uncertainty inherent in AI trajectories demands scenario-based thinking. Strategies that depend on a single predicted future are fragile.
Revisit assumptions regularly. The pace of change in AI means trajectory analysis has a short shelf life. Quarterly reviews prevent strategic drift based on outdated assessments.
AI trajectories intersect with several other strategic trend domains. Building a complete picture requires understanding these connections:
Geopolitical Power & Security Realignments: What Strategists Need to Know — Compute resources are increasingly central to geopolitical competition. Chip manufacturing concentration, export controls, and data sovereignty regulations all shape AI trajectories.
Macroeconomics & Capital Flows: Strategic Trend Analysis — Capital availability affects AI investment levels, while AI deployment affects productivity and labour markets. These economic dynamics interact with compute constraint evolution.
Societal Values & Demographics: Understanding Shifting Landscapes — Public attitudes toward AI, workforce impacts, and regulatory responses are shaped by—and shape—AI trajectories.
Monthly 'Top Trends' Report Template — A structured format for synthesising AI and compute developments alongside other strategic signals.
Portage Trend Index (Quarterly) — Quarterly synthesis tracking movement across major trend categories, including AI developments.
← Return to The Forces Shaping the Future: Portage's Strategic Trend Domains for the complete framework of strategic trend categories.
Start applying trajectory analysis to your current strategic priorities. Identify one decision or initiative where AI developments might materially affect outcomes. Map the relevant trajectories and constraints, then develop two or three scenarios to stress-test your assumptions.
If you want to systematise this approach, Portage's Trend Reports can help. The AI agent gathers research targeted to your key challenges, bringing together curated trends, your own files, and insights from across the web. You can configure reports to focus specifically on AI and compute developments relevant to your sector, delivered on your schedule.
Try generating a Trend Report on Portage to see how structured signal monitoring can inform your strategic process.
AI trajectories describe direction; compute constraints determine the pace and who can participate. Both are essential for strategic analysis.
Map your organisation's AI exposure before tracking general trends. Focus monitoring on developments that directly affect your strategic position.
Build scenarios around constraint evolution. What happens if compute becomes cheaper? More expensive? Concentrated in certain hands?
Distinguish capability from deployment. What's technically possible often takes years to become practically accessible in your context.
Integrate AI trajectory analysis into ongoing strategic monitoring. This isn't a one-time assessment—the landscape shifts continuously.
Stress-test strategic choices against multiple AI futures. Strategies dependent on a single predicted trajectory are fragile.