# How Innovation Happens — Lesson #26: Anticipate, Don’t Follow By ## Article Content I've been covering the link between future trends and innovation strategies for 30+ years. I went back through my blog posts, stretching to 2002, to dig out the nuggets of innovation insight I've discovered along the way. This series revisits those stories, tests what still holds up, and turns them into practical lessons for what comes next. "By the time a trend is obvious, the opportunity has already passed." I've come back to this idea repeatedly in my work because the mindset behind innovation usually changes before the results do. What people notice, question, challenge, and act on determines how ready they are for what comes next. Trend-following feels safe because everyone can see the evidence. But once a trend is obvious, competitors can see it too. The larger opportunities usually appear earlier, when the signals are scattered, the business case is imperfect and the idea still sounds uncertain. That is why anticipation matters more than prediction. You do not need to know exactly what will happen. You need to recognize the direction of change early enough to build capability before everybody else does. Following explains the present. Anticipation prepares you for the next move. So what does that mean in practice? It means challenging the assumption, habit, or hesitation that is keeping you anchored to the present. Here are three places to start. Action Items: Track weak signals, not only established trends. Ask what is becoming possible before asking what is already popular. Build one capability ahead of demonstrated demand. Explore the full How Innovation Happens series. Real-World Examples These researched examples show the lesson at work in the real world. NVIDIA CUDA: Building the software platform before the AI boom NVIDIA was known for graphics processors used in computer games. But a graphics chip can also perform many calculations in parallel, which hinted at applications far beyond rendering images. The opportunity was unclear: most developers did not yet have a convenient way to use graphics hardware for general scientific and engineering computing. In 2006, NVIDIA introduced CUDA, a programming model that opened the parallel-processing capabilities of GPUs to researchers and software developers. The company spent years supporting the surrounding tools, applications and developer community. It was investing in a capability rather than waiting for one mass-market use case to justify the effort. In 2012, the AlexNet neural-network breakthrough used NVIDIA GPUs to train an image-recognition system and helped show what accelerated computing could do for artificial intelligence. NVIDIA's own history identifies CUDA as a major milestone that preceded this turning point by six years. The crucial insight is that anticipation does not require predicting the exact winning application. NVIDIA recognized a direction of technological change and built a platform that made new applications possible. When deep learning took off, the company was not starting its developer ecosystem from zero. Why it matters: Anticipate the capability that will become valuable, even when the eventual winning application is still uncertain. Sources: ASML: Investing for decades in light that industry could not yet use Shrinking semiconductor features eventually runs into the physical limits of optical manufacturing. Extreme ultraviolet lithography promised a path forward, but the engineering was extraordinarily difficult. It required new optics, light sources, contamination controls and a complex network of specialized suppliers. For years, commercial success was far from certain. ASML began a dedicated EUV effort in the late 1990s and worked through international research partnerships. A prototype tool shipped in 2010, and the company bought light-source specialist Cymer in 2013 to accelerate development. Production-ready systems began attracting customer orders in 2016, long after the original research effort had begun. ASML describes the effort as four decades of science and engineering, billions invested and thousands of contributors across firms and laboratories. No one milestone alone made the technology inevitable. The advantage emerged because the firm sustained an interlocking set of capabilities through years when the payoff was uncertain. This is a counterexample to the habit of waiting until a trend appears in quarterly sales data. By the time advanced chipmakers needed EUV at scale, the underlying ecosystem could not have been assembled overnight. ASML had made time itself part of its competitive advantage. Why it matters: The best early investment may be in an enabling capability whose value compounds before the market is obvious. Sources: --- --- Source: https://jimcarroll.com/2026/11/how-innovation-happens-lesson-026/