Although several startups and tech companies are developing and selling ML platforms, a market leader has not yet emerged. Consequently, ML engineers continue to face many challenges when building and operationalizing models. In this post, Clemens Mewald posits that the adoption of ML platforms is stymied by a lack of dominant design – industry has not settled on the key components of an ML system, the relative importance of each component, or how they should be linked together. He also notes that AI developer tools have inconsistent form factors (e.g. that require users to interface with different model frameworks, orchestration engines, distributed processing tools, etc.) or inappropriate form factors (i.e. that provide the right level of abstraction, given user expertise). He predicts that ML platforms will not succeed until these issues are resolved.