a16z Podcast: Big Data Goes Really Big

Episode Summary

Executive Summary: The episode argues that big data is entering a “2.0” phase as it moves from on-premise infrastructure into the cloud, making analytics more accessible, collaborative, and cost-effective. Pratt Mogai and Peter Levine say this shift will democratize decision-making across companies, governments, and smaller organizations, while enabling faster, more agile responses, lower costs, and eventually new machine-learning and application-aware data products.

Main Topics: Why big data is slow to move to the cloud (Priority: 5/5): Big data has been harder to migrate than CRM or ERP because it is embedded across many workflows, involves complex data movement, and raises security concerns around customer, employee, and patient information. Big Data 1.0 vs. Big Data 2.0 (Priority: 5/5): Peter Levine frames the legacy on-prem model as Big Data 1.0 and the cloud-native model as Big Data 2.0, analogous to the shift from on-prem software to SaaS. Democratization of analytics and decision-making (Priority: 5/5): Both guests emphasize that cloud-based big data broadens access beyond elite data teams, letting more employees and smaller organizations ask questions and act on data directly. Business value: speed, cost, and culture (Priority: 4/5): The speakers argue that cloud big data improves agility, reduces capital expenditure, and creates a more collaborative culture by giving multiple teams a shared view of data. Use cases beyond corporations (Priority: 4/5): The discussion extends to government and social services, including predictive policing examples, showing that cloud analytics can improve public outcomes and lower costs. Technical and strategic barriers (Priority: 4/5): Enterprises face a stacked challenge: they must not only adopt cloud infrastructure securely but also choose among SQL, Hadoop, Spark, and other technologies without forcing data to bounce back and forth. Next wave after cloud big data (Priority: 3/5): Once big data is in the cloud, Levine predicts future layers such as machine learning, machine intelligence, and application-aware big data that is built directly into products and workflows.

Key Arguments: Big data is difficult to “lift and shift” because it is horizontally embedded across operations rather than confined to a single application. Legacy big data systems require expensive on-prem data centers and specialized infrastructure knowledge, limiting adoption to large incumbents. Cloud big data can make organizations faster by giving users real-time or near-real-time access to data instead of waiting months for reports. Democratized data access flattens hierarchies and encourages a more collaborative, Google-like culture across departments. The shift to cloud big data will benefit small and mid-sized companies that currently rely on spreadsheets and lack the budget or expertise for data centers. Moving data to the cloud is especially valuable when the compute happens where the data lives; hopping data back and forth creates cost and performance friction. The value proposition is not about one winning technology, but about solving workloads with the right mix of SQL, Hadoop, Spark, and cloud platform support. Government and civic organizations can use cloud analytics to improve safety and efficiency, as illustrated by predictive policing examples. Big data in the cloud is positioned as the foundation for future intelligent applications and machine-learning layers.

Data Points: Big Data 1.0 market size: $10 billion - Pratt describes the legacy on-prem big data market as a sizable but incumbent-dominated space. Deployment cycle time: 6 to 9 months - Used to illustrate how long it can take enterprises to rack, stack, and deploy big data projects on-prem. Annual spend: Millions of dollars per year - Refers to the cost of traditional on-prem big data infrastructure and operations. Data growth rate: 200% every year - Highlights how rapidly enterprise data is expanding relative to slow deployment cycles. Relative cost target: One-fifth the cost - Pratt says cloud big data can enable much cheaper rollout versus traditional approaches. Organization size impacted: Top 2,000 enterprises / hundreds of thousands of smaller organizations - The discussion contrasts large incumbent advantages with the broader democratization potential for SMEs.

Pivotal Quotes: "The implication of big data, cloud, making it really democratic is access to everyone, flattening of the organization, collaborative culture, and ultimately faster decision making." — Pratt Mogai: Pratt explains the organizational impact of moving analytics into the cloud. "Big data 2.0 is moving big data from on-prem into the cloud." — Peter Levine: Peter defines the new phase of big data and compares it to the SaaS shift. "We all have access to these great tools and services." — Peter Levine: Peter contrasts cloud access with the limited access of traditional on-prem systems.

Implications: Cloud big data could make analytics a standard capability for most organizations, not just tech giants. Expect faster decisions, lower costs, broader public-sector use, and new intelligent applications built directly on shared cloud data.