Oct 27, 2021 · 43m · mad
Top 10 Trends in AI, Machine Learning and Data for 2022
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this Data Driven NYC presentation, FirstMark venture capitalists Matt Turck and John Wu deliver a comprehensive breakdown of the 2021/2022 MAD (Machine Learning, AI, and Data) Landscape, identifying ten key macroeconomic, architectural, and funding trends shaping the enterprise technology ecosystem.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 56.5% of the talking time here. How this is scored →
Matt as informed peer 4.0 Guest teaching 1.9 Guest disagreement 0.0 Matt pushing back 0.0
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 37:18 John challenges standalone point-solution viability
John reframes the premise around standalone AI validation tooling, contending that isolated point solutions will struggle to compete against comprehensive end-to-end platforms.
Hardest push from Matt ▶ 9:33 Matt pushes back on consolidation inevitabilityMatt refuses the popular assumption that market consolidation is imminent, arguing that massive venture funding creates supply-demand imbalances that sustain vendor fragmentation.
Biggest teaching moment ▶ 27:55 John educates on Chinese homegrown AI infrastructureJohn educates the audience on China's government-driven 'Xing Chuang' localization movement, homegrown cloud architectures, and trillion-parameter models.
Matt holds his own ▶ 31:50 Matt details structured vs unstructured data convergenceMatt showcases high technical domain expertise by articulating the structural evolution and functional convergence of data warehouses and data lakes.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Opening Title Cards and Event Sponsorship | 5 | 0 | 0 | 0 | Matt introduces the webinar session, detailing Firstmark's early-stage venture investments in data unicorns like Ada and Cockroach Labs, as well as the history of the Data Driven NYC community. | |
| Evolution from Big Data to the MAD Ecosystem | 6 | 0 | 0 | 0 | Matt traces the decade-long evolution of the market map from Big Data and Hadoop to the modern MAD (Machine Learning, AI, and Data) acronym. | |
| Overview of the 2021 Top 10 Trends | 7 | 0 | 0 | 0 | Matt explains the macro trends driving the industry, including every company becoming a data company and the rise of cloud data warehouses like Snowflake enabling small firms to run modern data stacks. | |
| Trend 3 - Consolidation vs. Data Mesh and Hybrid Futures | 7 | 0 | 0 | 0 | Matt delivers an analytical breakdown of market consolidation versus point-solution fragmentation, explaining why high funding keeps potential acquirees independent. | |
| Trend 4 - The Explosive Funding Environment | 0 | 3 | 0 | 0 | John presents venture funding statistics for AI/ML startups before educating the audience on DataOps tooling, including data lineage and observability vendors like Superconductive and Anomalo. | |
| Trend 6 - It's Time for Real-Time Data | 5 | 2 | 0 | 0 | Matt details real-time data infrastructure market validation through Confluent's IPO and ClickHouse, after which John explains post-warehouse tooling such as metric stores and reverse ETL. | |
| Trend 8 - The Rise of AI-Generated Content | 0 | 4 | 0 | 0 | John presents commercial adoption examples of AI-generated content and large multimodal language models like Beijing's 1.75-trillion parameter WuDao model. | |
| Trend 9 - Evolution from MLOps to ModelOps | 0 | 3 | 0 | 0 | John explains the transition from MLOps to ModelOps, highlighting how ModelOps acts as a governance and explainability superset for non-technical business users. | |
| Trend 10 - The Emergence of a Separate Chinese AI Stack | 2 | 4 | 0 | 0 | John delivers a detailed overview of China's separate AI infrastructure stack and government-backed research, prompting lighthearted praise from Matt regarding John's language skills. | |
| Q&A - Data Evolution, Valuations, and AI Classification | 8 | 3 | 0 | 0 | In the audience Q&A, Matt demonstrates deep expertise explaining the technical convergence of Snowflake's structured data warehouse with Databricks' unstructured data lake. |
Statements from this episode (8)
Turck: Every company will need to become a data company
“To succeed in the future in the next few years and in the next few decades we think that every company is going to need to be not just a software company but also a data company.”
Turck: Big data, AI, and automation form a single continuous trend
“Ultimately it's all the same thing. It's all the same trend, which is basically turning companies from sort of like analog companies that run on, you know, sending Excel spreadsheets around with a lot of manual processes into companies that are intelligent and…”
Turck: Enterprise data market will remain fragmented, avoiding single-platform consolidation
“So the future that we see is, you know, as much as the Databricks and the Snowflakes of the world would want to be The one platform where all things data and all things AI happen we actually from our vantage point, see much more of a particular persistent subs…”
Turck: Confluent reached a $17 billion market cap post-IPO
“So Confluent which is the company behind Kafka which is like the real-time sort of message bus, like move, move data around in real-time had a wonderful IPO. They just checked today. There are like seventeen billion market cap.”
Turck: ClickHouse spun out of Yandex as an independent commercial company
“ClickHouse being a very popular open source project that was started at Yandex and just got spun out as a commercial for-profit for-profit company.”
Wu: Reverse ETL pipes warehouse data back into operational business systems
“Reverse ETLs change this paradigm by closing the loop. So these tools enable companies to pipe the transformed unified data from the warehouse back into the upstream business systems from which the data was generated.”
Wu: Point solutions proliferate in early-stage MLOps and ModelOps commercialization
“What, you know, within, within model ops and ML ops tooling is still in pretty early days in terms of commercialization, but from our perspective, we're seeing a lot more point solutions pop up around different parts of the machine learning and you know, broad…”
Turck: Snowflake and Databricks architectures are converging into direct competitors
“And the story of the last few years has been the great convergence of those two stacks. And in particular data breaks has worked on making their data lake more structured and look more like a data warehouse. And Snowflake is still in the early days of taking t…”