SpacetimeDB launched version 2.0 with a meme-style video mocking competitors and benchmarks that appear inflated. The database positions itself as an all-in-one database + application server, running application code inside the database instance. However, their benchmarks compare against traditional distributed databases using different trade-offs, making the comparisons misleading.

The article highlights that SpacetimeDB's storage engine uses an in-memory hash table protected by a single global read-write mutex, enabling linearizable writes but limiting concurrency. While writes are efficient due to local application logic and batching, readers may experience delays during high write throughput.

The author criticizes the dishonest benchmarking approach, noting that accessing data in-memory is inherently faster than network requests. They suggest focusing on technical breakdowns and honest trade-off explanations rather than exaggerated marketing claims.

Despite the questionable marketing tactics, the core concept of combining database and application server functionality presents interesting technical possibilities, though the implementation details around concurrency control remain underspecified.