The "Unsexy" Nature of Foundation Software and Its Real Value
A Feeling That Doesn't Sit Well
During my time working on RobustMQ, I've often felt a bit uneasy.
When I tell friends we're building a message queue, the reaction is often "Oh, you're still doing that." If I said we're doing AI, their eyes would light up: "So cool! AI is so hot right now!" That contrast sometimes makes me wonder: Did we pick the wrong direction? Message queues—foundation software—feel dull, unsexy, unimaginative. Unlike AI: funding news everywhere, disruptive applications everywhere, so much possibility.
Anyone building foundation software has probably felt this. You're seriously solving technical problems, optimizing performance, improving stability—but from the outside it looks boring. You say you've optimized zero-copy; people don't know what that means. You say you've achieved migration-free scaling; people think "what's so hard about that?" You say you've cut cost with tiered storage; they ask why not just use Kafka.
Why Foundation Software Feels "Unsexy"
On reflection, there are reasons.
It's too low-level for most people to notice. Users experience ChatGPT directly and feel AI's appeal. They don't interact with message queues. They see apps and features; they don't care what runs underneath. Like using a phone without caring about the chip architecture—you only care if it works well.
Value isn't obvious. AI generates images, writes text, holds conversations—immediate impact. A message queue done well gives you "the system is stable"; done poorly you notice "why is it so slow." Its value is negative: when it works, that's expected; when it fails, you notice. That invisible value is hard to appreciate.
High technical bar but low imagination. Distributed systems, consistency protocols, zero-copy optimization—all technically hard. But to outsiders it's just "making messages go faster and more reliably." AI can say "changing the world" and "disrupting industries"; message queues can't tell that story. The imagination gap is real.
But Does Foundation Software Have No Value?
I often reassure myself: foundation software may not be sexy, but it matters.
Those sexy AI applications—what do they run on? OpenAI uses 30+ Kafka clusters for ChatGPT, Flink for streaming, message queues for user feedback data flow. Those flashy real-time recommendations and personalized services—message queues move the data. Those "intelligent" systems—where does data come from, how does it move, how is it stored? Foundation software.
AI is the house; foundation software is the foundation. The foundation isn't sexy. But without it, you can't build. I've used this analogy often, but honestly sometimes I'm not fully convinced. Everyone looks at how pretty the house is; few care how deep the foundation goes.
Real Doubts
I wonder sometimes: Should we pivot to something sexier?
In message queues, Kafka is the de facto standard; Pulsar and RocketMQ are mature. Building a new queue—even if technically better—faces real barriers: market acceptance, ecosystem, migration cost. And truthfully, most user needs are already met; what's left are refinements (faster scaling, lower cost), not revolutionary demand.
Meanwhile, AI is full of opportunity and imagination. Models evolving, applications exploding, startups everywhere. AI-adjacent work gets funding, hires, and stories more easily.
So I've thought: Should we do something sexier? AI data platform, real-time feature engineering, intelligent data governance—at least touch AI, easier to pitch.
Back to the Core Question
Does RobustMQ have breakthrough innovation? No. Does it need it?
My honest answer: we'd like it, but we haven't found it. I wish we had something original that could truly break through, so RobustMQ isn't just "another message queue." We've thought a lot and haven't gotten there. The space has evolved for years; the main approaches have been explored.
I think our current design is the limit of what we can conceive. Without taking technical risk or going too aggressive, we're reasonably composing existing tech to solve known problems and fit more scenarios. There may be better ideas we haven't thought of; the future may bring new options. For now, this is what we can do.
Making Peace
Maybe foundation software really isn't sexy and doesn't have AI's imagination. Maybe that's just how it is.
Foundation software's value isn't in imagination—it's in reliability. Not in how new the concepts are, but in how well problems are solved. Not in the story you tell, but in whether the system runs stably. These qualities are "boring"—but important.
We chose message queues, chose foundation software. That may mean we'll never be that sexy. But if we build a message queue that's truly stable, high-performance, low-cost, and easy to use—one that helps other systems (including those sexy AI apps) run better—that has value.
Maybe that value won't be widely seen, won't make headlines, won't get much attention. But for the people who really use it, for the systems that depend on it, it matters.
A Bit of Stubbornness
Even though foundation software isn't sexy, I still want to keep going with RobustMQ.
Not because I've figured everything out, or because I'm sure we'll succeed. Just because we chose this path, we've thought through the design for so long—let's see how far we can go. Even if we never displace Kafka, even if we only find users in some niche, that's still value.
Sexy or not, imagination or not—that may not be up to us. What we can control: treat every technical choice seriously, refine every line of code, face every problem honestly. Do what we should, and leave the rest to time.
About RobustMQ
RobustMQ is a message queue written in Rust. Not sexy, but we're serious about it. The project is early; many ideas are still being validated. If you care about foundation software too, follow us on GitHub—let's build something "boring but useful" together.