I don’t know what your social media feed looks like these days, but mine is filled with March Madness, memes, and a relentless catalog of AI startups’ impressive (and rapidly attained) revenue figures.  Here is a very incomplete selection of what I’m talking about here: In less than thr ee years, Midjourney’s annual recurring revenue (or ARR) soared from zero to $200 million.  Over about 24 months, ElevenLabs, a voice AI leader, saw its ARR soar from zero to near $100 million.  In three months, Lovable, an AI app-building startup, went from zero to $17 million in ARR.  And, perhaps most famously, AI coding juggernaut Cursor went from nothing to $100 million in ARR in one year. (For the uninitiated, ARR is the metric that counts how much predictable revenue a company can anticipate generating over one year, frequently from subscriptions or contracts.) These are just some of the numbers I’ve watched glaze excitedly around social media. And the excitement seems very fair—bolstered by AI, companies seem to have the capacity to grow faster than we’ve ever seen before. But I did wonder: How seriously should I be taking revenue growth that’s just so darn fast? Can something that good materialize that quickly? "The resounding answer is a firm 'it depends,'" said Rebecca Lynn, Canvas Ventures managing director. "This is really where our job as investors becomes both interesting and hard. We see, time and again, that even pre-revenue companies can raise a lot of money, but can’t necessarily get their demo into the wild. We’re really careful about that." Lynn continued: "Now, the next piece is revenue—we’ve seen a lot of companies scale very quickly, hit something like $30 million, and then flatline. So what you really have to ask is: Where are the end consumers? We really need to understand their usage, because you want to see that revenue is not ephemeral." Because here’s the thing—AI revenue right now means lots of different things. There are subscriptions, enterprise contracts, consumer revenue, and use-based arrangements. Across this spectrum, predictability varies. For example, usage-based billing in AI is pretty big and has for some companies facilitated otherwise impossible growth. But that doesn’t mean that revenue will be there in five years—or in five months. In short, this isn’t the hard-won, steady SaaS revenue of yore. "We’re seeing different pricing and revenue models, which are good because they have some flexibility and can align with direct value creation," said Lily Lyman, Underscore VC partner. "It’s not false momentum. The revenue itself is real and it’s important. But it doesn’t necessarily mean there’s long-standing, enduring value being delivered in the same way that ARR used to mean with enterprise SaaS, because it was so hard to get those contracts." So, for VCs investing in AI right now, revenue is a metric that’s a moving target.  "As a community, we had a pretty widely accepted set of benchmarks: 'If you're growing at X percentage year-over-year and you’re generating this much revenue, then your company should be valued at Y,’" said Anna Barber, a partner at VC firm M13. "I think those formulas are changing and there are too many factors at play for us to say we have it all figured out." She added: "What I know is that I don't rely as much on revenue as a symbol of company health. The competitive environment and landscape has expanded so greatly, so I'm looking a lot more at trying to draw the connection between the team, the product, and the place they started—and the ability to capture a really big market long-term." The excitement about what’s possible is justified, but that revenue momentum could also be fragile. If you’re running an AI startup right now, it’s hard to know where your competition will come from, too—if some of these AI tools are as general purpose as they seem, possible applications (and competitors) seem endless. The good news is that we’ll know more soon enough.  "One great thing about AI is how rapidly it evolves," said Neil Sequeira, founder and partner at VC firm Defy. "This has only been happening for a year or two, right? In the next 12 to 18 months, we’ll have some pretty good answers for this wave of verticals. You’re already starting to see how companies quickly enter markets, and how quickly competition starts to increase. You’ll get some answers as to predictability as best you can, because nothing is predictable." Like all metrics, revenue has always been a number that represented something deeper, a simple explanation for something far more granular. And tracking investor discourse about a metric is often a small way of tracking how the world is changing. It’s also a reminder that, for AI startups right now, the question isn’t just if they can make a profit today (most aren't, of course)—it’s who will still be standing tomorrow. Brainstorm AI London…Beyond my life in your inboxes, I co-chair our Brainstorm conference series, and programming our upcoming London event with the Brainstorm team has been an absolute blast. At Brainstorm AI London in May, I’m honored and thrilled we’ll be joined by top investors from Accel, GV, IVP, and Headline, along with the dynamic investor-founder duo of Sequoia’s Shaun Maguire and Decart’s Dean Leitersdorf. If you’re crossing the pond in early May (or are already there, lucky you!), let me know.  See you Monday, Allie GarfinkleX: @agarfinksEmail: alexandra.garfinkle@fortune.comSubmit a deal for the Term Sheet newsletter here. Nina Ajemian curated the deals section of today’s newsletter. Subscribe here. This story was originally featured on Fortune.com ...read more read less