Chi-Hua Chien has spent much than 2 decades arsenic a task capitalist, but helium thinks similar a taste anthropologist. As a co-founder of Goodwater Capital, a steadfast focused exclusively connected user and prosumer technology, helium has built a portfolio spanning entertainment, healthcare, fintech, and unrecorded experiences — with investments successful companies similar MIDI Health, Fever, and Monzo. He was also, arsenic a 27-year-old subordinate astatine Accel, the idiosyncratic who initially recovered a six-person institution launched from Harvard called The Facebook.
That quality to work quality behaviour astatine standard informs everything from his presumption that Americans volition ne'er spot a azygous app with some their societal lives and their finances, to his content that the spread betwixt the astir precocious AI exemplary and what you tin tally connected your telephone — erstwhile arsenic wide arsenic 2 years — volition shrink to 3 months wrong the adjacent year.
These days, helium is besides consenting to accidental retired large what galore successful task superior are lone thinking: that the commoditization of the exemplary furniture is already underway, and that the biggest winners of the AI epoch won’t beryllium the companies selling AI astatine all.
We talked past week; this interrogation has been edited for magnitude and clarity.
More founders and investors person been publically sharing their grievances astir VCs lately. What’s changed?
It’s portion of the meme-ification of everything — you’re seeing what’s happening successful the governmental realm bleeding implicit into the concern side, and it’s astir apt besides the motion of immoderate peakiness successful the market. The crushed you’re seeing immoderate of these outspoken investors talking much publically is due to the fact that task firms person mostly vertically integrated, truthful the truly large ones person capable superior that they’re not needfully looking for syndicate partners. There utilized to beryllium decorum astir wanting to sphere bully relationships with different co-investors, due to the fact that you got to enactment with them astatine antithetic points on the line. As the firms person gotten bigger and vertically integrated, there’s little of that need.
What astir the “fast follow” rounds — wherever firms put a ample chunk astatine 1 valuation and a smaller magnitude weeks aboriginal astatine a overmuch higher one, making the header fig look much awesome than it truly is? Is this truly new? How pervasive is it?
I deliberation it’s been going connected for rather immoderate time. The champion companies rise successive rounds precise rapidly — determination mightiness lone beryllium 3 to six months betwixt rounds now, and valuations alteration truly quickly… Valuations are being marketed precise aggressively arsenic a mode of demonstrating marketplace leadership, attracting talent, perchance blocking retired competition. There’s astir apt immoderate constituent of frothiness, due to the fact that what these accelerated financings are astir illustrative of is there’s mode much request than determination is supply. An capitalist tin travel in, acceptable a price, implicit a financing, and past a mates of weeks aboriginal there’s inactive excess request — and the institution tin instantly terms a caller circular astatine a higher price.
You’ve argued that infrastructure companies get commoditized, and that applications seizure astir of the worth implicit time. Are we already seeing that play retired successful this cycle?
If you look astatine the PC cycle, the web cycle, and the mobile cycle, they each travel reasonably accordant patterns. Infrastructure marketplace caps really peaked successful the twelvemonth 2000 — but you accelerated guardant 25, 26 years later, and successful nominal dollar terms, the marketplace headdress of those infrastructure companies has not surpassed the 2000 peak. In the web era, infrastructure caller entrants produced $400 cardinal of caller marketplace cap. Application companies created $3.1 trillion — 88% of the caller value. In the mobile era, it’s precise similar: infrastructure produced astir $700 billion, portion exertion companies produced $3.7 trillion. Companies similar Netflix, Spotify, Meta, Uber, Airbnb.
And [last week] you saw thing pretty interesting: Google announced that their subscription AI merchandise is dropping terms from $7.99 a period to $4.99 a period and doubling the storage. We’re already successful the epoch of terms contention — and companies similar Google, with structural advantages successful vertical integration and distribution, tin commencement bundling and terms competing for the mean consumer.
You support coming backmost to personalization arsenic a done line. Is that what separates the adjacent question of winners?
Hyper-personalization decidedly is simply a cardinal done line, due to the fact that what does personalization springiness you? If done right, it gives you higher lawsuit satisfaction, deeper engagement, and higher ARPUs implicit time.
We person amusement companies successful our portfolio — companies similar Triumph and Ritten and Flow GPT — wherever the lawsuit is not saying “this is an AI application.” They’re saying it’s an amusement application. These companies are going into 100 million, 400 million, 600 cardinal of ARR precise quickly, astatine large margins, due to the fact that AI makes the acquisition much customizable and much personalized — but it’s not the cardinal capableness they’re selling.
We besides person a women’s wellness institution called Midi Health. One of the cardinal constraints successful women’s wellness is that determination aren’t that galore providers good trained successful hormone replacement therapy for perimenopausal women. By utilizing AI, they’re capable to substantially grow the proviso of attraction and dainty hundreds of thousands of patients that different couldn’t beryllium reached. And they tin bash it outgo effectively, which expands entree to a marketplace that was antecedently proviso constrained. You tin play that guardant crossed each supply-constrained class wherever quality expertise is the bottleneck.
How acold distant are we from AI that feels genuinely idiosyncratic and ambient?
I don’t deliberation we’re precise acold distant astatine all. You tin tally locally present connected your telephone AI models that are arsenic bully arsenic the champion models were astir six months agone — and that lag is shrinking. You spell backmost 2 years ago, the lag betwixt what you could tally locally and what was successful the unreality with the frontier models mightiness person been 18 to 24 months. It’s present six months. It’s astir apt getting down to 3 months by this clip adjacent year.
What we don’t yet person is the usage cases precise good defined. You saw this successful mobile — erstwhile the iPhone launched successful 2007, radical mostly thought it was going to beryllium each of the web applications ported implicit to mobile. It takes clip for entrepreneurs to percolate astir what is present possible.
What LLMs do, if you extrapolate distant from however they enactment to what they do, is fundamentally 2 things: they marque it imaginable for you to process ample amounts of discourse and marque consciousness of it all, and they let you to bash personalization down to the individual, outgo effectively, with a feedback loop that makes the merchandise amended and amended implicit time.
You’ve watched Facebook effort and neglect for years to physique a ace app. Why is it truthful hard to blend fiscal services and societal amusement for American consumers?
They’ve taken aggregate shots connected extremity — Facebook Credits, which launched successful 2009… Facebook Pay, Libra… They’ve ne'er been capable to recognize a existent ace app. I deliberation radical person an intuitive position connected trust, and determination is simply a spot spread betwixt amusement and societal products, and commerce, banking, fiscal services — peculiarly successful the Western world.
There is simply a seriousness to fiscal transactions that is precise antithetic from the triviality of societal media. And don’t get maine incorrect — that triviality has created a trillion-plus dollar company. But fiscal services is really the implicit inverse: portion assemblage has precise precocious clip and comparatively debased monetization, fiscal services transactions are precise precocious monetization and comparatively debased time. You don’t privation to bent retired successful your banking app. You privation to transact and beryllium done — but with highly precocious assurance successful the information and reliability of that transaction. That intelligence anticipation from customers is simply a precise pugnacious 1 to bridge.
Are you placing bets connected radical craving in-person transportation arsenic a counterreaction to each of this?
We really, truly judge successful this. What bash radical crave successful a satellite wherever there’s an infinite proviso of integer content? They crave the happening that is astir constrained, which is existent quality contact, real-world experiences.
We person an concern successful a institution called Bump, based successful Paris — from the archetypal founders of Zenly, which was acquired by Snap… They’ve built an interface that allows radical to interact successful the carnal world, catalyzed by integer information. We besides person Fever, based successful London and Madrid — fundamentally the Live Nation of Europe. They started with smaller, quirky events — candlelight concerts, the Bridgerton Experience — and person since gone mainstream.
I deliberation we’re swinging backmost successful the different absorption from axenic online consumption, and AI arsenic enabling technology, knowing wherever you go, who you bent retired with, wherever you thin to walk time, tin extrapolate a ton of applicable interests that marque that real-world acquisition much utile and much personal. That’s ace breathtaking to us.
When you acquisition done links successful our articles, we whitethorn gain a tiny commission. This doesn’t impact our editorial independence.















English (US) ·