Counting down to our AI rapture – TechCrunch


Welcome back to The TechCrunch Exchange, a weekly startups-and-markets newsletter. It’s inspired by what the weekday Exchange column digs into, but free, and made for your weekend reading. Want it in your inbox every Saturday? Sign up here

Happy Saturday, everyone. I hope that you are well, resting and relaxed.

Today we’re going to have some fun. Sure, down below we have our usual mix of venture capital rounds and notes and the like. But first, let’s talk about AI.

This week I got to chat to two different companies working on making machine intelligence a bit more concrete. One dealt with hardware, the other software.

On the hardware side of things, I chatted with IonQ’s Peter Chapman. IonQ is a quantum computing company that recently went public via a SPAC. However, instead of digging into redemptions and other blank-check minutiae, Chapman and I mostly discussed science fiction and what strong AI really means.

In simple terms, strong AI is not how Alexa works today. Alexa, per Chapman, works by having engineers code up lots of possible responses to queries. That scales for a while. But strong AI has to be able to write its own code, Chapman explained, which makes it fundamentally different than human-generated question-and-answer setups.

This fits into the quantum topic because quantum computing, Chapman said, is very good at the sort of code generation that strong AI will require. And, critically, it’s also pretty alright at parsing myriad probabilities at the same time and selecting between them.

All this is to say that quantum computing is reaching its early commercial stages, and companies like IonQ — named for its use of trapped ions in its technology — are helping usher in this new computer era. As quantum computing becomes more mainstream we should be able to get closer to the sort of AI that isn’t just ML models on a macro scale.

From the software side of things, I got on the phone with Intrinio CEO Rachel Carpenter. Her company has built a huge financial dataset that it makes available via an API. As a financial nerd, that’s cool. How much time you have spent reading SEC filings will determine if you care about that part of Intrinio or not.

But the startup is also building something called Thea, an AI service that works by weaving together neural networks into a custom natural language processing machine that can understand text. For folks looking to parse huge amounts of financial reporting, it’s a great product idea.

What struck me when talking to Carpenter was that Thea was initially trained on the larger internet. It’s not just a financial language parsing tool. It can do more.

Today the company is keeping Thea’s focus on its financial niche, per the CEO. But if Intrinio can spin up something that complex using partially open source services, we could see many more intelligent systems like Thea come to market in the coming years. Fuse that to increasingly commercial quantum computing tech and it seems, just maybe, that we’re possibly on the road to eventually, one day, maybe getting closer to actual artificial intelligence.

Yes, we were both born 50 years too early.

Venture capital this, venture capital that

As expected, the Q3 venture capital landscape was utterly bonkers. Flat crazy. Historically rich. Pick your phrasing.

So far Q4 is looking precisely and exactly the same. A sampling:

  • Notion has raised a $275 million round at a valuation of $10 billion, according to Alex Konrad over at Forbes. (Alex has been doing great work lately, I should note.) That’s effectively free capital. How so? Notion just sold 2.75% of itself for north of a quarter billion dollars. In terms of capital efficiency in dilution terms, that’s … cheap. Especially for a startup still too worried about its revenue scale to actually share the numbers. Notion still had a bunch of its last round in the bank before this round. So, it raised a quarter billy on spec, from investors who had that much capital lying around to place a  bet on the company managing an exit of at least $30 billion in time. Let’s see.
  • And this week Modern Treasury raised an $85 million Series C that values the “payments operations” fintech company at more than $2 billion. The company raised its Series B earlier this year when it was valued at roughly $300 million per PitchBook data. That’s a lot of value creation in a very, very short period of time! But it feels pretty fitting for what we’ve seen in recent months.

All this is to say that it doesn’t appear that Q4 is slowing compared to quarters two and three of this year. If 2022 doesn’t best 2021’s venture capital totals, I wonder how long it will be until we see this sort of investment again.

What a time to be alive.

In other news I wrote some more take-y stuff this week — here and here — if you want something with more bite.

Alex



Source