OpenAI Raising $6.5B at $150B Valuation with $250M Minimum

Today on the AI Daily Brief, more information about OpenAI's monster new round. Before that in the headlines, Apple's new iPhone is here, but no intelligence is in sight the AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show. Notes. Welcome back to the AI Daily Brief Headlines edition. All the daily AI news you need in around five minutes. One of the big exciting shifts for many in the AI space has been Apple's entrance into the field. Now of course, Apple being Apple, they couldn't just say they were doing AI, they had to have their own name Apple Intelligence. But why people are excited about the possibilities here is that it seems like it's effectively AI for normies, right? This is an approach to artificial intelligence that is extremely focused on simple day to day valuable use cases that abstracts away a lot of the complication and if it works, could bring a lot of people into getting value from this field. The challenge, of course, was that after announcing Apple Intelligence, Apple said that the iPhone would be launching without Apple Intelligence actually included, at least not right away. Still, yesterday they began testing their software in beta. For those who had signed up for pre release tests and had iPhones. From last year, they could begin testing out the Apple Intelligence features, which are promising a lot. As Yahoo Finance puts it, once the iOS 18.1 upgrade is installed with Apple Intelligence, the often bumbling Siri is supposed to become more conversational, versatile and colorful, with a glowing light that will rotate around the iPhone screen as it responds to requests. However, even that is limited. They continue While Apple is promising, Siri will be able to perform more tasks and be less prone to becoming confused. It won't be able to interact with other apps installed on the iPhone until another software update comes out at a still unspecified date. Basically, even the test version of Apple Intelligence that is coming out is still very incremental and this has led to a lot of reviews like this one from Axios. Promising, but still has a lot to learn. There's a lot of phrasing like this, only modest improvements left me excited for the future, but also impatient for it. Still, by and large the main community reaction right now is the fail of not launching this phone with these features. SJV on Twitter writes, It's hilarious to me that after all the hype, the Verge review doesn't even mention Apple Intelligence in its overview. Why? Because, as TechRadar concurs, the AI launch is a joke, anything but seamless or even late. But better. It is AI Absence of Intelligence Matthew Berman Simply put it I find it absolutely bonkers that the entire iPhone 16 launch is centered around Apple Intelligence, but the feature isn't available until later this year. So for now, iPhone upgraders will have to wait. And even when Apple Intelligence does get pushed, it may not be the complete version over in the world of TikTok like every other big tech company, TikTok owner ByteDance is trying to start making its own GPUs to stop having to rely on Nvidia. ByteDance is reportedly developing two AI focused GPUs that will enter mass production by 2026. TSMC will actually manufacture the products. Broadcom, which has also built AI chips for Google, is expected to design the chips, and they're expected to be produced on one of TSMC's N4N5 process technologies, which is similar to the one used by Nvidia's Blackwell series. ByteDance has reportedly spent over $2 billion on more than 200,000 Nvidia GPUs. And in addition to the high prices, you have to think that part of the calculus also might be the fact that of continued and continually increasing, it seems, restrictions on American chip exports to China. One other issue to contend with. As Tom's hardware puts it, the massive catch about ByteDance's initiative is that the company now relies on Nvidia's CUDA and supporting software stack for AI training and inference. Once it goes with its AI GPUs, it must develop its software platform and ensure its software stack is fully compatible with its hardware, meaning this is an even more complicated process. Over in governance land, an AI advisory body at the United nations released a report proposing seven recommendations to address AI related risks. The report was called Governing AI for Humanity. Some of the recommendations include the creation of a body similar to the Intergovernmental Panel on Climate Change to gather up to date information on AI. Another recommendation is a new policy dialogue also recommending that the UN take steps to empower poorer nations, especially those in the Global south, to benefit and contribute. Concerns they flag include automating, disinformation, generating deepfakes, replacing workers, and exacerbating societal algorithmic bias. Now, whether these recommendations will actually go anywhere remains an entirely different question. Finally today, an interesting one. In the space between AI and energy, nuclear has come back on the table in a big way, especially since the start of AI. And now Bloomberg is reporting that the owner of Three Mile island, which is a nuclear plant that was the site of a 1979 accident, is spending $1.6 billion to revive it with an agreement to sell all of the output to Microsoft. The owner, Constellation Energy Corporation, expects Three Mile island to go back into service in 2028. Now it should be noted that Three Mile island has two units on site. One was permanently closed after that 1979 nuclear accident, and the other reactor was shut down in 2019 because it couldn't compete economically. Constellation is planning to reopen the reactor, which shut down in 2019. As the Seattle Times writes, the decision is the latest sign of surging interest in the nuclear industry as power demand for AI soars. More than a dozen reactors went dark over roughly the past decade in face of increasing competition from cheaper natural gas and renewable energy. But growing demand for electricity from factories, cars and especially from data centers has spurred interest in nuclear plants that can provide carbon free power around the clock. That Friends is going to do it for today's AI Daily Brief Headlines Edition. Next up, the main episode. Today's episode is brought to you by Fractional. When we wanted to build an AI powered feature of Superintelligent, our AI tool finder, I went straight to Fractional. The Fractional team is a group of senior engineers in San Francisco working on some of the most exciting projects in applied AI. Working with them is basically like hiring an absolute top flight AI engineering team, but in a way that you can customize exactly for your particular needs. Like I said, that AI tool Finder feature that we built with them is already a key part of the super intelligent platform and we are working on something new as well. Fractional works with everyone from startups to the Fortune 500 and to request a free consultation, you can go to Fractional AI. If you want help identifying and building AI projects for your business, then I highly recommend you hit pause, open a web browser and go to Fractional AI to request a free consultation. Today's episode is brought to you by Venice. Venice is a private, uncensored, generative AI app. It accesses open source models to enable text, image and code generation without the fear of being spied on or having your data exploited. Discuss anything with Venice without concern about it being monitored, sold or given to advertisers and governments. Venice is different because your conversations and creations are kept securely within the browser, never stored or accessible by Venice. Unlike other AI apps, Venice won't tell you what's okay to say or not. Venice won't patronize you. It simply provides direct access to machine intelligence. No topics are off limits, no ideas are taboo. With Venice, you're in control of the AI as you should be Pro subscriptions are available for $49 a year or $8 per month. AI Daily Brief listeners receive a 20% discount on Venice Pro. Visit Venice AI NLW and enter the discount code NLW DailyBrief. That's NLW Daily Brief. All one word. Today's episode is brought to you by Superintelligent, which is of course our platform that helps you learn how to use AI tools and prepare. Perhaps even more importantly, gives you ideas on the best use cases that are actually going to help you achieve whatever it is you want to achieve. To recognize the end of summer and back to school, back to work. We are running our best promotion ever. When you sign up for Super Intelligent between now and the end of August using Code so back, your first month will be 100% free. The platform features over 600 fun, highly practical AI tutorials that get you using AI fast and with an eye to actually transforming how you get things done. We've just launched super for Teams, so if you have a group of people at your company that want to figure out how to use AI together, I highly suggest you check it out. But for those of you who are using superintelligent as an individual, once again, if you sign up for superintelligent between now and the end of the month using Code Soback, you will get your first month 100% free. Go to be super AI and check it out today. Welcome back to the AI Daily Brief. For a while now we have been following the story of OpenAI's latest round. For some time it wasn't exactly clear whether this was just an employee tender offer or a more significant raise, but that has become clear now. It is definitely a more significant raise and the dynamics of it show that whatever concerns people may have had around AI hype getting too far ahead in the summer and OpenAI starting to surrender its lead to competitors. The latest Zero1 model and its other activities seem to have answered those entirely, as we will see. So first of all, the latest funding round seems set to value OpenAI at around $150 billion. That's obviously significant increase from the 80 to $90 billion of the last round. The size of the total round has also increased. We had previously heard that Thrive Capital was leading the round, but now it appears that the total financing will be somewhere between five and seven billion dollars, led by Thrive, which is committing a billion. Microsoft, Nvidia and Apple are all looking to invest the total aggregate contribution from which should be between 2 and 3 billion. One big question is what's the combination of cash and other resources like computing power that's going into that? And then a couple of other likely investors named are Tiger Global Management as well as mgx, which is an Abu Dhabi based AI firm. In the case of Tiger, it appears that they're doing this through a special purpose vehicle which will aggregate other investors into a single investment which which as the information puts it, could help investors make good on their commitments without overweighting their main funds in OpenAI shares. Now one of the new details is an apparent quarter billion dollar minimum investment. For some context on that, according to PitchBook, $250 million is six times larger than the median venture fund raised in the US last year. VC Matt Turk framed it this way In a Tweet he wrote 250 million in venture capital over the years 1990s Entire assets under management of a firm across funds 2000s Total size of a series A fund 30 plus companies 2020-2021 Total size of one round in one late stage company 2024 Size of each participant's investment in one round of one company now what we get from Bloomberg is that even if you do have a quarter billion dollars to spend in this, you're not guaranteed to get into the round. Bloomberg writes that investors were set to find out today whether they'd actually get into the deal. One of their sources said that excess demand was in the billions of dollars. So if the number does come in at around 6.5 billion, that means there was potentially billions of dollars more that were left on the table. Bloomberg also did note that there was one notable participant who won't be investing, which is Sequoia. Bloomberg frames it potentially as a choice to focus on rival company Safe Superintelligence, which was of course founded by OpenAI co founder Ilya Sutskever, who left the company earlier this year. Maybe also this is Sequoia making good on their concerns that they expressed in the $600 billion challenge blog post from over the summer. The Financial Times put the logic of all this into very simple terms. Venture groups, they write, believe OpenAI will eventually be the world's dominant AI company and worth trillions of dollars. Ultimately, this is really all handicapping the odds that OpenAI wins, which really means OpenAI continues to lead in a market that they believe is going to be worth trillions of dollars. In that light, $150 billion starts to look, if not reasonable, at least a lot less insane. The Financial Times added that it appears that Andreessen Horowitz is also sitting out of the round, one of the investors that the Financial Times spoke with said that generative AI represented the biggest platform prize since cloud or the Internet, worth multiple trillions of dollars of economic value. Another interesting note from anonymous investors in this conversation has to do with the introduction of Apple as an investor, FT writes. More important still could be closer ties to strategic investors, said one investor in OpenAI. OpenAI have Microsoft, the biggest enterprise company on the planet. If I could pick another partner, it would be Apple, the biggest consumer company on the planet. I'm walking into a gunfight with Google and Facebook and I have Microsoft and Apple behind me. It's not such a bad thing from a distribution and branding perspective. Now I'm sure we will get more information soon as that round finalizes, but in the meantime, there was also an interesting conversation between Sam Altman and leaders AT T Mobile at an event yesterday where Altman used the recently introduced framework of Levels to explain that O1 does represent their level two and that while the shift to level two took time, he thinks that that means that level three will be coming faster.

We have these five levels of AI we talk about. The first was chatbots, the second, which we've just reached now, is reasoners. The third is agents. The fourth is sort of innovators, the ability to figure out new scientific information. And the fifth is full organizations. So this move from one to two took a while, but I think the most exciting one of the most exciting things about two is that it enables level three relatively quickly after and the agentic experiences that we expect this technology to eventually enable I think will be quite impactful.

The information recently explored this in somewhat more detail. Writing One of the most important applications of Strawberry, which is of course now called O1, is to generate high quality training data for Orion. OpenAI's next flagship large language model that's in development using Strawberry or O could help Orion reduce the number of hallucinations or error it produces. Researchers tell me that's because AI models learn from their training data, so the more correct examples of complex reasoning they see, the better. Amir from the information also pointed out an Altman Easter egg around Orion from September 13th I love being home in the Midwest. The night sky is so beautiful. Excited for the winter constellations to rise soon. They are so great. When you ask chatgpt what constellations he might be referring to, it says he is likely referring to prominent constellations that are visible during the winter months in the Northern hemisphere. The first one it listed is of course Orion. Since we're on the topic, A couple additional pieces of OpenAI news to round it out. The miniature version of OpenAI's new reasoning model is, as the week has gone on, getting more and more people excited. Amir Efrati from the information again writes OpenAI's Week of Good news keeps getting better. After releasing an initial version of its O1 Reasoning Model 01 preview, new details are emerging about a smaller but faster version of the model, O1 mini. Turns out that O1 mini is even better than O1 preview in math. According to developers, it is generally on par in most other ways. One reason they write for O1 Mini's relatively mighty performance is that OpenAI allows customers to use more tokens compared to O1 Preview. O1 Mini's smaller size means it processes information more efficiently and more cheaply. Because of that, OpenAI can let O1 mini think longer than O1 preview, and that can help prove what the company says is the best part of its reasoning models. More thinking time equals better answers, otherwise known as log linear compute scaling. The efficiency gains OpenAI got from developing a miniature form of a reasoning model appear to be just as big of an accomplishment as proving out that reasoning concept. Lastly today, congrats to friend of the show Leah Belsky, a former Coursera executive who I met years ago in my first stint with Learn Capital who has just been recruited and hired by OpenAI to be their first General Manager of Education education and lead their efforts to bring ChatGPT into schools and classrooms. I for one am definitely on the optimistic side about how AI can help education and so I love that there is going to be more focus there. For now though, that is going to do it for today's AI Daily brief. Until next time, peace.