Tech Brew Ride Home - Fri. 02/16 – Sora Brings Text To Video AI To Life

Episode Date: February 16, 2024

Huge watershed reveal of OpenAI’s Sora, its first text-to-video model, which can create up to a minute of 1080p video. We used to have a Today In Elon segment. We’re close to needing a Today In Sam Altman one. Increasing signs the Crypto winter is ending. And, of course, the Weekend Longreads Suggestions. Links: OpenAI introduces Sora, its text-to-video AI model (The Verge) OpenAI’s Sora Turns AI Prompts Into Photorealistic Videos (Wired) OpenAI’s newest model Sora can generate videos — and they look decent (TechCrunch) OpenAI’s Sam Altman Seeks US Blessing to Raise Billions for AI Chips (Bloomberg) Sam Altman owns OpenAI's venture capital fund (Axios) Crypto exits remain low but investors remain unfazed (TechCrunch) Crypto venture funding climbs for first time in nearly 2 years after bitcoin’s stellar run (CNBC) Weekend Longreads Suggestions: The fediverse, explained (The Verge) How an African streaming service dethroned Netflix (RestOfWorld) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Tech meme right home for Friday, February 16th, 2020. I'm Brian McCullough today.

Huge watershed reveal of OpenAI's Sora, its first text-to-video model, which can create up to a minute of 1080P video. We used to have a Today in Elon segment. We're close to needing a Today in Sam Altman segment. Increasing signs the crypto winter is ending, and of course the weekend long read suggestions. Here's what you miss today in the world of tech. I mean, these things are going to keep coming every week at this point,

but this one seems epoch-making. OpenAI has unveiled SORA, its first text-to-video model, which can create up to a minute of 1080P video from text prompts. Quoting the verge, SORA is capable of creating complex scenes with multiple characters, specific types of motion, and accurate details of the subject and background, according to OpenAI's introductory blog post. The company also notes that the model can understand how objects, quote, exist in the real world, as well as accurately interpret prompts and generate compelling characters that express vibrant emotions. The model can also generate a video based on a still image, as well as fill in missing frames on an existing video or extend it. The sort of generated demos

included in OpenAI's blog posts include an aerial scene of California during the Gold Rush, a video that looks as if it were shot from the inside of a Tokyo train, and others. Many have some telltale signs of AI, like a suspiciously moving floor in a video of a museum, and Open AI says the model, quote, may struggle with accurately simulating the physics of a complex scene, but the results are overall pretty impressive, quoting Stephen Levy in Wired. Other companies from giants like Google to startups like runway have already revealed text-to-video AI projects, but Open AI says that Sura is distinguished by its striking photorealism, something I haven't seen in its competitors, and its ability to produce longer clips than the brief snippets other models typically

do up to one minute. The researchers I spoke to won't say how long it takes to render all that video, but when pressed, they described it as more in the going out for a burrito ballpark than taking a few days off. If the hand-picked examples I saw are to be believed, the effort is worth it. OpenAI didn't let me enter my own prompts, but it shared four instances of Sora's power. none approached the purported one-minute limit. The longest was 17 seconds. The first came from a detailed prompt that sounded like an obsessive screenwriter's setup. Beautiful snowy Tokyo City is bustling. The camera moves through the bustling city street, following several people enjoying the beautiful snowy weather and shopping at nearby stalls. Gorgeous, Sakura petals are flying through the wind along with

snowflakes. The result is a convincing view of what is unmistakably Tokyo in that magic moment. when snowflakes and cherry blossoms coexist. The virtual camera, as if affixed to a drone, follows a couple as they slowly stroll through a streetscape. One of the passers-by is wearing a mask. Cars rumble by on a riverside roadway to their left, and to the right, shoppers flit in and out of a row of tiny shops. It's not perfect. Only when you watch the clip a few times, do you realize that the main characters, a couple strolling down the snow-covered sidewalk, would have faced a dilemma had the virtual camera kept running. The sidewalk they occupy seems to dead end. They would have had to step over a small guardrail to a weird parallel walkway on their right.

Despite this mild glitch, the Tokyo example is a mind-blowing exercise in world building. Down the road, production designers will debate whether it's a powerful collaborator or a job killer. Also, the people in this video who are entirely generated by digital neural networks aren't shown in close-up, and they don't do any emoting. But the SORA team says that in other instances, they've had fake actors showing real emotes. motions, end quote. And quoting TechCrunch. OpenAI's very much positioning SORA as a research preview, revealing little about what data was used to train the model short of around 10,000 hours of high-quality video, and refraining from making SORA generally available. Its rationale is the potential for abuse. Open AI correctly points out that bad actors could misuse a model like Sora in myriad ways.

Open AI says it's working with experts to probe the model for exploits and building tools to detect whether a video was generated by SORA. The company also says that should it choose to build the model into a public-facing product, it'll ensure that Providence metadata is included in the generated outputs, end quote. From the ambitions of Sam Altman file, which is pretty much full to burst thing at this point, Bloomberg says Altman is working to get U.S. approval for his chip project as the involvement of the United Arab Emirates and other countries risks raising national security and antitrust concerns. Quote, Altman has indicated he believes it's essential to work collaboratively with the U.S. government on approvals, timing and structure for the venture, according to one person familiar

with the matter who asked not to be named discussing private conversations. The CEO has met with Commerce Secretary Gina Riemondo and is working to arrange meetings with other officials, the people said, and commerce officials have held internal discussions on OpenAI's Middle East ambitions. The Wall Street Journal previously reported the two had met. Altman's ambitious fundraising push risks triggering a national security review of foreign investment by a committee chaired by the Treasury Department and could run up against the Commerce Department's controls on ship shipments to the Middle East.

Allman is also considering whether to create an issue new equity in a new company, separate from OpenAI, some of the people said, in a move that might raise antitrust concerns. That's part of why the plan needs U.S. government approval before moving forward. U.S. law forbids the same person from serving as a board director or officer at two companies that directly compete, and the Biden administration has amped up scrutiny of those so-called interlocking directorates. It is not known whether Open AI would contribute funds or have a formal relationship with the new startup, but antitrust enforcers at the Federal Trade Commission or Department of Justice may have concerns about Altman's involvement should the new company seek

to make chips exclusively for use by Open AI, end quote. A couple things on that. First, Altman has clarified to people that the $7 trillion investment figure we were bandying about for this chip project last week represents the total sum of investment. needed over the life of this chip project, including real estate over many years. So he's not asking for $7 trillion at like a $40 trillion valuation. But also, filings indicate he can make moves like this because Sam Altman is the sole owner of the OpenAI startup fund, which reported $175 million in total commitments back in May of 2023. Microsoft is an outside limited partner in the fund. Still a little odd, quoting Axios. Open AI's structural strangeness

permeates all aspects of the business. Open AI Startup Fund was launched in late 2021 to invest in other AI startups and projects. By last May, it reported $175 million in total commitments and a portfolio that included video editor Descript and Legal Tool Harvey. What set OpenAI startup fund apart, however, was that it wasn't and isn't owned by Open AI, nor even by its affiliated nonprofit foundation. Instead, it's legally owned by Altman. We wanted to get started quickly, and the easiest way to do that due to our structure was to put it in Sam's name. An open AI spokesperson tells Axios, we have always intended for this to be temporary. Well, temporary has been well over a year, and it's a significant risk. For example, what might have happened had Altman remained fired by Open AI? Could he have kept the

fund? Was there anything contractual to prevent it? No answer to that last question, but the company does add, we now know that we may need to re-examine our governance structure, which should proceed any changes to the fund, but our priority is to establish a new board first, end quote. So on threads, I posted sort of a joke, question mark theory for why Sam Altman is taking such big swings. It's because he has already achieved AGI and has it locked in a basement somewhere and is having it give him the playbook for taking over the world, to which Simon Owens replied, let's take it one step further and assume he's been killed and replaced by an AI that figured out how to generate a synthetic body that it can inhabit, or that the AI is using some sort of

blackmail to control him. Quick check in on the state of crypto. According to pitchbook, Q4 2023 saw only 12 crypto exits, the lowest number since Q4 of 2020, and consistent with the quote, low-level activity seen throughout 2023, quoting TechCrunch. The lack of crypto startup exit volume and value can be linked to a related decline in total venture investment into a start Web 3 companies. When liquidity is light, investment return prospects can darken. The good news for crypto founders is that despite slim chances at selling their company, venture capital investment ticked 2.5% higher in Q4-20203 compared to the third quarter, though deal volume fell a similar percentage. The fourth quarter was consistent with the low-level

activity seen throughout 2020. The report stated, and with only 12 exits during that time frame, it was the lowest number since Q4 2020. More deal value, despite limited exits, does imply a level of optimism among crypto investors that we might consider to be surprising. But with crypto prices rising, key regulatory hurdles cleared, and other positive signals, casting a bit of a warm light on Web 3 more generally, more investment doesn't shock us. The exit question, however, remains recent investment totals be damned. Looking at yearly data, crypto-focused venture capital generated exits worth $1.2 billion in 2012, just $500 million between 2019 and 2020. In 2022 and 2023, the numbers came to 1.5,000,

and $1 billion, the outlier was 2021 with $88 billion worth of crypto exit value, end quote. Yeah, that's because Coinbase went public in 2021. But let me reiterate something we just said there. Pitchbook also says VC investment in crypto-related companies rose 2.5% quarter over quarter in Q4, 2023 to $1.9 billion. The first time that crypto startup funding rose since Q1 of 2022, quoting CNBC. It's a welcome stat for crypto entrepreneurs who've been, bruised by the last couple of years of the so-called crypto winter, which made it much harder for founders to raise money. Venture funding for crypto firms slumped significantly in 2022 after a rise in interest rates from major central banks led to a flight of investors from riskier assets like

tech stocks and cryptocurrencies. Problems for crypto ventures were compounded that year by major collapses of crypto companies like Du Kwan's controversial algorithmic stablecoin Terra and Sam Bankman-Fried's FTX. Major venture funds like Andresen Horowitz, Sequoia Capital, and Tiger Global were bruised by the downturn in crypto deals. In some cases like the fall of FTX, funds had to write off their entire stake. It's no secret investors have been writing more checks. Lee said in ACNBC interview, now we're starting to see it in the data. Lee said that crypto venture funding has bottomed with a rise in crypto asset prices and public market valuations of crypto-related companies such as Coinbase, Marathon Digital, and Microstrategy, end quote.

Time for the week and long-rate suggestions. Just two again this week. First up, David Pierce has a handy explainer for a term you've heard us say a lot recently. What is the Fedaverse exactly? Read this explainer from the Verge, quote. So what is the Fediverse? It's an interconnected social platform ecosystem based on an open protocol called Activity Pub, which allows you to port your content data and follower graph between networks. What? I know. Let me try again. The Fedaverse is as if you took X, TikTok, Snapchat, Instagram, and Facebook and made them all interoperable. so you could post anything from anywhere, and all your followers would be guaranteed to see it. And if you wanted to leave one platform for another, you could bring all your content, all your followers, all your everything with you.

Oh, so it's a new social network. Why didn't you just say that? It's actually lots of new social networks. A bunch in the Fediverse already exists, and lots more are coming. But the interconnection is the thing. Instead of having all your Facebook stuff on Facebook and all your X stuff on X, the Fediverse allows them to interoperate. That's the really big shift here, end quote. And then let me turn you on to a player in the streaming wars you probably haven't heard of. Showmax is, well, not a combination between Showtime and Max. It's a South African streaming company that has accumulated 2.1 million subscribers in Africa. That puts it ahead of Netflix's 1.8 million in Africa and Amazon Prime video's 300K

subscriptions on the continent, quoting rest of world. Showmax, which was spun out of Africa's largest entertainment company, multi-choice in 2015, has seen its market share rise to nearly 39% compared to Netflix's dropping to 33.5%. Showmax has overtaken Netflix amid intense competition brewing in Africa's video streaming industry, where global firms, major telecoms, and several country-specific apps are fighting for consumers' bucks. In January, Prime Video announced it was downsizing its Africa and Middle East operations and halting the commissioning of original content. The company stood out among competitors due to its deep understanding of local preferences and a careful

curated mix of content suited to African interests. Showmax has partnered with HBO and Comcasts to bring high-quality international English language content to its viewers and developed strong relations with local creators in African countries. The platform also has streaming rights to the English Premier League, for which it offers a dedicated subscription plan. In 2020, Showmax acquired exclusive licensing rights to stream HBO's content in Africa, which has become one of the biggest advantages for the company, although we now watch a few local shows on Showmax. My favorite shows were billions and curb your enthusiasm. Johannesburg-based designer Bello Cindy told rest of world. In March 2023, Comcast, the parent company of Universal

Pictures, NBC Peacock, Sky, DreamWorks Animation, and Telemundo, bought a 30% stake in ShowMex through NBC Universal. This gave the African company access to the advanced technology, the funds to survive in a competitive market, and blockbuster English language content, including titles from the BBC Lionsgate, ITV, Paramount, Sony, and Warner Brothers. Comcast's presence will help further strengthen Showmax's position, CEO Mark Jury told rest of world, it allows us to invest in far greater numbers of local productions. And so when you take those local productions coupled with the international content from around the world and the best of live sport, we know that we've got a winner, end quote. Showmax users in Nigeria and South Africa told rest of the world,

they prefer the platform over others because of its vast mix of local and international content. Showmax feels like watching Nigerian TV without having a TV. Dice Amy Nasin, a Lagos-based human resources manager told rest of the world. I like the reality TV shows you find on Showmax, the Real Housewives shows and Big Brother Nijja. I also enjoyed the HBO series on the platform. Showmax's original series include multiple award-winning shows like Ties Joburg Diary and Donker Boss. Showmax's biggest advantage is parent company multi-choice's vast network. It runs a cable TV business and has been building TV channels across Africa since 1995. Marie Laura Mungai, founder of consulting firm Restless Global told rest of the world,

multi-choices revenue in the 2023 fiscal year was $59.1 billion, or $3.1 billion, most of which came from its direct broadcast satellite service, DSTV, end quote. I know it feels like I'm telling you about new language models every day. There are actually like four other new AI models I could have told you about this week, but I'm trying to balance real groundbreaking news with just new headlines. SORA seems like a bigger deal than most. because it's crossing the threshold of a new modality in a convincing way for the first time.

No live show on Monday, as it is a bank holiday here in the U.S., but I will post that interview Chris and I did with Blue Sky CEO Jay Graber on Monday. Enjoy that, and I'll talk to you again on Tuesday.