Big Tech: too big to f-AI-l? Why even VC-backed genAI "unicorns" can't compete
Recent developments at Inflection AI & Stability AI beg the question: does Big Tech already have a "lock" on the generative AI market opportunity?
Sure, it’s Eclipse day. But it’s also time for your weekly brAIn dump! As always, first, “the trAIler” - 10 key media-related AI headlines from last week (the “AI:10”). Then, “the mAIn event” (my feature story is about Big Tech eating VC-backed generative AI’s lunch). Next, “the cocktAIl” - my special AI event mixology. Finally, “the AI legal case tracker” - updates on the key generative AI copyright infringement cases. So LET’S GO!
I. the trAIler - 10 “quick hit” AI headlines from last week
(1) It’s “fair” to train on copyrighted works without consent because they’re “publicly available,” right? That’s what OpenAI’s cinematic quality AI video generator and Suno’s AI music generator want us to believe, as they and all Big Tech generative AI relentlessly scrapes the web to train their LLMs. But is it fair (both legally and ethically)? If it were, why are these companies so cagey about what they mean by “publicly available”? Read this Axios article for perspective. Also check out Ezra Klein’s latest “must listen” podcast on the subject (via this link). (Long-time readers know how I feel - using copyrighted material for training without consent or payment is an outright “taking.” Creators must be paid - and ultimately will be).
(2) Oh snap! French collab Common Corpus rebuffs OpenAI’s claims that it’s impossible to create tools like ChatGPT without training on copyrighted works. Sure there are limitations, but this initiative is a step in the right direction (one of many “ethically sourced AI” initiatives). Read more about this French collab’s approach here.
(3) By the way, love that phrase “ethically sourced AI” (some colleagues and I think we invented it last week). In any event, maybe it’s time to label AI generated content to foster content trust and safety in the same way we use nutrition labels to promote food safety. Read more about labeling efforts here via Forbes.
(4) Sure they can run, but they can’t hide! AI deepfakes are everywhere. But now a growing number of “forensic AI” companies are in the hunt to track them down and keep you safe. Read about the various approaches here via The Washington Post - and heed the author’s ultimate warning for this election: “Assume it’s fake.” Yikes!
(5) Who wants deepfakes (or any fakes), when you can have the real thing? Musicians en masse - over 200 of them (including icons ranging from Smokey Robinson to Billie Eilish) - signed an open letter to Big Tech pleading for them to “cease the use of AI to infringe upon and devalue the rights of human artists.” They certainly have a point. Read more about it all here via Axios.
(6) Guardrails, what guardrails? Hackers can force AI chatbots to break their own rules. Shocking, I know. Read more here.
(7) Wanna know how OpenAI’s cinematic video quality AI generator does its thing? Now you can get a peek behind the curtain to see how the sausage is made (well, kind of, because even Big Tech gen AI CEOs don’t know exactly how their LLMs generate their specific outputs). Read here via The Hollywood Reporter.
(8) OpenAI only needs a 15-second sample to clone your voice (or any voice, for that matter). So what prevents anyone from cloning anybody’s voice? Several startups are banking on just that, essentially using “the honor system” as their legal air cover for the coming mass infringement. Read more here from The Verge.
(9) Less “uncanny,” more “valley.” Read how the advertising industry is making AI generated images look less like AI via The Wall Street Journal. And check out Invesco QQQ ETF’s clever genAI-inspired marketing campaign via Bloomberg titled “The Future Isn’t Scary, Being Locked in Your Home by AI Is.”
(10) Apple’s AI “secret sauce” revealed? Experts weigh in about how they believe Tim Cook and his crew might use Google’s Gemini to power new applications on their upcoming iOS and iPhones. Read about it here via WIRED.
II. the mAIn event - does Big Tech already have a “lock” on the generative AI market opportunity? and if so, why? (inquiring VC minds want to know)
A funny thing happened on the way to several venture capital backed generative AI companies becoming $1 billion plus tech “unicorns.” Reality, that’s what. And reality can, in fact, bite.
Remember this chart my team and I created a couple months ago? The numbers haven’t changed that much in terms of investment. But the “substance” behind those numbers certainly has. Here’s what’s happened since.
Just in the last few weeks, there have been a number of significant developments that reflect a growing pullback in confidence of VC-backed generative AI companies (and their investors) about their ability to compete against Big Tech behemoths.
Exhibit A is once high-flying generative image unicorn Stability AI, a now ironically-named company that’s been anything but stable in the past several months. CEO Emad Mostaque resigned late March to focus instead on what is called “decentralized AI” due to ongoing resource-devouring copyright infringement litigation, among other unstable developments (read more here from The Verge).
Exhibit B is even more surprising. Unicorn/genAI darling Inflection AI was essentially absorbed into its largest benefactor, Microsoft. No, this wasn’t an outright acquisition in a literal sense. Microsoft is too smart for that; it knew the antitrust Feds likely would attempt to slow it down. Rather, always-savvy Microsoft effectively did the same thing by hiring all key employees without buying the company’s assets themselves (read about this unusual deal here via Reuters).
What’s going on here?
It’s really quite simple. Several leading VC-backed genAI companies - even those that have raised hundreds of millions of dollars (and even billions of dollars like Inflection AI has) - have concluded that they simply can’t compete with the massive resources necessary to make it on their own. Sure, OpenAI and Anthropic plow forward with their “independence.” But how independent are they really when Microsoft is by far OpenAI’s largest investor (with over $10 billion) and Amazon is Anthropic’s (with $4 billion). Oh yes, these “startups” are also wholly dependent upon their Big Tech parents’ massive compute and cloud infrastructure.
GenAI LLMs are resource intensive (devouring, really), and the compute power necessary for these indies to compete requires billions upon billions of dollars. Even if these VC-backed companies had the means to pay for that horsepower (which they don’t at nearly Big Tech’s scale), they need to wait in line behind the owners of that compute power - which is Big Tech! Case in point Microsoft, which just announced that it, together with favorite child OpenAI, plan to build a $100 billion data center (yes, I said $100 BILLION!).
But wait, there’s more to this “David getting squashed by Goliath” story. As Elon Musk just recently bemoaned (and he bemoans a lot of things these days), he’s never seen a battle for engineering talent like he sees now with generative AI (in his words, “It’s the craziest talent war I’ve ever seen"). Make no mistake, money talks. And these Big Tech giants are certainly willing to spend it to attract the best and brightest in this generative AI “spAIce race” (case in point: Inflection AI). Top talent goes where the money goes. Not always. But most of the time.
None of this means that genAI startup founders won’t ultimately receive their desired riches. Many of them certainly will. But it does mean something very different than what happened during the Internet revolution. The Internet spawned entirely new VC-backed companies that ultimately became the big winners. Google. Amazon. Facebook/Meta. These former VC-backed indies ARE Big Tech, and now they, together with Big Tech brethren Microsoft and Apple, have left little room for new players to join their Big Tech party.
Big Tech is simply too big to f-AI-l in these generativeAI wars. These giants are only getting bigger and more powerful as they either effectively “own,” swallow up or overwhelm VC-backed unicorns. And that increasingly concentrated power has profound implications on how genAI will play out both in the marketplace - and in our lives.
What do you think? Send me your feedback and thoughts at peter@creativemedia.biz.
III. the cocktAIl- your AI mix of “must attend” AI events
After all, it’s always happy hour somewhere!
(1) AI LA’s “A.I. on the Lot” is only one month away, so save the date! May 16th, 2024 is a “must attend” event. Check it out and sign up here. It’s where all of LA’s generative AI “movers and shakers” meet, learn, and collaborate. You gotta be there.
(2) Check out Digital Hollywood’s first generative AI-focused virtual summit, “The Digital Hollywood AI Summer Summit,” on four days in July (July 22nd - 25th). The sessions are outstanding. Learn more here via this link.
(3) My next virtual roundtable - brought to you by Digital Hollywood - is coming in the next couple weeks. Last time it was The Police’s Stewart Copeland talking about AI and music. This month’s features a great “show and tell” about how generative AI will transform video production - both for Hollywood and for Madison Avenue agencies.
check out Creative Media and our AI-focused services
IV. the AI legal case tracker - updates on key AI litigation
Rather than lay out the facts of each case - and the latest developments - in every newsletter, click on this “AI case tracker” tab on “the brAIn” website. You’ll get all the up-to-date information you need. These are the cases I track (there were many important developments this past week).
(1) The New York Times v. Microsoft & OpenAI
(2) Sarah Silverman, et al. v. Meta
(3) Sarah Silverman v. OpenAI
(4) Universal Music Group, et al. v. Anthropic
(5) Getty Images v. Stability AI and Midjourney
The 7 Big Tech companies are trying desperately to grab land before any of us peasants realize what's going on -- but despite huge spends and impressive demos they are still paying at least 50x more than they should be for the results achieved.
They wont be able to keep this up for long, and once they pause for a season to try and recoup ROI Hugging Face & the open source community at large is going to make all their gains meaningless. The one and only thing they have going for them is brand recognition. We'll see how long that holds.