Five for Friday: Issue #8
On Amazon's Nova models, China's reasoning models, OpenAI's "12 days of Shipmas", ElevenLabs' conversational agent platform, and Musk's injunction against OpenAI
Welcome back to Five for Friday.
Buckle up! This week's headlines paint a masterclass in tech industry power plays:
Price wars: Amazon bulldozes into the AI arena with aggressive pricing
Global competition: Chinese labs match Silicon Valley stride for stride
Marketing tactics: OpenAI wraps new releases in holiday paper
Product innovation: ElevenLabs’ conversational AI agent platform
Corporate governance: Musk and OpenAI's courtroom drama
If like me, you sometimes prefer listening over reading, why not try the podcast below which has been generated from this article using NotebookLM.
#1 Amazon Goes Supernova
Tech giant unveils full-stack AI offensive
Amazon turned more than a few heads this week with the unveiling of Nova, a family of AI models that's positioning itself as a serious challenger to established players like OpenAI, Anthropic, and Google.
The Nova lineup introduces six distinct models, ranging from the streamlined text-only Micro to the comprehensive Pro version. There's also Canvas for image generation and Reel for video creation, with the advanced Premier model set to join the family in 2025. What's catching everyone's attention? Amazon claims these models are 75% more cost-effective than current market offerings.
The Nova launch is part of a broader AI strategy that includes some impressive hardware developments. Amazon's new Trainium2 chips are showing 4x performance gains and they have slashed costs by 50% compared to market leader Nvidia. The company has also announced Project Rainier to build what could be the world's largest distributed AI compute cluster. Add in their $8 billion investment in Anthropic, and its clear Amazon is finally ready for its multipronged move into AI.
Perspectives:
This is classic Amazon: First they match capabilities, then they slash prices, finally they dominate through scale. We've seen this movie before in cloud computing and e-commerce. They're betting that by driving down costs, they'll capture market share and make up the difference in volume. However, while Amazon has deep pockets, maintaining this price advantage while funding ongoing AI development will prove very expensive.
While the $8 billion investment in Anthropic shows Amazon's commitment to working with the AI start-up, the Nova launch suggests they've learned from watching the fraying OpenAI-Microsoft relationship. Having their own models provides strategic independence and bargaining power.
#2 Made in China 2.0
Chinese labs challenge OpenAI's o1 with open source reasoning models
The East-West (China-US) AI race is heating up. Just months after OpenAI's o1 model, capable of chain-of-thought “reasoning” stormed onto the scene in September and shot straight to the top of AI leaderboard, Chinese AI labs are serving up some serious competition.
First out of the gate a few weeks ago, was DeepSeek’s R1 model. The model claims to outperform OpenAI's o1 on key benchmarks, such as AIME (advanced high school math problems), MATH (high school competition-level math, and Codeforces (competitive programming challenges).
Not to be outdone, Chinese academics unveiled LLaVA-o1 last week, a pioneering vision-language model (i.e., capable of interacting with text and images). As opposed to most vision-language models which respond without an organised ‘thought process’, LLaVA-o1 takes a structured approach to solving problems with a four-stage reasoning process.
The timing couldn't be more interesting. Models that feature reasoning features such as o1, DeepSeek-R1 and LLaVA-o1 will likely become more commonplace as the tech world debates whether AI progression could potentially be slowing due to limits to the “scaling laws”.
However, these achievements come with a distinctly Chinese flavor. Both DeepSeek-R1, and LLaVA-o1, will both politely decline to discuss certain sensitive topics faster than you can say "Great Firewall." It's a reminder that while the code might be universal, the context isn't.
Perspectives:
What makes this particularly intriguing is that both DeepSeek-R1 and LLaVA-o1 are open source. Just a year ago, the gap between closed and open source models seemed unbridgeable, with OpenAI, Anthropic, and Google in a league of their own. Yet the pace of open source advancement has been relentless, with each new release closing the capability gap. The question is now when, not if, they'll catch up.
All of Chinese models come with distinctly Chinese characteristics. While Western labs often prioritise capabilities and safety, Chinese models ship with additional built-in compliance to national values and regulations. This suggests we're heading toward a bifurcation in AI development, where models might share technical foundations but optimise for dramatically different priorities.
#3 OpenAI's Seasonal Gift Guide
OpenAI plans festive flood of product launches from Sora to Santa voice
OpenAI's Sam Altman is trading his usual black T-shirt for a Santa suit, announcing a "12 Days of Shipmas" — a daily parade of AI launches and demos that is Silicon’s Valley’s version of a tech advent calendar.
Starting December 5th, the AI giant plans to unwrap everything from "big ones" to "stocking stuffers". The headline gift appears to be Sora, their the long awaited text-to-video model.
The o1 reasoning model is also expected to graduate from its ‘preview’ phase. This will most likely mean that o1, previously limited to text-only chats, will now have access to the full swathe of ChatGPT capabilities such as voice, image generation, and advanced data analysis.
ChatGPT's Advanced Voice Mode might also get some upgrades. Code snippets spotted in the wild hint at everything from a "Live Camera" feature for real-time video analysis to a seasonal Santa voice that might have you saying "Ho Ho GPT."
Perspectives:
OpenAI's holiday marketing blitz isn't just seasonal cheer — it's likely another calculated step in their consumer brand building strategy. Having already made "ChatGPT" practically synonymous with AI in the public consciousness, moves like "12 Days of Shipmas" and acquiring the chat.com domain further cement their consumer-friendly image. But there's a fine line between clever marketing and gimmickry. Could this rapid-fire approach risks overwhelming users and diluting the impact of genuinely significant releases?
When it comes to video, OpenAI is playing catch-up. While its initial preview made waves, competitors such Runway, Kling, and Google have already realised several generations of video models. However, with most current models limited to generating clips of 10 seconds or less, there's still plenty of room for innovation. If rumors about Sora's longer sequence capabilities prove true, OpenAI might still find ways to differentiate itself in this nascent market
#4 AI Finds Its Voice
ElevenLabs debuts conversational agent platform and podcast generator
ElevenLabs is bringing a double dose of AI voice innovation to the table this week. The AI voice company has launched a new conversational AI platform that streamlines the creation of custom voice agents, targeting everything from customer service to interactive gaming characters.
The platform's standout feature is its natural conversation handling, including real-time interruption detection and turn-taking capabilities that make AI interactions feel more human. It integrates seamlessly with leading AI models like Gemini, Claude, and GPT, supports 31 languages, and offers competitive pricing starting at $0.10 per minute, dropping to $0.015 at scale.
For developers, the platform comes packed with practical tools including SDKs for Python, JavaScript, React, and Swift, plus WebSocket API access. Perhaps most notably, it features native Twilio integration, allowing businesses to easily deploy these AI agents directly into their existing phone systems.
This launch follows their recent release of GenFM, a clever addition to their ElevenReader app that transforms written content into AI-hosted podcasts. Google's NotebookLM went viral earlier this year with similar text-to-podcast capabilities, but it remains limited to English-only content. GenFM supports 32 languages, allowing users to convert their PDFs, articles, and ebooks into audio content with AI co-hosts — perfect for consuming content while multitasking.
Perspectives:
ElevenLabs, despite their technical prowess in voice, is playing in a field dominated by giants like OpenAI, Anthropic, and Google, who have significantly deeper pockets and larger teams. Their simultaneous push into both enterprise and consumer markets, while ambitious, raises questions about their ability to maintain competitive development pace and support quality across both segments.
The company’s decision to absorb language model costs during beta points to a classic market-building strategy in an emerging space. While text-based AI agents are now commonplace, voice-based interactions are still in their infancy. By subsidising costs during this crucial adoption phase, they're effectively paying to accelerate market development — a bold move that could pay off if they establish themselves as the go-to platform before larger players fully enter the space.
#5 Profit & Prejudice: Act IV
Musk challenges OpenAI's shift from non-profit roots
Silicon Valley's favorite soap opera just got a new plot twist, and this time it's not about who's getting voted off the OpenAI island. Elon Musk, apparently not busy enough with his collection of companies, has decided to add "legal drama instigator" to his resume by filing for an injunction against his former AI besties.
The Tesla titan is trying to throw a wrench into OpenAI's works, specifically targeting their potential shift to a fully for-profit model. His legal filing reads like a tech world restraining order, asking the court to stop OpenAI from doing everything from sharing board secrets with Microsoft to allegedly playing investment matchmaker (or rather, anti-matchmaker) by discouraging investors from funding competitors like his own xAI.
At the heart of this corporate custody battle is OpenAI's gradual transformation from a non-profit to what Musk's lawyers dramatically describe as a "Frankenstein, stitched together from whichever corporate forms serve the pecuniary interests of Microsoft." OpenAI counter that Musk's claims are about as solid as a cloud server, pointing to old emails showing he was actually all for Team Profit back in the day.
Meanwhile, the industry watches from the sidelines of this billion-dollar battle royale, perhaps wondering if they should grab popcorn or their lawyers' phone numbers.
Perspectives:
To a cynical observer (i.e., yours truly), the timing of Musk's injunction feels more strategic than principled. With xAI seeking $6 billion in funding and openly competing with OpenAI, its hard not to see this legal move as a calculated attempt to kneecap a competitor while simultaneously portraying xAI as the righteous defender of open-source AI. It’s also difficult not to point out the irony in Musk, a champion of free-market capitalism, suddenly playing the role of non-profit crusader.
The "Frankenstein" metaphor in Musk's legal filing, while dramatic, unintentionally highlights a key truth: OpenAI's hybrid structure has become an unwieldy experiment in corporate governance. Altman’s ouster and return as CEO was a major fiasco and its clear that regardless of the lawsuit's outcome, the current governance setup is likely unsustainable. It's a stark reminder that in the tech world, where founder-kings often reign supreme over pliant boards, getting governance right isn't just a legal nicety — it's crucial for responsible AI development.
Justin Tan is passionate about supporting organisations and teams to navigate disruptive change and towards sustainable and robust growth. He founded Evolutio Consulting in 2021 to help senior leaders to upskill and accelerate adoption of AI within their organisation through AI literacy and proficiency training, and also works with his clients to design and build bespoke AI solutions that drive growth and productivity for their businesses. If you're pondering how to harness these technologies in your business, or simply fancy a chat about the latest developments in AI, why not reach out?