Meta’s open-source era may be over
December 11, 2025
Welcome back. OpenAI may be getting nervous about its own models. The company warned in a report on Wednesday that the cyber capabilities of its frontier models are getting stronger. OpenAI is planning as if each of its new models could reach “high” levels of cybersecurity capability, as measured by its risk framework (one step below the “critical” threshold, where a model is so risky that it's not released to the public). To prepare for the problem they’re creating, OpenAI is establishing a Frontier Risk Council, an advisory group dedicated to assessing these risks. Still, as leading-edge models get better at finding the cracks in security systems, bad actors will inevitably find a way to leverage them.
1. Meta’s open-source era may be over
2. Tech firms target AI’s power problem
3. Amazon, Microsoft make big commitments in India
BIG TECH
Meta’s open-source era may be over
After spending billions and enduring significant staff turnover to revamp its AI unit, Meta is weighing a move to make its new model, nicknamed Avocado, closed source, multiple outlets reported.
Since open-source Llama 4 received mixed reviews following its April release, Meta CEO Mark Zuckerberg has embarked on an ongoing crusade to reverse the company’s AI fortunes. Meta aqui-hired Scale CEO, Alexandr Wang, and co. went on an AI hiring spree to build, in Zuckerberg’s telling, “the most elite and talent-dense team in the industry.” Wang’s team, one of a few AI teams within Meta, is known internally as TBD Lab. While pay packages reportedly stretching up to nine figures have attracted talent, the group has reportedly not cohered.
Meta’s AI restructuring led some researchers to depart for other AI labs. Meta’s chief AI scientist, Yann LeCun, left in November to found his own startup. CNBC has reported that LeCun was rankled by the 600 layoffs that Meta Superintelligence Labs (MSL) underwent in October. Just this week, MSL employees Sang Michael Xie and Vitaliy Chiley decamped from the company.
Those still at Meta have reportedly clashed on direction. Senior executives said some of Meta’s AI efforts should be oriented toward improving the company’s social media and advertising businesses, while Wang argued Meta should catch up with frontier models offered by OpenAI and Google before focusing on products, per The New York Times. The Times also reported that budget cuts to Meta’s metaverse team were rerouted to Wang’s unit.
TBD Lab’s new frontier model, being closed source, would represent a major strategic shift for Meta, which has long touted its open-source AI efforts. The risks of open-source development became clear when DeepSeek’s R1 model incorporated components of Llama’s architecture, which angered some at Meta, CNBC reported.
Open-source cuts both ways, though. TBD Lab is using Alibaba’s Qwen model as part of the training process for Avocado, Bloomberg reported.
Meta has fallen behind other open-source AI models, so it’s pivoting to the even more competitive field of closed source AI models. Keeping Llama open-source could have created a wedge for Meta to differentiate itself from other hyperscalers, but as things currently stand, it’ll be up to Wang to lead Meta’s proprietary model past those offered by OpenAI, Google, and Anthropic. In the words of Wang’s unit within Meta, the odds of that happening are very much TBD.
TOGETHER WITH ROCKET
Build an App from Your Pocket
Meet Rocket Mobile—the full power of Rocket.new, now on iOS.
Whether you're commuting, between meetings, or in line for coffee, you can create, edit, and deploy full-stack apps from your phone. From fixing a bug to shipping a new feature in seconds, Rocket Mobile turns idle moments into progress.
Built with Face ID security, one-tap actions, and a mobile-first UI, it's redefining what’s possible on the go. No laptop required.
HARDWARE
Tech firms target AI’s power problem
AI is doubtlessly causing an energy crisis. Investors are placing their bets on how to solve it.
On Wednesday, Fervo Energy, an enhanced geothermal energy startup, raised $462 million, bringing its valuation to $1.5 billion. Fervo’s backers include a slew of major investors, climate tech firms and Google, which has a previous partnership with the company to supply its data centers with electricity.
The funding will help Fervo continue its efforts to construct a 500-megawatt power plant in Utah, as well as break ground on other sites.
“Folks realize that there’s an urgency to adding new power generation to the grid, and that the AI race is going on right now,” Fervo CEO Tim Latimer told the Wall Street Journal.
And Fervo isn’t the only one raking in cash to alleviate the energy problem. Boom Supersonic, a company that builds clean natural gas turbines, announced $300 million in funding on Tuesday.
The company said in its press release that Crusoe, an AI infrastructure firm, will be the launch partner for its 42 megawatt turbine to deliver “reliable energy to AI data centers.” The firm also revealed that it has a backlog of more than $1.25 billion for its turbines.
“Supersonic technology is an accelerant—of course for faster flight, but now for artificial intelligence as well,” CEO and founder Blake Scholl said in a press release.
With AI data centers facing a projected power shortfall of up to 13 gigawatts by 2028, tech firms and investors alike are scrambling for solutions.
These financing rounds are among the many examples of companies funding ambitious ideas to address the problem before it goes from bad to worse. For instance, bets on fusion and nuclear energy have ramped up, and Meta has even jumped into energy trading as a means of shoring up US power plants.
And some tech giants are even targeting literal moonshots: Google, Jeff Bezos’ Blue Origin and Elon Musk’s SpaceX are all exploring ways to deploy data centers in space, harnessing the sun’s energy directly without burdening the already overburdened power grid.
Impending crises are breeding grounds for invention. With so much money on the line, companies have no choice but to make their AI bets work (or else potentially face devastating economic fallout). Rather than slow down the breakneck pace of their growth, they’re seeking the energy they need by any means necessary. However, whether or not the speed of climate innovation can keep pace with AI’s hypergrowth is still unclear.
TOGETHER WITH FUNNEL
Prove your impact. Optimize campaigns with Conversion APIs
The privacy era has broken the flow of truth between your ads and the platforms measuring them, making attribution harder and budgets less effective.
Conversion APIs (CAPI) offer a secure, reliable way to reconnect your conversion data to major ad platforms, keeping insights accurate, compliant, and complete.
In this recorded webinar, you’ll learn why first-party data is now essential and how Funnel’s Data Activation helps you unlock better campaign performance.
What you’ll learn:
✅ Why today’s data landscape doesn’t work without first-party data
✅ How Funnel’s CAPI enables seamless first-party data integration
✅ How to work with CAPI and LinkedIn for stronger B2B optimization
✅ How agency Trickle implemented CAPI to improve results
MARKETS
Amazon, Microsoft make big commitments in India
AI-linked firms are racing to gain a foothold in India.
Amazon announced plans to invest $35 billion in India by 2030, with a large portion of the push being driven by AI-enabled selling and shopping. The news came shortly after Microsoft said it would invest $17.5 billion to advance India’s cloud and AI infrastructure.
India is the world’s most populous nation, contains a relatively skilled and low-cost labor pool, and has not seen much success from home-grown AI developers. These factors may explain why it is such an attractive investment target for Big Tech and AI firms. In any case, lots of dry powder is being deployed.
In October, Google announced it would invest $15 billion in India over the next five years to construct an AI hub in the southern city of Visakhapatnam.
Intel will explore making chips in India after striking a partnership with the Indian corporation Tata.
OpenAI and Google have both offered discounted subscriptions for Indian users accessing ChatGPT and Gemini.
Perplexity offered free subscription trials to customers of the Indian telecom giant Bharti Airtel.
Anthropic announced plans to open an office in Bengaluru.
There’s a lot to like about India from the perspective of AI and Big Tech. For one, the nation could offer fresh compute — Microsoft, OpenAI, and Amazon are all building or have built high-powered data centers in India. There’s also the local talent pool, which multiple firms are going after through local offices and outreach initiatives. The consumer market is also huge, which OpenAI and Perplexity have seen success in tapping. The only threat to the bonhomie may be the Indian government, which just proposed charging OpenAI and Google for training models on copyrighted content.
LINKS
Microsoft partners with UW, Providence for AI cancer insights
DeepSeek reportedly training models on banned Nvidia chips
Google deploys MCP servers to allow agents to connect to cloud services
McDonald's takes down AI slop holiday ad following backlash
Oracle posts weak cloud sales as AI growth takes longer than expected
Google partners with international publications for AI-powered audience engagement
PubMatics, Kontext strike a deal to inject ads into chatbots
ChatGPT and Adobe: Adobe classics like Photoshop, Express and Acrobat are now available in OpenAI’s flagship chatbot.
SimGym: A system that creates “digital customers” to test your site, complete tasks and run A/B tests.
Basedash Autopilot: Finds answers buried in your company’s data stack to surface insights you may be missing.
Vybe: a vibe-coding platform for creating secure, production-ready internal apps.
A POLL BEFORE YOU GO
What do you think is the best bet for powering AI? |
The Deep View is written by Nat Rubio-Licht, Jack Kubinec, Jason Hiner, Faris Kojok and The Deep View crew. Please reply with any feedback.
Thanks for reading today’s edition of The Deep View! We’ll see you in the next one.
“The tip of the wing of the plane of [the other image] looks too dark. It looks like it has a shadow on it but there should be no shadow since the sun is shining directly on it. ” “Didn’t see any duct tape on the fake wing so knew it was fake 😂” “[The other image’s] wing didn’t have any safety stickers i.e. ‘No Step’” |
“What kind of magic camera takes a photo that clear and colorful from out a plane window?” “I was sure the scratches on the window made it real!” “I guess the oddly placed engine is real after all. It doesn’t look tucked under the wing where it should be.” |
Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning.
If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.