Tech
Mark Zuckerberg, Alexandr Wang, and Shengjia Zhao
Mark Zuckerberg, Alexandr Wang, and Shengjia Zhao (@zuck/Threads)

Reports: Meta training its new AI using rival models; switching to closed models in quest for profits

A pair of reports from The New York Times and Bloomberg detail the ongoing struggles between Meta’s upstart AI division and the rest of the company as it seeks to monetize its massive investments in AI.

Jon Keegan, Rani Molla

A pair of new reports about internal struggles at Meta add new information to how Mark Zuckerberg’s hard pivot to AI is going.

The New York Times details some of the friction between the company’s old guard and Alexandr Wang, the 28-year-old upstart who now leads Meta’s AI division.

One detail: Meta asked the company’s longtime CTO, Andrew Bosworth — considered to be one of the Meta’s top executives — to cut $2 billion from the budget of the division he leads, Reality Labs. The segment is responsible for the company’s AR glasses and the metaverse, the feature that the company changed its name in homage to in 2021. The budget cut from Bosworth’s division will go to the AI division, whose leader joined the company in June, though Meta said next year’s budget isn’t final.

A report last week saying the company is planning 30% budget cuts for the money-losing Reality Labs caused Meta’s stock to surge higher.

Another detail from the Times’ reporting is that according to sources, Bosworth and Chris Cox, the company’s chief product officer, wanted Wang’s team to concentrate on using Instagram and Facebook data to help train Meta’s new foundational AI model — known as a “frontier” model — to improve the company’s social media feeds and advertising business.

But Wang, who is developing the model, pushed back. He argued that the goal should be to catch up to rival AI models from OpenAI and Google before focusing on products, the sources said.

Closed is the new open

Separately, a Bloomberg report out today explains Meta’s effort to build not just a “superintelligent” AI model, but one that is also super profitable. Per the report, Zuckerberg “spends much of his time and energy” working day to day with his new team of AI all-stars, known as “TBD Lab.”

The report also has details of how Meta is building its next model, code-named Avocado. The TBD team is reportedly using third-party models to help train Avocado, including those of its rivals Google and OpenAI. The team is “distilling” from Google’s Gemma, OpenAI’s open-weight model gpt-oss, and the Qwen model from Alibaba, per the report. Use of a Chinese model like Qwen for training could complicate Meta’s efforts to sell its AI for use in national security applications.

A major shift away from open-source models toward proprietary closed ones also seems to be part of Meta’s new strategy. This is a notable departure from Zuckerberg’s passionate, repeated praise of open-source AI, as the Meta chief has recently signaled that the company will be using more closed models. A proprietary model would make it easier to charge for Meta’s AI services compared to its previous strategy of giving away its Llama models for free.

More Tech

See all Tech
tech
Jon Keegan

DeepSeek releases new V4 series models highlighting efficiency and long context

Chinese AI lab DeepSeek has released a major new version of its eponymous open-source AI models that are nipping at the heels of leading frontier models in some areas.

The most significant DeepSeek-V4 Pro and DeepSeek-V4 Flash both have a 1 million-token context — the amount of information the model can actively work with in a single session — which is a crucial feature for complex, long-running coding tasks.

DeepSeek rebuilt how the models process information under the hood, making them substantially more efficient — and that efficiency is what makes the large context window actually usable.

Also, the new models’ coding skills have closed the gap with the major frontier models from Anthropic, OpenAI, and Google.

The authors of the model acknowledge some of V4’s shortcomings, such as its lower scores on reasoning benchmarks, saying that V4 “trails state-of-the-art frontier models by approximately 3 to 6 months.”

As open-weight models, V4 can be run on any user’s own hardware, making the V4 models among the top-performing open-source models out there. V4’s large context and token efficiency are especially significant among open-source models.

But like with earlier DeepSeek models, don’t ask it about Tiananmen Square.

DeepSeek Unveils Flagship AI Model a Year After Breakthrough

DeepSeek Unveils Flagship AI Model a Year After Breakthrough

DeepSeek rebuilt how the models process information under the hood, making them substantially more efficient — and that efficiency is what makes the large context window actually usable.

Also, the new models’ coding skills have closed the gap with the major frontier models from Anthropic, OpenAI, and Google.

The authors of the model acknowledge some of V4’s shortcomings, such as its lower scores on reasoning benchmarks, saying that V4 “trails state-of-the-art frontier models by approximately 3 to 6 months.”

As open-weight models, V4 can be run on any user’s own hardware, making the V4 models among the top-performing open-source models out there. V4’s large context and token efficiency are especially significant among open-source models.

But like with earlier DeepSeek models, don’t ask it about Tiananmen Square.

$28.5T
Rani Molla

SpaceX thinks its total addressable market (TAM) is a whopping $28.5 trillion for its businesses, according to an S-1 filing for its upcoming IPO reviewed by Reuters. And most of that market isn’t rockets. The company says roughly 90% could come from AI — largely selling artificial intelligence tools to businesses.

“We believe that our enterprise strategy, which is focused on serving the digital needs of the world’s largest industries with Al solutions, positions us competitively to pursue this rapidly ⁠growing opportunity,” ​SpaceX said in the filing. “We believe we have identified the largest actionable total addressable market in human ​history.”

TAM, of course, assumes capturing every possible customer. But even a small slice of a $28.5 trillion market would be enormous.

tech
Rani Molla

Tesla Cybercab production has begun

On Tesla’s earnings call earlier this week, CEO Elon Musk said production of the company’s steering-wheel-less Cybercab had begun. Since then, Musk and Tesla have posted videos showing the gold two-seater rolling off the line at its Texas Gigafactory and onto the road.

The Cybercab — meant both for consumers and Tesla’s Robotaxi network — is widely seen as central to the company’s future. “The future of the company is fundamentally based on large-scale autonomous cars and large scale and large volume, vast numbers of autonomous humanoid robots,” Musk said last year.

Whether these cars actually make it to consumers is another question. For now, regulations generally require steering wheels, and Tesla still has to prove the vehicles can reliably drive themselves.

On the earnings call, Musk said production would be “very slow” but would ramp up and go “kind of exponential towards the end of the year and certainly next year.”

tech
Rani Molla

Meta signs deal to use Amazon Graviton chips

Meta said it will deploy “tens of millions” of Amazon Web Services Graviton CPU cores to power so-called “agentic” AI systems — tools that can reason, plan, and act on their own. The move makes Meta one of the largest customers of Amazon’s in-house chips.

The deal also underscores a broader shift in AI infrastructure, as companies move beyond Nvidia GPUs and use different chips for different tasks.

Meta, which is working on its own custom inference chips, also has chip deals with Advanced Micro Devices and Nvidia.

Meta Partners With AWS on Graviton Chips to Power Agentic AI

Meta Partners With AWS on Graviton Chips to Power Agentic AI

The deal also underscores a broader shift in AI infrastructure, as companies move beyond Nvidia GPUs and use different chips for different tasks.

Meta, which is working on its own custom inference chips, also has chip deals with Advanced Micro Devices and Nvidia.

tech
Rani Molla

Oracle rises after Wedbush’s Dan Ives calls the stock a buy with 25% upside

Oracle extended its premarket gains Friday after Wedbush Securities’ Dan Ives initiated coverage with an “outperform” rating and a $225 price target — about 25% upside to its pre-initiation level — calling the enterprise software and cloud infrastructure company a “foundational infrastructure provider for the AI revolution.”

Ives argues investors are misreading Oracle’s heavy capital spending and negative free cash flow as risky, despite being backed by a massive $553 billion backlog of contracted demand. He says the company’s “secret sauce” is a two-part strategy: building high-performance cloud infrastructure for AI workloads while connecting those models directly to companies’ own data.

“We believe Oracle is in the early innings of a significant repositioning as it executes on this generational opportunity,” Ives wrote.

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, Robinhood Derivatives, LLC, or Robinhood Money, LLC. Futures and event contracts are offered through Robinhood Derivatives, LLC.
©2026 Sherwood Media, LLC