Even after three years since the onset of the generative AI explosion, the majority of AI startups still find profitability by catering to enterprises rather than individual users.

While general-purpose LLMs like ChatGPT were swiftly embraced by consumers, many targeted consumer GenAI offerings have not yet made a significant impact.

“Many of the initial AI applications related to video, audio, and photography were extremely impressive,” remarked Chi-Hua Chien, co-founder and managing partner at Goodwater Capital, during TechCrunch’s StrictlyVC event in early December. “But then Sora and Nano Banana emerged, and the Chinese open sourced their video technologies. Consequently, many of those opportunities vanished.”

Chien likened some of these applications to the basic flashlight, which became a popular third-party addition right after the iPhone’s release in 2008, but was subsequently integrated into iOS.

He maintained that, similarly to how it took a few years for the smartphone ecosystem to mature before revolutionary consumer applications surfaced, AI platforms require a comparable phase of “stabilization” for enduring AI consumer products to thrive.

“I believe we are on the brink of what could be compared to the mobile landscape of the 2009-2010 period,” Chien stated. That era marked the inception of enormous mobile-first consumer enterprises like Uber and Airbnb.

Chien suggested we might be observing early signs of that stabilization as Google’s Gemini achieves technological equivalence with ChatGPT.

Elizabeth Weil, founder and partner at Scribble Ventures, reflected Chien’s perspective regarding the formative phase of GenAI, portraying the current landscape of consumer AI solutions as existing in an “awkward teenage middle ground.”

What is required for consumer AI startups to mature? Perhaps a novel device that goes beyond the smartphone.

“It seems improbable that a gadget which you interact with 500 times a day yet only registers 3% to 5% of your surroundings will ultimately facilitate the full realization of AI’s potential,” Chien commented.

Weil concurred that a smartphone may be too restrictive for re-envisioning consumer AI products mainly because it lacks an ambient nature. “I don’t anticipate we’ll be developing for this in five years,” she stated, gesturing to her iPhone as she displayed it to the audience.

Startups and established tech firms have been competing to create a new personal device that could replace smartphones.

OpenAI and Apple’s previous design head, Jony Ive, are reportedly developing a “screenless,” pocket-sized device. Meta’s Ray-Ban smart glasses are operated by a wristband that senses subtle movements. Meanwhile, numerous startups are attempting—often with lackluster outcomes—to introduce a pin, pendant, or ring applying AI differently than smartphones do. 

Nevertheless, not every AI consumer offering will hinge on the introduction of a new device. Chien proposed that one possibility could be a personal AI financial adviser tailored to individual needs. Simultaneously, Weil predicts that a personalized, “always-on” tutor will become prevalent, delivering its specialized instruction straight from a smartphone.

Although they are excited by AI’s capabilities, Weil and Chien voiced doubts about the arrival of several still-under-the-radar AI-driven social networking startups. Chien described these companies as building platforms where numerous AI bots interact with the user’s content.

“It turns social media into a single-player experience. I’m not convinced it works,” he stated. “The appeal of social networking lies in the awareness that genuine humans are present on the other side.”

Chai Discovery, a biotechnology startup supported by OpenAI, revealed a $130 million Series B funding round at a valuation of $1.3 billion on Monday.

The funding was spearheaded by General Catalyst and Oak HC/FT, according to the company. Additional investors include Menlo Ventures, OpenAI, Dimension, Thrive Capital, Neo, Yosemite venture fund, Lachy Groom, SV Angel, along with new backers Glade Brook and Emerson Collective. The total funding for the firm now exceeds $225 million.

The company is part of an expanding sector that views AI as a quicker pathway to drug development. In August, Menlo Ventures disclosed it was leading Chai’s $70 million Series A round. The investor characterized Chai as a startup creating foundational models specifically designed for drug discovery, aiming to predict interactions between biochemical compounds for potential reprogramming into cures.

Chai aims to “develop the ‘computer-aided design suite’ for molecules.” Last year, the startup introduced the Chai 1 AI model and has now launched Chai 2, its newest model. The company states that Chai 2 is showing marked enhancements in success rates for de novo antibody design, which refers to the creation of custom antibodies from the ground up, rather than altering existing ones.

“Our newest models are capable of designing molecules possessing desirable characteristics typical of actual drugs and addressing complex targets that have previously been unattainable,” Josh Meier, co-founder and CEO of Chai, stated in a prepared remark.

Previously, Meier, who specializes in machine learning, was involved in research and engineering at Facebook and, before that, worked for OpenAI, as per his LinkedIn profile. Chai Discovery was established in 2024, the profile indicates.

The podcast recording service Riverside has released its own version of a year-end summary similar to Spotify’s “Wrapped.” This recap, named “Rewind,” generates three personalized videos for podcasters.

Rather than presenting metrics such as how many minutes of content you recorded or the total episodes produced, Riverside crafted a 15-second montage of laughter, featuring rapid clips of my podcast co-host and I amusing each other. The subsequent video is in a similar vein, this time showcasing a compilation of us repeatedly saying “umm.”

Next, Riverside analyzes its AI-created transcripts of your audio to identify the one word you uttered the most frequently (we presume they filter out common words like “and” or “the”).

Ironically, during my podcast focused on internet culture, my co-host and I mentioned “book” more than any other word (this was likely influenced by our exclusive “book club” segments… or because my co-host has a forthcoming book that we promote relentlessly).

Another show on our network, Spirits, noted that they said “Amanda” most frequently (not due to an obsession with me, but because they also have a host named Amanda).

Within the podcast network’s Slack channel, we shared our Rewind videos. There’s something inherently amusing about a video featuring people repeating “umm.” Yet, we also understand what these videos signify: our creative resources are becoming increasingly filled with AI functionalities, many of which are unnecessary. Riverside’s Rewind highlights the futility of such tools — why would I require a video of my co-host and I repeatedly saying “book”? It’s good for a quick chuckle, yet lacks depth.

While I appreciated Riverside’s AI recap, its introduction occurs during a time when my industry colleagues are losing the chance to create, edit, and produce new podcasts, due to the very AI tools that crafted our Rewind videos. Although AI can automate tasks such as eliminating “umms” and dead air, podcasting itself is not purely mechanical.

AI can swiftly produce a transcript of my podcast, which is crucial for accessibility, helping automate what was once a very time-consuming and monotonous task. However, AI lacks the ability to make editorial choices regarding the arrangement of audio or video to effectively tell a narrative. Unlike the human editors I collaborate with, AI is unable to discern when an off-topic discussion in a podcast is entertaining and when it should be removed due to lack of interest.

In spite of the emergence of individualized AI audio tools, such as Google’s NotebookLM, their performance as creative instruments has recently encountered notable setbacks.

Last week, The Washington Post began to implement customized, AI-generated podcasts focused on the day’s news.

One can understand why this might appear to be a “good” strategy for profit-driven executives — instead of compensating a team to undertake the detailed tasks of researching, recording, editing, and distributing a daily show, why not automate the process? However, that is not feasible.

The resulting podcasts produced fabricated quotes and inaccuracies, which poses a significant risk for a news organization. According to Semafor, the Post’s internal evaluations revealed that between 68% and 84% of the AI podcasts did not meet the publication’s standards. This indicates a profound misunderstanding of how LLMs function. Training an LLM to differentiate reality from fiction is impossible since it is designed to deliver the statistically most probable response to a prompt, which is not always the most truthful – particularly in urgent news situations.

Riverside successfully created a delightful year-end product, yet it serves as a reminder. AI is permeating every sector, including podcasting. But amidst this period of the “AI boom,” as businesses experiment with new technologies, we must learn to differentiate when AI is beneficial and when it merely produces trivial excess.

Disney’s three-year agreement with OpenAI is exclusively focused, Disney CEO Bob Iger informed CNBC. The partnership, finalized last week, will integrate Disney’s beloved characters into the AI company’s Sora video generator. After the initial year of exclusivity concludes, Disney is permitted to form similar partnerships with other AI entities.

This arrangement provides OpenAI with a prominent content collaborator, enabling users to access over 200 characters from Disney, Marvel, Pixar, and Star Wars to generate content on Sora. Currently, it stands as the sole AI platform legally authorized to do this.

For Disney, this partnership serves as a means to explore the potential of generative AI and its intellectual property, allowing the company to evaluate the success of its collaboration with OpenAI before seeking further partnerships.

“No generation of humanity has ever obstructed technological growth, and we have no desire to do so,” Iger stated to CNBC. “We have always believed that if change is inevitable, including disruptions to our existing business frameworks, we ought to embrace it.”

Notably, on the same day that Disney revealed its agreement with OpenAI, the company dispatched a cease-and-desist notice to Google, accusing the tech giant of copyright violations. Google refrained from confirming or denying Disney’s claims but indicated that it will “engage” with the company.

Nvidia persists in broadening its influence in open source AI through two primary avenues: a new acquisition and a model rollout.

The semiconductor powerhouse revealed on Monday that it has acquired SchedMD, the top creator of the widely used open source workload management system Slurm. Nvidia stated that the firm will continue to manage the program, which is tailored for high-performance computing and AI, as open source, vendor-neutral software.

Slurm was first introduced in 2002, with SchedMD founded in 2010 by lead Slurm developers Morris Jette and Danny Auble. Auble currently serves as the CEO of SchedMD.

The specifics of the agreement were not made public. Nvidia chose not to further elaborate on the announcement beyond its blog entry.

Nvidia has collaborated with SchedMD for over a decade and mentioned in its blog that the technology is essential infrastructure for generative AI. The company plans to continue its investments in the technology and “accelerate” its integration with various systems.

Additionally, the semiconductor firm unveiled a new series of open AI models on Monday. The company asserted that this collection of models, named Nvidia Nemotron 3, is the most “efficient family of open models” for creating accurate AI agents.

This series includes the Nemotron 3 Nano, a compact model for specific tasks, the Nemotron 3 Super, designed for multi-AI agent applications, and the Nemotron 3 Ultra, which is aimed at more complex tasks.

“Open innovation serves as the cornerstone of AI advancement,” Jensen Huang, founder and CEO of Nvidia, stated in the company’s press release. “With Nemotron, we’re converting advanced AI into an open platform that provides developers with the transparency and efficiency required to create agentic systems on a large scale.”

Recently, Nvidia has made strides to enhance its open source and open AI initiatives.

Last week, the company introduced a new open reasoning vision language model, Alpamayo-R1, aimed at autonomous driving research. The company also announced that it had included additional workflows and guides related to its Cosmos world models, which are open source under a permissive license, to assist developers in better utilizing the models for physical AI development.

This surge in activity reflects Nvidia’s belief that physical AI represents the next frontier for its GPUs. Nvidia aims to be the preferred supplier for numerous robotics and self-driving vehicle companies seeking the AI and software to develop the underlying intelligence for the technology.

As Ford moves away from producing large electric vehicles, the company is introducing a fresh product line to utilize its batteries.

On Monday, Ford announced that rather than abandoning plans to manufacture batteries for those vehicles, it will redirect that production capacity towards a new battery storage venture. These storage solutions, which will employ more affordable lithium iron phosphate batteries, will serve to power data centers and alleviate demand on the electric grid. 

Ford anticipates that the battery storage systems will commence shipping in 2027 and aims to establish 20GWh of annual output. 

The company plans to invest around $2 billion into this new initiative over the next two years. According to the strategy, Ford will modify the existing manufacturing capability at its Kentucky facility. The automaker intends to produce LFP batteries using technology licensed from China’s CATL, along with battery energy storage system modules and 20-foot DC container setups at this site.

Ford will align itself with several automakers already active or planning to enter the battery storage market. Tesla has spent the previous decade marketing battery storage solutions and deploys about 10GWh each quarter. General Motors also provides a range of home and commercial battery storage options.

Lisa Drake, Ford’s vice president of technology platform programs and EV systems, stated that the “predominant” market opportunity for this new venture will be commercial grid customers. However, data centers will be a secondary focus, with plans to eventually offer home storage products, according to Drake.

“It became apparent when we surveyed the market that most of these customers preferred an LFP prismatic type of container system,” Drake remarked during a press call. “And considering that we already possessed a license to manufacture that technology in the U.S., along with our extensive experience in large-scale manufacturing over a century, it was a natural fit for us.”

Ford’s BlueOval Battery Park Michigan in Marshall, scheduled to start LFP battery production in 2026, remains on schedule, according to the company. These LFP batteries, which will also utilize CATL technology, will be integrated into Ford’s forthcoming mid-sized electric truck. There will be one modification at the Michigan facility, as Ford indicated it will also manufacture smaller Amp-hour cells for “residential energy storage solutions,” hinting that Ford’s strategy will extend beyond just commercial clients.

The BlueOval Battery Park Michigan has undergone various transformations in its short lifespan. In February 2023, Ford announced a $3.5 billion investment to construct the factory intended for LFP batteries to support its expanding electric vehicle lineup. However, construction was abruptly paused in September 2023. Two months later, a revised plan was revealed, targeting a production capacity of 20 gigawatt-hours, roughly 43% less than initially proposed.

Ford is halting the manufacturing of the fully-electric F-150 Lightning as part of a wider corporate reorganization of its electric vehicle strategy, the company disclosed Monday. Instead, Ford will offer an “extended range electric vehicle” version of the truck, which includes a gas generator that can recharge the battery pack to enable the motors to run for over 700 miles. 

The company did not specify when the updated F-150 Lightning will be available for purchase, or its pricing details. 

This strategic shift will come with a significant financial impact for Ford. The firm expects to incur a $19.5 billion loss as it reconfigures its EV business approach. Most of these costs, including an $8.5 billion write-down of EV assets, will be accounted for in the fourth quarter. Ford indicated that $5.5 billion in cash charges are anticipated through 2027.

The reorganization will influence numerous factories and employees. It also signifies that Ford’s next-generation all-electric truck — internally referred to as “T3” — is now off the table. The T3 was intended to be a completely new design, unlike the Lightning, which integrated electric vehicle technology into a gas vehicle framework. Ford confirmed to TechCrunch that it will also scrap plans for a next-generation commercial van, while the current E-Transit model will continue production.

“Ford has decided against producing certain larger electric vehicles where the business case has weakened due to lower-than-anticipated demand, elevated costs, and regulatory shifts,” the company stated.

The company remains committed to launching a mid-sized all-electric pickup truck in 2027, as confirmed Monday. The platform for this truck — developed from a skunkworks initiative led by former Tesla officials Doug Field and Alan Clarke — will also serve as the foundation for future Ford models. Ford stated it is still on track to start producing cost-effective lithium iron phosphate batteries in 2026. These LFP batteries, to be manufactured at the BlueOval Battery Park Michigan in Marshall and utilizing technology licensed from China’s CATL, will be incorporated into the mid-sized truck.

“Instead of investing billions more into large EVs that currently lack a path to profitability, we are redirecting that investment into higher-return sectors, more hybrid trucks and vans, extended-range electric vehicles, affordable EVs, and completely new ventures like energy storage,” Ford president Andrew Frick mentioned during a press call.  

Ford unveiled the F-150 Lightning in 2021, two years following its announcement of an all-electric Mustang, the Mach-E. Ford hinted at a $40,000 price for the Lightning, which was intended to be a flagship model in the company’s $22 billion initiative focused on electric vehicles.

However, like many large electric trucks, the F-150 Lightning faced challenges in the U.S. market. A contributing factor was that the anticipated $40,000 price was largely inaccessible for most consumers, as that base model was specifically aimed at fleet clients. Over the past two years, Ford managed to sell approximately 7,000 Lightnings each quarter, peaking at nearly 11,000 in the fourth quarter of 2024.

Electric vehicles have encountered significant obstacles since the launch of the F-150 Lightning. Tesla initiated an aggressive price war to mitigate declining sales, thereby impacting the slim (or negative) margins of traditional automakers. The reelection of Donald Trump, along with the Republican takeover of Congress, has resulted in a reversal of numerous Biden-era policies intended to promote electric vehicle sales.

Following the earlier announcement this year regarding a framework for an open AI ecosystem, the nonprofit Creative Commons has expressed its support for “pay-to-crawl” technology — a method designed to automate payment for website content accessed by automated systems, such as AI web crawlers.

Creative Commons (CC) is widely recognized for leading the licensing initiative that enables creators to share their works while maintaining copyright. In July, the organization unveiled a strategy to create a legal and technical structure for sharing datasets between companies that own the data and AI developers seeking to utilize it for training.

Now, the nonprofit is cautiously endorsing pay-to-crawl models, indicating it is “cautiously supportive.”

“If implemented effectively, pay-to-crawl could offer a mechanism for websites to support the production and dissemination of their content, while managing substitute uses, ensuring that content remains accessible to the public where it might not otherwise be shared or might succumb to more stringent paywalls,” a blog post from CC stated.

Led by companies such as Cloudflare, the concept of pay-to-crawl aims to impose charges on AI bots each time they extract information from a site to compile content for model training and updates.

Previously, websites permitted web crawlers to index their content for search engines like Google without restrictions. They reaped the benefits of this partnership through visibility in search results, attracting visitors and clicks. However, with the advent of AI technology, this relationship has transformed. After a consumer receives an answer from an AI chatbot, they are often disinclined to visit the original source.

This change has already proven detrimental to publishers by drastically reducing search traffic, and there are no signs of this trend reversing.

Conversely, a pay-to-crawl approach could assist publishers in recovering from the financial impact caused by AI. Moreover, it could be advantageous for smaller web publishers lacking the leverage to negotiate bespoke content agreements with AI providers. Major arrangements have been made between firms like OpenAI and Condé Nast, Axel Springer, and more, as well as with Perplexity and Gannett; Amazon and The New York Times; and Meta with various media publishers, among others.

CC has provided several stipulations regarding its support for pay-to-crawl, highlighting that such systems might centralize power on the internet. It could also restrict content access for “researchers, nonprofits, cultural heritage institutions, educators, and other entities operating in the public interest.”

It proposed a set of guidelines for the responsible implementation of pay-to-crawl, including avoiding making it the default option for all websites and steering clear of blanket regulations for the entire web. Additionally, it advocated for pay-to-crawl systems to permit throttling, not merely blocking, and to maintain public interest access. They should also be transparent, interoperable, and constructed with standardized components.

Cloudflare is not the sole corporation invested in the pay-to-crawl arena.

Microsoft is also developing an AI marketplace for publishers, while newer startups like ProRata.ai and TollBit are beginning to explore similar avenues. A consortium known as the RSL Collective has introduced its own specification for a new standard referred to as Really Simple Licensing (RSL), which would regulate what areas of a website crawlers could reach but would not block their access. Cloudflare, Akamai, and Fastly have since embraced RSL, which has the backing of Yahoo, Ziff Davis, O’Reilly Media, and others.

CC was also among those who announced its endorsement of RSL, along with CC signals, its wider initiative aimed at developing technology and resources for the AI era.