A (0:04)

Welcome to the Tech We Write Home for Thursday, January 29, 2026 I'm Brian McCullough. Today meta earnings good the street likes the AI spending. Microsoft earnings Bad why is growth in their cloud business not as robust all of a sudden? Elon is following through on taking Tesla all in on robots and AI. The AI inspired layoffs are not just for the tech industry anymore. And are you actually technical enough to run that? Claude Bottom here's what you missed today. In the world of tech, Everyone seems to be deploying AI agents right now. But the thing is, sometimes AI agents mess up. They go off script or make changes you didn't authorize, forcing you to diagnose and fix it. The good news is Rubrik Agent Cloud helps prevent that extra work. It's the only platform that allows you to monitor, govern and rewind AI agent actions. It's like an undo button for AI it runs in the background the whole time and makes sure things stay on track. On a singular platform, Rubrik helps you unleash more agents faster while mitigating risk. So if you're running AI agents and want to rest a little easier at night, check it out. Right now my listeners get exclusive early access to Rubrik Agent cloud. Head to rubrikrik.com that's R U B R I K.com I'm going to run down a bunch of earnings news here, all in one segment, starting with Meta, who reported earnings yesterday that were fine. Meta stock is up nearly 8% this morning, and that comes even as Meta said it expects its 2020 capex spend to be $115 to $135 billion. That compares to a $110.6 billion analyst estimate and the $72.2 billion it spent on capex just last year. This is all, of course, driven by investments in their superintelligence labs and quoting TechCrunch in 2025 we rebuilt the foundations of our AI program, Mark Zuckerberg said on an investor call Wednesday, referring to the company's recently restructured AI lab. Over the coming months, we're going to start shipping our new models and products, and I expect us to steadily push the frontier over the course of the new year. End quote. But while Zuckerberg didn't give specific timelines or products, he highlighted AI driven commerce as a particular area of focus for Meta. This also has implications for commerce, zuckerberg continued. New agentic shopping tools will allow people to find just the right set of from the businesses in our catalog. That proposal echoes broader interest in AI powered shopping assistants across the industry. Both Google and OpenAI have built platforms for agent enabled transactions, with companies like Stripe and Uber signed on as partners. But while other AI labs have already built significant technical infrastructure, Meta believes its access to personal data will prove uniquely valuable. We're starting to see the promise of AI that understands our personal context, including our history, our interests, our content and our relationships, zuckerberg said on the call. A lot of what makes agents valuable is the unique context that they can see, and we believe that Meta will be able to provide a uniquely personal experience. So that was Meta good. The street seems to buy into that AI narrative and are sanguine about continued spending to get there, I guess. But conversely, Microsoft not good, down almost 12% this morning 12% in one day is pretty pretty not good, right? So what's up here? Quoting cnbc, the software maker showed slowing growth in its Azure cloud segment, which declined to 39% growth from 40% growth in the company's first fiscal quarter. Investors closely watched the segment as a stand in for measuring enterprise AI demand. At the same time, capital expenditures and finance leases in the quarter jumped 66% to $37.5 billion, surpassing the $34.31 billion expected by analysts polled by Visible Alpha as the company supports demand for its cloud and AI segments. The company also said it's str struggling with compute capacity constraints as demands continue to outweigh supply. Microsoft finance chief Amy Hood said Azure would have grown 40% if the company had funneled all of its new GPU chips in the first and second quarter into its Azure business. Microsoft also said its cloud contracted backlog grew 110% to $625 billion, around 45% of which comes from OpenAI alone. So that's basically money that should come into Microsoft, some in theory if no bubbles burst. Also, I want to note a third company, Skhenix, has surpassed Samsung in annual operating profit for the very first time. SK Hynix posted a record profit of around $33 billion in 2025, while Samsung posted 30 and a half billion dollars. The reason here? Well, they can't sell those memory chips fast enough. Bunch of Tesla news this morning. Tesla plans to scrap its Premium S and X models and convert its California factory into an Optimus manufacturing hub as Elon Musk refocuses the company on Robotics and AI. Quoting the FT, Tesla plans to scrap two models and invest $2 billion in Elon Musk's X AI as the electric vehicle pioneer accelerates a charge into robotics and artificial intelligence following its first drop in annual revenue. In the clearest sign yet of where Musk is steering Tesla, the company said it would end production of the premium S and X models next quarter and convert its California factory into a manufacturing hub for its optimus robots. The announcements came as Tesla's fourth quarter results laid bare the damage to the carmaker from a year dominated by the Trump administration's ending of EV tax incentives, Musk's polarizing politics and the continued rise of low cost. Chinese rival byd Announcing the decision to ditch the premium S and X models, Musk said that is slightly sad, but it's part of our overall overall shift to an autonomous future. The world's richest man has gambled Tesla's future on self driving cybercabs and AI enabled humanoid robots. The group has begun calling itself a physical AI company. Tesla has also pressed ahead with a $2 billion investment in Musk's XAI as part of its January funding round, despite lukewarm shareholder support for that move. SpaceX, the billionaire's rocket company, has already invested $2 billion in Xai as he seeks more financial support for the capital intensive business. The move to link TESL closely to Musk's AI group came after a non binding shareholder resolution in November pushed for such an investment. While the measure received more votes in favor than against, the combined number of abstentions and no votes showed a majority of shareholders were not behind the move. End quote. Maybe it's not just tech companies. Dao Chemical maker Dow says it plans to cut 4,500 staff to save costs and will rely on AI to boost productivity, resulting in 1.1 to $1.5 billion in charges. Dow currently employs around 34,000 people. Quoting the Journal, Dow said it is embarking on a program dubbed Transform to Outperform, which would employ artificial intelligence and automation to reduce expenditure and catalyze growth and productivity, yielding an additional $2 billion in operating earnings before interest depreciation, and AM Chairman and CEO Jim Federling said the program represents a comprehensive and radical simplification of our operating model. The company currently employs about 34,000 people and operates sites in 29 countries. Dow anticipates several charges and other costs associated with the program, including $600 million to $800 million in severance costs and 500 to $700 million in other one time costs. In 2027, Dow said the overhaul would result in $1.2 billion in incremental operating EBITDA with charges of about DOL $500 million in 2028. The program would add $300 million in operating EBITDA with no further charges, end quote. Waymo says one of its robo taxis struck a child at 6 mph near an elementary school in Santa Monica on January 23, resulting in minor injuries, quoting TechCrunch. The NHTSA has opened an investigation into the accident and Waymo said in a blog post that it will cooperate fully with them throughout the process. Waymo said its robo taxi struck the child at 6 mph after braking hard from around 17 mph. The young pedestrian suddenly entered the roadway from behind a tall SUV moving directly into our vehicle's path, the company said in its blog post. Waymo said its vehicle immediately detected the individual as soon as they began to emerge from behind the stopped vehicle. Following contact, the pedestrian stood up, immediately walked to the sidewalk and we called 911. The vehicle remained stopped, moved to the side of the road and stayed there until law enforcement cleared the vehicle to leave the scene, waymo wrote in the Post. News of the crash comes as Waymo faces dual investigations into its robo taxis illegally passing school buses. The NHTSA opened a probe into the problem in October, shortly after the first report of the incident in Atlanta, Georgia, and the National Transportation Safety Board opened its own investigation last week after around 20 incidents were reported in Austin, Texas. According to the NHTSA, the accident occurred within two blocks of the elementary school during normal school drop off hours, the safety regulator said. There were other children, a crossing guard and several double parked vehicles in the vicinity. The NHTSA's office of defects investigation is investigating whether the Waymo AV exercised appropriate caution given, among other things, its proximity to the elementary school during drop off hours and the presence of young pedestrians and other potential vulnerable road users, End quote. Picture this. You've been chosen to lead AI onboarding for your team. That includes everyone from seasoned technical pros to AI newcomers. No pressure, right? That's where ARIA comes in. Aria's no Code, Low Code and Pro Code platform makes it easy for your organization to embrace AI. Regardless of technical experience. With areas unified security layer, advanced threat detection and robust compliance measures, your AI ecosystem can stay safe while your team innovates. Worried about staying on budget, ARIA's cost optimization tools can help you manage and forecast your AI spend so there are no surprises. Get started@aria.com morning brew that's aria.com morning brew with everything you balance at work, wouldn't it be nice to have your own personal chief of staff? Something that can capture meetings, calls in person and online conversations without forcing anyone to use their phone or break eye contact. Well, that's exactly what Plaud is, a dedicated AI note taking device designed to capture in person conversations so you can stay in the moment. Plod devices are small, secure and can automatically deliver accurate transcripts, clear summaries with action items right in your phone. And with over 20 hours of continuous recording you can be more present without losing key details or phone battery life. For a limited time, you can get 10% off any plaud device when you check out with code tbrh. Visit Plaud AI TBRH to learn more. That's P L A U D A I TBRH CodetBRH US based AI startup Acri has released Trinity Large, a 400 billion parameter open weight model that it says compares to Meta's Llama 4 Maverick 400B on some benchmarks. I'm flagging this because look, it's not just the Chinese. We can make impressive open weight models right here in the good old US of A. Quoting TechCrunch Many in the industry think the winners of the AI model market have already been decided. Big tech will own it. Google, Meta, Microsoft, a bit of Amazon, along with their model makers of choice, largely OpenAI and anthropic but tiny 30 person startup Acri AI disagrees. The company just released a truly and permanently open Apache license general purpose foundation model called Trinity and ACRI claims that at 400 billion parameters it is among the largest open source foundation models ever trained and released by a US company. Acri says Trinity compares to Meta's Llama 4 Maverick 400B and Zai's GLM 4.5, a high performance open source model from China's Tashinga University, according to benchmark tests conducted using base models. Very little post training. Like other state of the art SOTA models, Trinity is geared for coding and multi step processes like agents. Still, despite its size, it's not a true SOTA competitor yet because it currently supports only text. More models are in the works. A vision model is currently in development and a speech to text version is on the roadmap, CTO Lucas Atkins told TechCrunch. In comparison, Meta's Llama 4 Maverick is already multimodal, supporting text and images. But before before adding more AI models to its roster, ACRI says it wanted a base LLM that would impress its main target customers, developers and academics. The team particularly wants to woo US companies of all sizes away from choosing open models from China. Ultimately the winners of this game and the only way to really win over the usage is to have the best open weight model, atkins said. To win the hearts and minds of developers, you have to give them the best. The benchmarks show that the Trinity base model, currently in preview while more post training takes place, is largely holding its own and in some cases slightly besting llama on tests of coding and math. Common sense knowledge and reasoning the progress ACRI has made so far to become a competitive AI lab is impressive. The large Trinity model follows two previous small models released in December the 26B parameter Trinity Mini, a fully post trained reasoning model for tasks ranging from web apps to agents and the 6b parameter Trinity nano, an experimental model designed to push the boundaries of models that are tiny yet chatty. The kicker is ACRI trained them all in six months for $20 million in total using 2048 Nvidia Blackwell B300 GPUs. This out of the roughly $50 million the company has raised so far, said founder and CEO Mark McQuaid. That kind of cash was a lot for us, says Atkins, who led the model building effort. Still, he acknowledged that it pales in comparison to how much bigger labs are spending right now. Finally today, some news about Claudbot. First, it's no longer called claudebot. Anthropic's lawyers finally came calling. I guess so. Now it is officially Mult Bot. But second, security researchers have a warning for all of you out there tinkering with this thing. They caution that Moltbot Bot requires a specialist skill set to use safely, as data exposure risks persist even when it is set up correctly. Quoting the Register Would you be comfortable handing the keys to your identity kingdom over to a bot, one that might be exposed to the open Internet? Claudebot, now known as Mult Bot, has gone viral in AI and developer circles in recent days, with fans hailing the open source AI personal assistant as a potential breakthrough. The long and short of it is that Multbot can be controlled using messaging apps like WhatsApp and Telegram in a similar way to the Genai Chatbots everyone knows about. Taking things a little further, its agentic capabilities allow it to take care of Life Admin for users, such as responding to emails, managing calendars, screening phone calls or booking table reservations, all with minimal intervention or prompting from the user. All that functionality comes at a cost, however, and not just the outlay so many seem to be making on Mac Mini purchases for the sole purpose of hosting a Multbot instance. In order for Multbot to read and respond to emails and all of the rest of it, it needs access to accounts and their credentials. Users are handing over the keys to their encrypted messenger apps, phone numbers and bank accounts to this Agentix system. Naturally, security experts have had a few things to say about that. First, there was the furor around public exposures. Multbot is a complex system, and despite being as easy to install as a typical app on the face of it, the misconfigurations associated with it prompted experts to highlight the dangers of running Multbot instances without the proper know how. Jameson O'Reilly, founder of Reddit teaming company Dvaan, was among the first to draw attention to the issue, saying that he saw hundreds of cloudbot instances exposed to the Web potentially leaking secrets. He told the Register that the attack model he reported to Multbot's developers, which involved proxy misconfigurations and local host connection auto authentication, is now fixed. However, if exploited, it could have allowed attackers to access months of private messages, account credentials, API keys and more anything to which claudebot owners gave it access. According to his Shodan scans supported by others looking into the matter, he found hundreds of instances exposed to the Web. If those had open ports allowing authenticated admin connections, it would allow attackers to access the full breadth of secrets in Moltbot, eric Schwacke, director of cybersecurity strategy at SALT Security, told the Register. A significant gap exists between the consumer enthusiasm for claudebot's one click appeal and the technical expertise needed to operate a secure agentic gateway. While installing it may resemble a typical Mac app, proper configuration requires a thorough understanding of API posture governance to prevent credential exposure due to misconfigurations or weak authentication, many users unintentionally create a large visibility void by failing to track which corporate and personal tokens they've shared with the system. Without enterprise level insight into these hidden connections, even a small mistake in a prosumer customer setup can turn a useful tool into an open backdoor, risking exposure of both home and work data to attackers. The security concerns surrounding Multbot persist even when it is set up correctly, as the team at Hudson Rock pointed out this week. Its researchers said they looked at Multbot's code and found that some of the secrets shared with the assistant by users were stored in plain text markdown and JSON files on the user's local file system. The implication there is that if a host machine such as one of the Mac Minis being bought en masse to host Multbot, were infected with infosteeler malware, then it would mean the secret stored by the AI assistant could be compromised. Hudson Rock is already seeing malware as a service. Families implement capabilities to target local first directory substructures such as those used by Molbot, including Redline, Luma, and Vidar. It is fathomable that any of these popular strains of malware could be deployed against the Internet exposed Multbot instances to steal credentials and carry out financially motivated attacks. If the attacker is also able to gain write access, then they can turn Multbot into a backdoor, instructing it to siphon sensitive data in the future. Trust malicious sources and more. Claudebot represents the future of personal AI, but its security posture relies on an outdated model of endpoint trust, said Hudson Rock. Without encryption at rest or containerization, the local first AI revolution risks becoming a goldmine for the global cybercrime economy and crash Quote. Nothing more for you today. Talk to you tomorrow.