Tech Brew Ride Home - AI 50% Off!

Episode Date: August 13, 2026

Anthropic's investors talked up a $2T+ October IPO, even as data showed Fable 5 barely selling. Google cut prices on Gemini 3.7 Flash, OpenAI previewed a 14× faster tier, Trump enlisted private hackers, and Twitch fed Amazon's AI. Links Google's Gemini 3.7 Flash targets coding and agents with a 50% introductory price cut (VentureBeat) OpenAI previews Ultrafast, an API tier powered by Cerebras that runs GPT-5.6 Sol up to 14× faster and generates up to 750 output tokens per second (9to5Mac) President Trump signs a memo letting the US government partner with private companies to conduct cyberattacks abroad against criminal groups targeting Americans (Bloomberg) Sources: Anthropic's investors expect it to float at a $2T+ valuation in an October IPO and to hit $100B to $120B in annualized revenue by the end of 2026 (Financial Times) Ramp data: Fable 5 drew just 6% of Anthropic's API tokens in its first month and 75% of GPT-5.6 Sol's model revenue, suggesting corporate willingness to pay for frontier AI has hit a ceiling (The Decoder) Databricks closed a $5B funding round at a $190B valuation, six months after raising $5B at a $134B valuation, and says it has crossed $7B in revenue run rate (CNBC) Twitch says it intends to use videos streamed on its platform to help train Amazon's generative AI content models and adds a setting for creators to opt out (TechCrunch) Subscribe to the ad-free feed.

Welcome to the TechBrewrite Home for Thursday, August 13th, 2026. I'm Brian McCullough today. Anthropic investors talked up a $2 trillion plus October IPO, even as data showed Fable 5 maybe has hit a price ceiling. Google cut prices on Gemini 3.7 Flash. OpenAI previewed a 14x faster tier. Trump enlisted private hackers and Twitch wants to feed Amazon's AI with your videos. Here's what you missed today in the world of tech. Every day shareholders meet to discuss important matters about the companies you invest in. Now you can make your voice heard too. Vanguard Investor Choice makes it easy to set your proxy voting preference for your eligible Vanguard index funds. Whether you hold a Vanguard fund directly or through another brokerage firm, all it takes is a few clicks to select your proxy voting preference and be heard on important shareholder topics like executive pay and director elections.

Visit vanguard.com slash investor choice to learn more. It's your shares, it's your voice, it's easy. Vanguard investors own shares of our index funds, and those funds own shares of the companies they invest in. Vanguard Marketing Corporation distributor. Well, I mean, this is what we're going to have every day for the foreseeable future, right? More models, new models. Google has unveiled Gemini 3.7 Flash, what it calls its most intelligent workhorse model for coding and agents. But what's interesting is the pricing. They are pricing it at 75 cents per 1 million. input and $3.75 per 1 million output tokens at launch. And let me underline that. We're seeing the

race to compete on pricing in real time now, aren't we? People online seem to be at least at first blush gushing about the capability of this model and then loving the price at which they can do that capability. Remember, Google has the advantage that they can basically subsidize this sort of thing. Quoting Venture Beat. Google is rolling out Gemini 3.7 Flash, a new version of its workhorse, AI model that puts coding agenetic workflows and knowledge work at the center of the upgrade while temporarily cutting API prices in half. The release arrives just three weeks after the release of Gemini 3.6 Flash an unusually short turnaround that Google attributed to developer feedback and algorithmic improvements.

For enterprise developers, the more consequential story may be the combination of those intelligence gains with the lower inference costs. Through the end of 2026, Gemini 3.7 Flash costs 75 cents per million input tokens and $3.75 per million output tokens. Starting January 1st, pricing rises to $1.50 per million input tokens and $7.50 per million output tokens. That means the current discount is temporary, but it gives teams deploying high-volume coding and business agents several months to evaluate whether Google's claimed reductions in retries and manual oversight translate into lower total operating costs. The launch also underscores Google's rapid iteration on its flashline, while

its next flagship pro model remains absent. Google did not provide a release date for Gemini 3.5 Pro with Thursday's announcement. Writers reported, despite the model having previously been described as undergoing partner testing. Axios similarly noted that 3.7 Flash arrives before the anticipated pro release. Google describes Gemini 3.7 Flash as its most intelligent workhorse model yet for coding and agents. The company says the model is better at adapting when it encounters roadblocks, clarifying intent when necessary and following instructions with greater fidelity. Those improvements matter beyond benchmark scores in an enterprise coding agent, a model that makes

fewer unnecessary changes, recovers from errors and executes multi-step plans more reliably, can reduce the number of human interventions needed to complete a task. The same principle applies to business agents operating across documents and applications, where an incorrect tool call or poorly interpreted instruction can derail an otherwise useful workflow. Google says 3.7 Flash thinks more diligently applying more effort to multi-step planning and tool calls. Its stated goal is more disciplined execution with fewer retries and less manual supervision. That represents an interesting evolution from Gemini 3.6 Flash. Google's developer documentation described 3.6 as reducing reasoning steps, conversational turns, and tool calls compared with earlier models

while attempting to limit execution loop spiraling. With 3.7, the emphasis shifts toward putting sufficient effort into planning, while improving the quality of execution, potentially a more useful optimization than simply minimizing the number of steps an agent takes. Google DeepMind said in a post accompanying the release that 3.7 flash shows gains in debugging and issue resolution generates more functional web layouts and applications with fewer prompts and improves reasoning and accuracy on real-world business workflows, end quote. Well, you know, competing on price can also mean competing on execution and speed, because those can all mean the same thing in the end, quoting 9 to 5 Mac.

OpenAI is previewing a new way to run its most capable GPT 5.6 model at dramatically higher speeds. The company says its new ultra-fast service tier can run GPT 5.6 sole up to 14 times faster than standard processing. Ultra-fast mode launches first through the OpenAI API and its power. by Cerebrus. It can generate up to 750 output tokens per second, potentially bringing frontier level performance to workflows where latency matters as much as model intelligence. Opening I introduced GPD 5.6 Seoul in June alongside the balanced Terra and speed-focused Luna models. The full family became broadly available in July, including through ChatGPT, Codex, and the API. The company sees ultra-fast supporting live or near-production tasks, including voice,

customer support, commerce, developer agents, financial research, and security response. OpenAI says its own developers have used it to analyze logs and traces during incidents as well as compress research cycles that previously ran overnight into multiple iterations during the workday. Access remains limited to a select group of customers, while OpenAI evaluates how the added speed changes real-world products and expands capacity. Businesses can join the ultra-fast waitlist by sharing their workload, latency requirements, expected usage and other details, end quote. When I get into the nation-state hacking game, well, the U.S. government will now employ you to do just that.

Quoting Bloomberg, the U.S. government is partnering with private sector firms to launch cyber attacks abroad, expanding the scope of national security operations that until now had been largely conducted by government agencies. Private companies under the direct control and oversight of the federal government, will be able to conduct offensive cyber operations aimed at disrupting criminal organizations that target Americans, according to a national security presidential memorandum, signed by President Donald Trump, and release Wednesday. The memo, which will create a new program run by the Department of Homeland Security, marks a significant shift for the U.S. where such operations are usually carried out by the government. The program will be overseen in coordination with

the Justice Department. For months, the Trump administration has been laying the groundwork for private sector firms to legally conduct offensive cyber operations. The idea is highly polarizing due to fears that such activities could lead to unintended consequences and escalation. Advocates argue that tapping a larger pool of professionals would expand the country's ability to counter cyber threats. The move is about granting cyber letters of Mark, said Ari Redboard, head of government affairs at the blockchain intelligence firm, TMR Labs, referring to a 19, century government license that allowed privately owned ships to attack pirates and other designated threats. Sorry, this is me cutting in here to put on the history hat. No, letters of Mark were government-sponsored

piracy, essentially. Quoting again, the federal government's partners would be able to conduct surveillance operations and cyber attacks against foreign cyber-enabled transnational criminal organizations, the memo said. To qualify for the program, companies will have to maintain a bond or escrow of at least $1 million that will be forfeited if they don't comply with contractual agreements, end quote. I guess right now somewhere on a Discord channel, the modern version of Francis Drake just got their marching orders. With this summer's record-breaking heat, it's hard to cool down enough to get a good night's sleep. That's where the pod comes in. The pod by eight sleep is a smart mattress cover that goes over your existing mattress and actively heats or cools each side

of your bed independently. It connects to your wearable, analyzing your daily activity. Then it predicts how you'll sleep and adjusts the pods temperature automatically. What I love about the eight sleep is the simplest of things, the ability to be cool when it's time to go to sleep and warm, when it's time to wake up. And my side is entirely under my control. My wife can be whatever temperature she wants on her side of the bed too. Use code ride home at eight sleep.com slash ride home for up to $350 off. That's code ride home at 8Sleep.com slash ride home. AI threats are like a hydra that learned how to type, cut one head down, and two more pop right back up. Which means the old way we've been handling cybersecurity just won't cut it anymore.

That's where Rubrik comes in. They've engineered a platform that secures your data, controls your AI, and protects your identity. See how Rubrik helps you stay ahead of threats at rubric.com slash mb. That's rubr-r-k-com slash mb. In a different time, Databricks would be the most famous startup in all the land, quoting CNBC. Databricks on Thursday said it closed a $5 billion funding round at a $190 billion valuation to invest in enterprise capabilities. The company said that it crossed $7 billion in revenue run rate and grew more than 80% year over year in its second quarter. CEO Ali Goetzey told CNBC's John Fort on Thursday that demand is crazy.

What's happening basically is everybody's using these agents, AI agents, and the whole world is laser-focused on agents, AI, and the sort of, you know, that core part of it, Goetze said on CNBC's squawk on the street. Goetze highlighted strength in the AI software company's lake-based database unit, genie business agent and its AI gateway tool, which, which helps control model use and costs. The company's recent lake-based database for AI agents has already surpassed $100 million in revenue run rate, Databricks said.

The company said its lakehouse data warehousing tool has surpassed a $1.5 billion run rate. The funding round comes six months after the private data analytics software company raised $5 billion in funding and $2 billion in new debt at a $134 billion valuation, end quote. We're now getting that steady drumbeat of rumblings about the Anthropic IPO that usually means the IPO is probably coming to tweet, quoting the FT. Anthropic investors expect the AI startup to float at evaluation of $2 trillion or more in

October, a dizzying figure that would eclipse SpaceX and make the AI Labs debut the largest ever initial public offering. Half a dozen of the company's backers told the FT that Anthropics rapidly rise revenue would enable it to more than double its current valuation in a planned autumn float. A listing at that level could unlock billions of dollars in gains for the five-year-old company's early investors, but would also test public markets that are growing more nervous about the AI boom. Anthropics backers say booming demand for the lab's advanced AI models and tools justifies their lofty expectations. Investors expect the clodmaker's

annualized revenue to be between $100 and $120 billion by the end of 2026, using the startup's preferred measure, which infers full-year sales from recent performance, up by more than 10 times over the course of 2026. If Anthropic is growing 800% a year, you'd think at the incredibly low end they would trade at 30 times revenue, said one investor in the group that would make them a $3 trillion company. Anthropics' lack of a publicly listed U.S. peer that would provide a benchmark for its valuation is an issue, but companies that are seen as AI beneficiary. Such as Data Intelligence Group Palantir and cloud company Nebius have traded this year at roughly 55 times revenue. Several investors said senior Anthropic executives had yet to fix the valuation target for the IPO even in private conversations, but investors have built their own financial models, end quote.

Well, helping or hampering that, I guess, is this, quoting the decoder. Anthropics Fable 5 is considered the most capable AI model on the market, but new sales data shows U.S. companies are barely adopting it. Spending data from financial services provider ramp shows that companies are barely buying Anthropics' most powerful model through its API. In its first month after launch, Fable 5 accounted for only about 6% of the tokens purchased from Anthropic. Measured against total spending on Anthropic models, that number sat at 11.4%. OpenAIs flagship model, GPD 5.6, Seoul, captured 25% of tokens and 23% of spending at OpenAI. Overall, according to Ramp, Fable 5 brought in only about 75% of the model-related revenue that GPD 5.6 Sol generated, despite costing significantly more per token.

Ramp economists are a Karazanian attributes Fable 5's slow uptake to its price. The model costs about $10 per million input tokens and $50 per million output tokens, making it roughly twice as expensive as GPD 5.6 sole or other anthropic flagship models. Garzanian sees this as a new ceiling on what companies are willing to spend on AI, arguing that the extra performance simply isn't worth the cost. It's likely more complicated than that, though. The Performance Edge Fable Five offers may not matter for many use cases, or it's barely measurable and daily work. This points to a basic problem with calculating AI return on investment. How does a company put a number on the value in AI model delivers, especially when trying to measure the gap between one model generation and the next? It's a complicated and messy equation.

But the data doesn't say that a fable five-class model represents the upper limit of what companies would pay for AI per se. Models that are dramatically more capable could also deliver dramatically higher and more importantly tangible value. Companies will buy what pays off. But as long as that value stays abstract, their willingness to pay appears to be limited. According to Ramp, 43 and a half percent of U.S. companies paid for anthropic subscriptions or tokens in July up one 1.1 percentage points from the previous month. Open AI reached 39.7%, but grew only by 0.23 percentage points, lagging overall AI adoption growth. XAI posted its fastest growth since July 2025, rising 0.94 percentage points, up 4%. New customers keep signing up with American model providers,

but advanced users, who's growing spending OpenAI and Anthropic increasingly depend on, are shifting toward open source models. Ramps data shows those models now trail frontier models by only a few months, and as a result, growth at the two leading AI labs is slowing, end quote. Woo, might want to have that IPO happen sooner rather than later. Finally today, Twitch says it intends to use video streamed on its platform to help train Amazon's generative AI content models, but has added a setting for creators to opt out, quoting TechCrunch.

The streaming platform Twitch will now use creators' content to help train generative AI models for its parent company Amazon, this move has inspired swift and concentrated backlash from the Twitch community, especially because creators are opted in to having their content used for this AI training by default. For Amazon, these stream recordings are incredibly valuable offering thousands of hours of audio and video content to help train AI models. But Twitch users worry that since creators have to manually opt out, they might be surrendering their content to train Amazon's AI content models without even knowing it. This is especially concerning on a platform like Twitch, where creators are often recording live streams of themselves and their voices

for many hours per week. In a stream of the official Twitch channel, Twitch Head of Community, Mary Kish and Chief Product Officer Mike Minton addressed a live audience of nearly 3,000 aggrieved users, many of whom were posting anti-AI sentiments in the chat. Why is it not opt-in? That's what everybody is spamming in chat. I get it. Let me opt-in versus making me opt out. Minton said, well, there's an honest answer. If the this was opt-in, nobody would opt-in. That's honestly the answer. Twitch knows that its community of streamers is largely opposed to the use of generative AI since the most prevalent generative AI products are trained on books, images, videos, and other materials scraped from the internet

without consent. Even Twitch's approach to breaking this news shows that the company is braced for backlash. Instead of telling the community that Amazon would begin training on Twitch users' content, Twitch frame this change as, quote, adding a setting that lets you opt- opt out of having your channel content used to train generative AI content models across Amazon, end quote. So in our new house, we have a smart lock for the first time ever, and that means I find myself with an empty pocket for the first time. For 30 years, I kept my keys in my right pocket and my phone in my left so as not to scratch my phone with my keys. Well, now with the empty pocket, I thought, why not move the phone to my right, since I am right-handed? But, dear listener, I am here to tell you it is basically impossible to change 30 years of muscle memory overnight.

The amount of times each day I think I've lost my phone because I hit my left pocket and it's not there. Talk to you tomorrow.