Tech Brew Ride Home - Mon. 06/02 – How DoorDash Has Quietly Been Killing It

Episode Date: June 2, 2025

We know WWDC might be underwhelming this year, but to what degree? Is Samsung about to pick Perplexity as its horse in the AI race? AI based acquisition and wrapups continue to be the new hotness in VC investing. And how DoorDash has quietly been killing it in the delivery space. Sponsors: AGNTCY.ORG QualiaLife.com/ride and code RIDE Links: Apple Developer Event Will Show It’s Still Far From Being an AI Leader (Bloomberg) Samsung Nears Wide-Ranging Deal With Perplexity for AI Features (Bloomberg) Google quietly released an app that lets you download and run AI models locally (TechCrunch) Early AI investor Elad Gil finds his next big bet: AI-powered rollups (TechCrunch) Meta Aims to Fully Automate Ad Creation Using AI (WSJ) DoorDash CEO Tony Xu is taking on the role of industry consolidator in food delivery (CNBC) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Tech meme right home for Monday, June 2nd, 2025. I'm Brian McCullough today. We know WWDC might be underwhelming this year, but to what degree? Is Samsung about to pick perplexity as its horse in the AI race? AI-based acquisition and wrap-ups continue to be the new hotness in VC investing, and

how DoorDash has quietly been killing it in the delivery space. Here's what you miss today in the world of tech. Guess what a week from today is? It's WWDC next Monday. So what can we expect from Apple's big event this year? Who do you think I'm going to turn to? Quoting Mark a Gurman. People within the company believe that the conference may be a letdown from an AI standpoint. Others familiar with the company's plan announcements worry they could make Apple's shortcomings even more obvious. There are also signs the event will be smaller scale than the last two WWDCs, The Vision Pro was announced in 2023, and Apple Intelligence was shown in 2024. The company desires to make up for that next year at WWDC, 2006, when it hopes it can try to convince consumers that it's an AI innovator.

That's risky, though. Taking a gap year is hard to do when your competitors are moving at light speed. The most significant AI announcement at WWDC will be the opening of Apple's foundation models to third-party developers. This move will let app creators tap into the company's on-device technology that it currently, uses to handle lightweight tasks such as text summarization. We're talking about large language models that have about 3 billion parameters, a measure of their complexity and learning capacity, far less than the cloud-based ones used by OpenAI and even Apple itself for its in-house cloud-powered AI features. Developers will be able to use the foundation models to make custom app features,

marking an important but incremental step in Apple's AI journey. Sticking to the subject of AI, the company will also introduce a new power management mode, pitch a reboot of a its Translate app now integrated with AirPods and Siri, and quietly rebrand several existing features in apps like Safari and Photos as AI-powered. Apple hopes it'll have a better story to tell in the coming years when more features are planned to be ready. The projects underway include LLM Siri, a bold redo of the assistance architectures that should eventually make it more like the chat GPT voice mode. The hope is to finally give Siri a conversational interface. A revamped version of its shortcuts app, which today lets users create actions such as

launching certain features within apps or playing a particular playlist. The new version will let consumers create those actions using Apple intelligence models. This has long been planned for 2025, but delays may push it into 2026. Also, Apple's AI doctor service, codenamed Mulberry, and a redesigned health app. This project remains deep in development, but probably won't be shown this year. It's currently slated to be released in 2026 as part of a spring update to iOS 26. a chat GPT chatbot competitor that can pool in data from the open web, which some employees have dubbed knowledge. The early project is being led by Robbie Walker, who saw Siri removed from his command several weeks ago.

It was moved over to Vision Pro creator Mike Rockwell. Employees familiar with the work say it's already been plagued by some of the same problems that delayed the Siri overhaul. Apple could always give a preview of future technologies at WWDC, but it will be more cautious about doing that this year. When the company showed off Apple Intelligence in 2024, much of the technology wasn't close to ready. It lost some credibility that it now hopes to win back. As I reported in my recent big take story about Apple's AI efforts, the company is prioritizing announcing features that it knows for sure will launch in the fall. In terms of more traditional scoops, I'm also told that the company has homed in on Lake Tahoe as its next moniker, making it MacOS Tahoe. It's a famous resort area

and a vacation destination and a second home site for many Apple employees, end quote. German then goes on to outline how the various OS releases, as we've previously mentioned, will be renamed for the year going forward. This will be iOS 26, for example. They're doing 26 and not 25 because, you know, the OSs don't come out until later in the year anyway. But remember, remember what I asked about naming the phones for years as well? Well, quoting Mark, one lingering question since my report has been whether Apple will make a similar switch, for its devices. In other words, will the iPhone 17 end up being called the iPhone 26? As of now, I am skeptical that will happen. The version numbers on iPhones are well known to consumers,

and it would be jarring to change them. The other factor here is that people tend to hang on to iPhones for more than one year. Tying the hardware to a specific year would literally make older iPhones outdated, end quote. All right, agree to disagree, but I can see the logic there. German also has sources saying Samsung is nearing a deal to invest in AI startup perplexity. The plan would be to maybe preload perplexity's app on Samsung devices, add perplexity search to Samsung's browser, and more. Quote, the firms have also discussed weaving perplexities technology into Samsung's Bixby personal assistant, said the people who asked not to be identified because the talks

are private. Samsung is planning to announce the perplexity integrations. As early as this year, the people said, with the goal of including the service, as a default assistant option in the Galaxy S-26 phone line that's slated to launch in the first half of 2026. The tech giant is also expected to be one of the biggest investors and a new round of funding for perplexity, the people said. The startup is in advanced discussions to raise $500 million at a $14 billion valuation, Bloomberg News has reported. The broad tie-up may help Samsung reduce its reliance on Alphabet's Google and pave the way for it to work with a mix of AI developers, similar to Apple's strategy for its devices and services.

For Perplexity, the arrangement would mark its biggest mobile partnership to date and follows a recent integration deal with Motorola. Samsung and Perplexity have also discussed building an AI-infused operating system and AI agents' apps that can tap into functionality from Perplexity and a range of other AI assistants, the people said. Apple has also shown interest in working with Perplexity. The iPhone Maker has discussed using Perplexity as an alternative to Google Search, as well as a substitute for chat chbti integration in the Siri voice assistant Bloomberg News has reported.

It's unclear how perplexity's relationship with Samsung, one of Apple's fiercest rivals, would affect that, end quote. Google has quietly released an Android app called Google AI Edge Gallery on GitHub, letting users run some AI models from hugging face locally on their phones. Quoting TechCrunch, called Google AI Edge Gallery, the app is available for Android and will soon come to iOS. It allows users to find, download, and run compatible models that generate images, answer questions, write and edit code, and more. The models run offline without needing an internet connection, tapping into supported phones processors. AI models running in the cloud are often more powerful than their local counterparts, but they also have their downsides. Some users might be wary of sending personal or sensitive data to remote data centers or want to have models available without needing to find a Wi-Fi or cellular connection. Google AI Edge Gallery, which Google is called.

an experimental alpha release can be downloaded from GitHub. The home screen shows shortcuts to AI tasks and capabilities like Ask Image and AI chat. Tapping on a compatibility pulls up a list of models suited for the tasks such as Google's Gemma 3N. Google AI Edge Gallery also provides a prompt lab users can use to kick off single-turn tasks powered by models like summarizing and rewriting text. The prompt lab comes with several task templates and configurable settings to fine-tune the model's behaviors. Your mileage may vary in terms of performance, Google warns. Modern devices with more powerful hardware will predictably run models faster, but the model size also matters. Larger models will take more time to complete a task, say answering a question

about an image than smaller models do, end quote. The journal says meta aims to help brands fully create and target ads using AI by the end of 2026, including images and budgetary goals, thereby expanding its current AI tools. In other words, they want AI to do the whole thing, soup to nuts. All you have to do to advertise is prompt the AI and it does everything else. Quote, using the ad tools meta is developing. A brand could present an image of the product it wants to promote along with a budgetary goal and AI would create the entire ad, including imagery, video, and text. The system would then decide which Instagram and Facebook users to target and offer suggestions on budget. People familiar with the matter said. Meta also plans to enable

advertisers to personalize ads using AI so that users see different versions of the same ad in real time based on factors such as geolocation, the people said. A person seeing an advertisement for a car in a snowy place, for example, might see the car driving up a mountain, whereas a person seeing an ad for that same car in an urban area would see it driving on a city street. Increasingly sophisticated AI tools are starting to enable businesses to create full photo and video ads without having to spend the money and higher production crews. Google released a new version of its video generation tool called VO at its last annual developer conference last month, that allows people to create short videos from a text prompt.

In the not too distant future, we want to get to a world where any business will be able to just tell us what objective they're trying to achieve, like selling something or getting a new customer, how much they're willing to pay for each result, and connect their bank account, and then we just do the rest for them, Mark Zuckerberg said during the company's annual shareholder meeting last week on a separate podcast's appearance, he called it a redefinition of the category of advertising. Generating ads from scratch with AI could be a boon to small and mid-sized businesses, which represent most of the advertisers on meta's platforms, and often lack big budgets for ad creation, end quote.

Yeah, just this weekend, I designed some new image ads for resume writers on Google ads. Took me all of five minutes. Set it and forget it. I told you this was becoming a trend. Elad Gill, backer of perplexity, character AI, Airbnb, Coinbase, and Stripe has invested in Enam company. which aims to use AI to reinvent businesses via PE roll-ups, quoting TechCrunch. The idea is to identify opportunities to buy mature people-intensive outfits like law firms and other professional services firms, help them scale through AI, then use the improved margins to acquire other such enterprises and repeat the process.

Gill has been at it for three years, he says. It just seems so obvious, said Gil over a Zoom call earlier this week. This type of generative AI is very good at understanding language, manipulating language, manipulating text, producing text, and that's audio, that's video, that includes coding, sales outreach, and different back office processes. If you can, quote, effectively transform some of those repetitive tasks into software, he said, you can increase the margins dramatically and create very different types of businesses. The math is particularly compelling if one owns the business outright, he added. If you own the asset, you can transform it much more rapidly than if you're just selling software

as a vendor, Gill said. And because you take the gross margin of a company from, say, 10 to 40%, that's a huge lift. Suddenly, you can buy other companies at a higher price than anyone else because you have that increased cash flow per business. You have enormous leverage on the business on a relative basis, so you can do roll-ups in ways that others can't, end quote. So far, Gill has backed two companies pursuing the strategy, according to the information. One is a one-year-old company called a NAM company focused on worker productivity, which has been valued at more than $300 million by its backers, including Andresen Horowitz and Open AI's startup fund. Though Gill says he can't discuss specifics of the private deals, he suggests the approach

represents something new. There used to be these technology-enabled roll-ups 10 years ago, and most of them kind of ended up being not really that much of a user of technology, he says. It was kind of like a thin veneer painted on to increase the valuation of the company. I think in the case of AI, you can actually radically change the cost structure of these things, and quote. Finally today, you know who's been quietly killing it? Doordash. This weekend I read on CNBC a fascinating profile of DoorDash CEO Tony Zhu, who has transformed the once-scrappy delivery startup into a $90 billion industry consolidator

through an acquisition spree. Quote, since starting DoorDash on the campus of Stanford University in 2013, the now 40-year-old CEO has navigated the notoriously cutthroat and low-margin business of food delivery, building a company that Wall Street today values at close to $90 billion. The stock has emerged as a tech darling this year, jumping 23% while the NASDAQ is still down for the year, largely on tariff concerns. More than four years after its IPO net profits remain slim, but that's not getting in the way of Zhu's mission to become an industry consolidator using a combination of cash and new debt to fuel an acquisition spree at a time when big tech deals remain scarce. Earlier this month, DoorDash scooped up British food delivery

startup deliveroo for about $3.9 billion and restaurant technology company seven rooms for 1.2 billion dollars. What we've delivered for a customer yesterday probably isn't good enough for what we will deliver for them today, Zhu told CNBC's Squawk Box after the deals were announced. This week, Doordash announced the pricing of a $2.5 billion convertible debt note and said the proceeds could be used in part for acquisitions. The San Francisco-based company has a history with scooping up competitors to grow market share. In 2019, it bought food delivery competitor Caviar for $410 million from Square, now known as Block. About two years later, Dorda's DoorDash said it was paying $8.1 billion for international delivery platform Wolt. The deal was

its last big transaction until this month. When DoorDash entered the food delivery market, it had to face off against the likes of Grubhub and Seamless, which later joined forces. That combined entity was bought late last year by Restaurant Owner Wonder Group. In 2014, Uber launched Uber Eats, which is now DoorDash's biggest competitor in the U.S. It's a very competitive market, and I think merchants do have a choice, Zhu said in the CNBC interview. what we're focused on is always trying to innovate and bring new products to match increasing standards and expectations from customers. DoorDash differentiated itself early on by cornering suburban markets that had fewer delivery options, while other players attacked city centers. When COVID

shut down restaurant dining in early 2020, DoorDash capitalized on the booming demand for deliveries, revenue more than tripled that year and grew 69% in 2021. Colleagues and early investors credit a customer-first focus for much of Zhu's success, Goko Rajaram, who joined DoorDash through its caviar acquisition describes you as the best operational leader in the U.S. after Amazon founder Jeff Bezos. Restaurants haven't universally viewed DoorDash as an ally. Commissions can reach as high as 30%, which is a hefty cut to fork over. Many restaurants have reluctantly paid the high fees because of DoorDash's dominant market share, which reached an estimated 67%. In 2021, the company introduced three tiers of pricing with a basic option at 15%

for more price-sensitive businesses. DoorDash needs the high fees in order to stay in the black. The company's contribution profit as a percentage of total marketplace volume hovers below 5%. End quote. Nothing more for you today. Talk to you tomorrow.