CEOs Are Telling Their Employees to Embrace AI—or Become Irrelevant

CEOs are urging their teams to embrace AI tools or otherwise be rendered obsolete—and they’re doing so publicly. First, Shopify CEO Tobi Lütke published on X what was intended to be an internal memo on the necessity of employees adopting AI. He posted the lengthy directive on social media because “it was in the process of being leaked and (presumably) shown in bad faith,” he wrote. The move seemed to inspire a bit of FOMO from Micha Kaufman, CEO of gig work marketplace Fiverr, who posted screenshots of a very similar corporate directive a day later. Kaufman says his Fiverr AI mandate was sent out via email to employees on Monday, making its timing almost identical to Lütke’s. On the topic of hiring, the men echoed each other. “It does not make sense to hire more people before we learn how to do more with what we have,” Kaufman wrote. “Before asking for more head count and resources, teams must demonstrate why they cannot get what they want done using AI,” Lütke said. The CEOs are imparting a similar message, presumably meant to spark a fire at their respective companies: AI is coming to reshape the global workforce, and with that, no job is safe from upheaval. It’s a message that has been resonating for a while, as fears of so-called AI agents capable of automating jobs descend on the corporate rank-and-file. Only now, though, top corporate leaders are publicly declaring the inevitable. As Lütke put it: “Using AI effectively is now a fundamental expectation of everyone at Shopify.” He argued that not using AI amounts to stagnation. Performance reviews will now consider employees’ AI usage. “Stagnation is almost certain, and stagnation is slow-motion failure. If you’re not climbing, you’re sliding,” he wrote. Kaufman struck a somewhat alarmist tone, writing: “So here is the unpleasant truth: AI is coming for your jobs. Heck, it’s coming for my job, too. This is a wake-up call.” The Fiverr chief executive seemed to suggest that ignoring AI now means an inevitable battle for relevance later. “If you do not become an exceptional talent at what you do, a master, you will face the need for a career change in a matter of months. I am not trying to scare you. I am not talking about your job at Fiverr. I am talking about your ability to stay in your profession in the industry,” he wrote. Kaufman was unavailable to comment. It’s possible that corporate leaders are trying to squeeze every ounce of productivity out of their workforces, now that automating certain tasks is more accessible than ever. To be sure, Shopify is obsessed with tracking productivity: An internal tool at the company, GSD, tracks the status of every project. GSD, which stands for “get shit done,” is now inseparable from AI: “The prototype phase of any GSD project should be dominated by AI exploration,” Lütke’s announcement said. BY SAM BLUM @SAMMBLUM

OpenAI Releases GPT-4.1, a New Family of Models Designed for Coding

OpenAI’s lineup of powerful AI models is growing. Today, the company behind ChatGPT has announced the GPT-4.1 “model family,” a collection of three new AI models “purpose-built for developers.” The models can now be used via OpenAI’s API. According to OpenAI, the GPT-4.1 family will consist of three models: the normal-size GPT-4.1, the smaller and more modestly priced GPT-4.1 mini, and the even smaller and cheaper GPT-4.1 nano. These models will be available only through OpenAI’s API, and can’t be accessed through ChatGPT’s model picker. OpenAI says this is because the latest version of its GPT-4o model incorporates many of the same improvements in ChatGPT. The GPT-4.1 models have been trained on data with a knowledge cutoff of June 2024, and are said to outperform OpenAI’s top models at coding, effectively following complex instructions, and understanding large datasets. If developers want their AI applications to access more current information, they’ll need to connect the model to the internet. The GPT-4.1 models will also be able to process up to one million tokens of context at a time, significantly more than models from Google and OpenAI rival Anthropic. OpenAI says that the new GPT-4.1 models have been optimized to help software developers in coding and programming. The company adds that models will assist in front-end coding, will make fewer extraneous edits to code, and follow formats and structures more reliably. AI-assisted coding is one of the most common use cases for generative AI, and has been a focus for Anthropic. According to OpenAI, the GPT-4.1 model is 26 percent less expensive than GPT-4o when answering questions or performing tasks, but the real cost efficiencies can be found in GPT-4.1 nano. The nano model is said to be OpenAI’s fastest, cheapest model ever released, which could make it an attractive option for small businesses looking to dip a toe in the AI waters without making a big financial investment. The company also announced that the GPT-4.1 models would replace GPT-4.5 Preview, a model released in February that had high levels of capability in emotional intelligence, but was extremely expensive compared with most AI models. An OpenAI representative said that GPT-4.1 offers “improved or similar performance on many key capabilities at lower latency and cost,” making GPT-4.5 unnecessary. BY BEN SHERRY @BENLUCASSHERRY

Venture Capital Has Never Been This Obsessed With AI, New Data Shows

U.S. venture capital is becoming increasingly focused on a select cohort of investment prospects, with artificial intelligence the industry’s clear priority, data released today by the research firm PitchBook shows. The first quarter of 2025 saw $91.5 billion in U.S. venture capital activity spread out across an estimated 3,990 deals, PitchBook’s new data indicates. That’s a massive increase in allocated capital compared with Q1 of 2024—up by more than double from $42.4 billion—despite a (very slightly) smaller number of deals, down from 3,995 year-over-year. Notably, a big chunk of the Q1 2025 money came through in a single massive outlier deal: the $40 billion, SoftBank-led funding round that OpenAI just closed. It’s a world of haves and have nots, says Kyle Stanford, PitchBook’s director of American venture research. “The U.S. market has become very bifurcated between a handful of companies able to raise an endless amount of money and the rest of the market, which continues to struggle through a capital shortage,” Stanford says. The primary culprit? Artificial intelligence, says Stanford—a sector that remains beloved by VCs. “Seventy-one percent of total deal value in the U.S. went to AI investments,” Stanford says. “That amount is highly biased with OpenAI’s $40 billion round. Though excluding that deal, AI still captured 48.5 percent of the total invested during the quarter on one-third of completed deals.” No quarter since at least 2015 has seen AI so totally dominate venture capitalists’ checkbooks. This isn’t a new phenomenon, as 2024 saw a rise in VC deal values and deal volume compared with 2023, with much of that venture activity coming from AI deals. Several of those investments included established AI powerhouses such as OpenAI, xAI, and Anthropic. The gold rush of AI deals provided “a false sense of growth,” PitchBook warned at the time—a trend which does not appear to have waned in the subsequent three months. Meanwhile, the exit environment is also still well below its Covid-era highs, meaning investors have less liquidity to pump back into new investments. “Exit activity showed signs of excitement in Q1 with the high-profile IPO of CoreWeave, the announcement of a $32 billion acquisition of Wiz (yet to be completed), and several other well-known brand IPO filings,” says Stanford. “However, outside of those few transactions, the liquidity market remained subdued. Just 12 companies completed public listings, and liquidity worries abound within the market.” BY BRIAN CONTRERAS @_B_CONTRERAS_

Google Claims Its New Gemini AI Is Smarter and Cheaper Than OpenAI’s Best

On April 4, Google made its most powerful AI model yet, Gemini 2.5 Pro, available to software developers. The new model outperforms offerings from OpenAI and Anthropic across several benchmarks while being cheaper to use. Google first revealed Gemini 2.5 Pro in late March, the first in a line of Gemini 2.5 models. You can expect other models in the 2.5 line to have names that denote their price and capabilities, similar to Gemini 2.0 Flash, a lower-cost model revealed in February. All models in the 2.5 line will be thinking models, meaning they can reason through the best way to answer a question by using an internal dialogue. Gemini 2.5 Pro was initially only available on the Gemini app and website, but is now available for commercial use through an API. According to Google, Gemini 2.5 Pro exhibits high levels of capability in math and science. The model outperformed OpenAI’s, Anthropic’s, xAI’s, and DeepSeek’s latest models in benchmarks that test high-level science and math. And in a benchmark meant to test the model’s agentic coding skills, Gemini 2.5 Pro came in second, behind only Anthropic’s Claude 3.7 Sonnet. In brief, 2.5 Pro seems to be a whiz at those subjects. In a blog post announcing Gemini 2.5 Pro’s API debut, Google senior product manager Logan Kilpatrick wrote that the model had been “priced competitively.” Like most other AI APIs, developers will need to pay a fee to Google every time the model processes a new input and creates a new output. This is done through a process called “tokenization,” in which input data is broken up into a series of “tokens,” to be processed by the model. The number of tokens in an input/output determines the size of the API fee. Essentially, more data means more tokens, which means more money. Google lists the precise pricing scheme in the blog post. Developers can also “ground” Gemini 2.5 Pro’s outputs with Google search, enabling the model to access information from across the internet, instead of being restricted to its training data. They’ll get 1,500 free searches every day, but will have to pay $35 for every thousand searches after that. How does all this compare with the competition? OpenAI’s current flagship model, GPT-4.5, is much more expensive. Gemini 2.5 Pro is also slightly cheaper than GPT-4o, OpenAI’s most popular model. And Gemini 2.5 Pro is cheaper than Anthropic’s latest model, Claude 3.7 Sonnet. Ultimately, if you’re building an AI agent that needs to be able to quickly and efficiently search through the internet, like an assistant for buying plane tickets, Gemini 2.5 Pro’s capabilities and integration with Google could make it an attractive option. And if you just want to explore Google’s AI offerings in general, downloading the Gemini app will let you chat with Gemini for free. BY BEN SHERRY @BENLUCASSHERRY