Top News | Meta builds an AI CEO: According to the Financial Times, Meta is developing a realistic AI version of CEO Mark Zuckerberg, trained on not just his public statements but also his tone, speech patterns, thought processes, and personality. Just a few weeks ago, WSJ also reported on Zuckerberg’s plans to create a “CEO Agent” that can help him prioritize and independently clear tasks off of his plate. Now we’ve got more details on the project, part of Meta’s overall attempt to develop “personal superintelligence,” along with photo-realistic 3D AI avatars that interact with users in real time. Zuck apparently also hopes that his new digital persona will help him connect better with his employees, who can now ask “him” questions and solicit direct feedback. So would you want a realistic simulation of your boss to whom you could share thoughts and queries? Or is one supervisor enough? Wayback Machine hits hard times: The Internet Archive project allows individuals to access snapshots of earlier versions of webpages, preserving information and data that would otherwise be lost to history when sites go offline or domains get updated. But Wired reports that a number of major media platforms, publishers, and news organizations — including Gannet’s USA Today and Reddit — are blocking the web crawler that’s used to populate the Wayback Machine. A collective of more than 100 journalists and researchers published a letter in support of the project; signatories include cable news mainstay Rachel Maddow and reporter Taylor Lorenz. Japanese firms unite for AI project: Four major Japanese corporations and three banks have joined forces on a new AI initiative, working together to develop a domestic, frontier Japanese AI model. SoftBank, NEC, Sony Group, and Honda Motor Co. are all founding members of the project, along with “mega-banks” MUFG, Sumitomo Mitsui, and Mizuho. SoftBank and NEC will focus on developing the proprietary AI technologies, while Honda, Sony, and others will work on ways to integrate the new technology into their businesses and workflows. Asahi Shimbun reports that the group plans to develop Physical AI applications as well, such as robotics projects. When we were in Tokyo in January for TWiST Japan, the country’s lack of domestic frontier models — and the potential that they could permanently fall behind, and be forced to rely on less secure models from other countries — was a frequently raised topic of concern.
| TWiST 500 | Of course, we write about OpenAI all the time in this space. Often, the conversation focuses on its mercurial leader, Sam Altman. But this week, we’re turning to more of a behind the scenes player, Chief Revenue Officer Denise Dresser. On Sunday, she sent a memo to employees laying out the company’s latest strategic direction, as they pivot from a “something for everyone” approach to more of a specific focus on enterprise customers. | Dresser explains that OpenAI is seeking a wider moat around its AI products, to prevent users from jumping between OpenAI models and competing products from high-profile rivals (like say… ANTHROPIC!) Dresser stressed that “multi-product adoption makes us harder to replace,” and encouraged staffers to think of OpenAI as a single platform with a suite of helpful tools and “entry points,” rather than “a company with separate product lines.” | More controversially, Dresser makes the case to her own team that Anthropic has been artificially inflating their widely-reported $30 billion in run-rate revenue. She argues that Anthropic’s special accounting “makes revenue look bigger than it is, including grossing up rev share with Amazon and Google.” | When Anthropic models are sold through a platform like Amazon Bedrock or Google Cloud Vertex AI, a customer plays the cloud provider directly, and then a portion of this revenue is shared with Anthropic. The company uses a “gross” accounting approach, which books the full customer-facing price of this transaction as their revenue. They then label the portion of that revenue that stays with Amazon or Google as a revenue share expense. Dresser argues that this makes the top number appear artificially, even deceptively large. | OpenAI, on the other hand, uses a net reporting model. The company only lists the portion of the revenue it actually receives from cloud partners, not the total. So the numbers are smaller but, in Dresser’s view, more accurate in terms of the money reliably coming in from these relationships. | Dresser also accuses Anthropic of making a “strategic misstep” by not acquiring enough compute. Anthropic declined to comment on the memo, but it clearly wasn’t aimed at them anyway. If OpenAI staffers got sufficiently encouraged by the missive, it did its job. - Lon | A message from Squarespace | Turn your idea into a beautiful website! Go to http://www.squarespace.com/twist for a free trial. When you’re ready to launch, use offer code TWIST to save 10% off your first purchase of a website or domain. | This Week in Startups | E2274: It’s the (alleged) rug pull threatening to tear the Bittensor community apart! Jason and Alex are discussing the highly-visible Templar subnet, which had trained the massive Convenant-72B AI model using an innovative, decentralized approach. But when CEO Sam Dare (allegedly!) sold off $10 million of subnet tokens, he threatened the project’s future, and caused the largest single-day drop in the price of TAO ever. What really happened? Who’s in the wrong? And how can the Bittensor ecosystem prevent this from happening again? We’re digging in on a new TWiST. | E2273: Anthropic’s Mythos is so dangerous… you’re NOT ALLOWED to use it! Jason and Alex explore the potential for Project Glasswing, and whether this modern-day Manhattan Project can keep us safe from the cyberthreats to come, along with guest AI expert and Neurometric CEO Rob May. Plus we’re checking out Death By Clawd — the website that tells you how long your startup has to live before it’s Claude-shotted — along with creator, Gyani aka “The G-Man.” | E2272: It’s an OpenClaw all-expert panel on TWiST, with three of our favorite developers and vibecoders: Ryan Carson, Alex Finn, and Yazin Alirhayim. They’re talking about Anthropic cutting off Claude subscriptions from OpenClaw, checking out Google’s Gemma 4 release, and demoing their latest projects: ClawChief, Agent Henry, and Sidecast. | TWiST Partner Offers | Sentry: Your team should be focused on shipping features — not chasing down bugs. New users can get $240 in free credits when they go to sentry.io/twist and use the code TWIST Deel: Founders scale faster on Deel. Set up payroll for any country in minutes, hire anyone anywhere, get visas handled fast, and get back to building. Visit deel.com/twist to learn more. NetSuite: The business landscape is very chaotic right now. That’s why you need NetSuite, by Oracle. Get the free business guide Demystifying AI at Netsuite.com/TWiST
| LAUNCH Accelerator 36! | Our 36th LAUNCH Accelerator cohort has kicked off. If you’re an early stage investor and want to take a look at this cohort first-hand during an exclusive pitch session, please email bianca@launch.co! | Our Favorite Tool | If your work depends on conversations — interviews, meetings, calls — you need a Plaud NotePin. You can check it out at Plaud.ai/twist and use code TWIST for 10% off! | SF Live-Work Space Now Available | Need a flexible living and working environment in San Francisco? This thoughtfully designed loft-style residence at 787 Bryant St., the heart of the vibrant SOMA district and the city’s creative hub, is now available for rent or purchase. Check the listing for more details. | The TWiST500 newsletter is the new, updated, and improved TWiST Ticker. |
|
