Tesla is broadening its robotaxi offerings to Dallas and Houston, as stated in a company social media update.

The update simply mentions that “Robotaxi is now launching in Dallas & Houston 🤠” and features a 14-second clip depicting Tesla cars operating without human monitors or drivers in the front seats.

The firm now provides robotaxi services in three Texan cities, having initiated operations in Austin last year and commencing driverless rides in January 2026. In a filing from February, Tesla reported that its Austin robotaxis were involved in 14 accidents since their debut.

Additionally, it has a more restricted ride service that employs human drivers in the San Francisco Bay Area.

Tesla might not have a substantial number of vehicles operating in these new markets at this point, with crowdsourced data from the Robotaxi Tracker website only noting one vehicle in each city (compared to 46 active vehicles recorded in Austin).

Veteran venture capitalist Ron Conway announced on Friday that he has been “recently diagnosed with a rare cancer type.”

In a message on X, Conway stated that he “will be stepping back from some of my regular activities,” but he plans to “continue supporting” the founders associated with his firm SV Angel: “With a more concentrated and balanced agenda, I can focus on treatments while assisting SV Angel founders at critical junctures as we always do!”

Conway further mentioned that SV Angel will remain “unchanged,” noting that his son Topher Conway “has been making all of our investment choices for most of the last decade.” He also highlighted that another son, Ronny Conway, became a managing partner in 2024.

“They bring knowledge from nearly every significant technology cycle in Silicon Valley and are currently dedicated to collaborating with founders shaping the future of AI,” Conway remarked.

He added that he is not disclosing “the particular type of cancer” in his diagnosis to avoid “speculation” regarding the prognosis, yet he expressed that he remains “optimistic.”

“I am fortunate to have an amazing team of UCSF doctors in San Francisco, and as you know, I never back down from a battle,” Conway stated.

Cerebras Systems, a startup that CEO Andrew Feldman refers to as “the quickest AI hardware for training and inference,” has initiated the process to become a publicly traded company.

The firm had previously attempted an initial public offering in 2024, but this was postponed due to a federal examination of a funding from G42, an investment firm based in Abu Dhabi, and was ultimately retracted. Last year, Cerebras secured $1.1 billion in a Series G funding round, followed by a $1 billion Series H in February, achieving a valuation of $23 billion, as reported by the Wall Street Journal.

In recent times, the company announced a partnership with Amazon Web Services to implement Cerebras chips within Amazon’s data centers, in addition to a reported $10 billion deal with OpenAI.

In a recent discussion with the WSJ, Feldman stated confidently, “Clearly, [Nvidia] did not want to forfeit the fast inference market at OpenAI, and we successfully acquired that from them.”

According to the filing, Cerebras generated $510 million in revenue in 2025, with a net income of $237.8 million (adjusted for specific one-time items, it resulted in a non-GAAP net loss of $75.7 million).

The company has not revealed the amount it aims to collect from the IPO. A representative indicated that the launch is tentatively scheduled for mid-May.

Even though the Pentagon has recently classified Anthropic as a supply-chain risk, the company is still engaging with senior figures from the Trump administration.

Indications of a warming relationship — or at least a sense that not all factions of the administration aimed to sever ties with Anthropic — surfaced previously, with reports indicating that Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell encouraged leaders of major banks to experiment with Anthropic’s new Mythos model.

Jack Clark, co-founder of Anthropic, appeared to affirm this, stating that the ongoing dispute regarding the supply-chain risk label is a “narrow contracting issue” that would not hinder the company’s readiness to inform the government about its newest models.

On Friday, Axios disclosed that Bessent and White House Chief of Staff Susie Wiles had a meeting with Anthropic CEO Dario Amodei. The White House characterized this as an “introductory meeting” that was “productive and constructive.”

“We talked about collaboration opportunities, as well as common approaches and protocols to tackle the challenges related to scaling this technology,” the White House stated.

In a similar vein, Anthropic released a statement confirming that Amodei had engaged in a “productive discussion with senior administration officials on how Anthropic and the U.S. government can collaborate on crucial shared priorities such as cybersecurity, maintaining America’s leadership in the AI race, and AI safety.”

The company expressed that it’s “looking forward to continuing these conversations.”

The conflict between Anthropic and the Pentagon appears to have initiated after unsuccessful discussions regarding the military’s application of Anthropic’s models; the AI firm aimed to preserve safeguards surrounding the use of its technology for fully autonomous weaponry and extensive domestic surveillance. (OpenAI swiftly revealed a military agreement of its own, triggering some consumer backlash.)

In turn, the Pentagon labeled Anthropic a supply-chain risk — a designation typically reserved for foreign adversaries that could severely restrict the government’s utilization of Anthropic’s models. The company is contesting that label in court. 

Nonetheless, it seems the remainder of the Trump administration does not share the Pentagon’s antagonism, with an administration source informing Axios that “every agency” except the Department of Defense is interested in using the company’s technology.

Many predicted that AI would eliminate apps. However, the number of new app launches is skyrocketing.

As per a recent report from market intelligence firm Appfigures, global app launches in the first quarter of 2026 increased by 60% compared to the previous year across both Apple’s App Store and Google Play. This figure rose to an impressive 80% when considering only the iOS App Store. Thus far in April 2026, the total number of app launches has surged by 104% across both platforms compared to the same period last year, with a nearly 89% increase on iOS.

Greg “Joz” Joswiak, Apple’s Senior Vice President of Worldwide Marketing, humorously remarked in a recent discussion that claims regarding the demise of the App Store in the AI era “may have been greatly exaggerated.”

These insights appear amidst worries that the growth of AI chatbots and agents might lead users to abandon apps – a suggestion already posited by industry figures, such as Nothing’s CEO Carl Pei, who is intent on developing a smartphone for the AI generation. The New York Times highlighted last year the possibility of new computing platforms overtaking smartphones, such as smart glasses, ambient computing gadgets, or redesigned smartwatches with AI functionalities.

OpenAI is reportedly even collaborating with renowned Apple designer Jony Ive on an AI hardware device.

However, there’s another angle to consider: AI may facilitate app creation, leading to a revival of the App Store. This new wave of app development might be spearheaded by creators who have innovative concepts but lack the technical know-how to develop mobile software.

Data from Appfigures shows that some app categories are experiencing more new launches than others.

Mobile games continue to dominate global new app releases as of Q1 2026, consistent with previous years. However, this year, “productivity” apps have entered the top five. The “utilities” category has ascended to the second position, while “lifestyle” apps have climbed from the fifth position last year to the third position now. Finally, “health and fitness” applications complete the top five categories.

The hypothesis suggests that AI-driven tools, such as Claude Code or Replit, could be contributing to this increase in new launches. It appears we may be reaching a tipping point regarding AI usability, enabling individuals to utilize these resources to create their own mobile applications more efficiently — or even embark on building their very first apps.

The influx of new apps awaiting review by Apple may also be implicated in some of the tech giant’s recent errors. This week, Apple removed the rewards app Freecash from the App Store for violating rules, after allowing it to rise in the Top Charts and remain within the top five for several months. Additionally, Apple faced an unexpected threat from a malicious cryptocurrency app, a fake version of Ledger Live, that siphoned $9.5 million in crypto from users’ accounts.

Although significant issues like these can create negative publicity for the App Store, the company still effectively prevents and rejects hazardous or spammy apps. According to Apple’s latest analysis from 2024, the firm removed or denied over 17,000 apps for bait-and-switch violations that year; rejected more than 320,000 app submissions identified as spam, duplicates, or misleading; and acted to block over 37,000 potentially fraudulent apps from being available to users on the App Store.

Nonetheless, Apple commentators like John Gruber have long contended that the App Store requires a vigilant team to monitor for fraudulent or scammy applications that are gaining traction or generating significant revenue.

If AI-assisted coding becomes the catalyst for this recent wave of app launches, the necessity for such vigilance will only increase as more new apps enter the market, not all of which will be harmless.

Jack Zhang, at 34 years old and three and a half years into his startup journey, found himself in a meeting with one of Silicon Valley’s top investors. Michael Moritz from Sequoia had welcomed him into his home — which had, as Zhang remembers, a couple of floors and a view directly facing the Golden Gate Bridge — to discuss a potential sale.

Stripe was interested in acquiring Airwallex for $1.2 billion. At that juncture, the Melbourne-based company was generating about $2 million annually. The figures were hard to ignore: a revenue multiple close to 600. Moritz emphasized that Patrick Collison was a once-in-a-generation founder. The acquisition would evolve into something remarkable. Zhang considered this. He spent two weeks wandering San Francisco, feeling agitated and unable to think clearly. Eventually, he agreed.

He then flew back home, nearly 8,000 miles away.

“I deeply contemplated my motivations for building Airwallex,” he mentioned earlier this week during a conversation with this editor from overseas. “Having been in business for three and a half years, the company was growing exponentially in 2018, and I had only just begun to feel what entrepreneurship was about — and that’s what I had always dreamed of.”

Two of his three co-founders opposed the deal, which contributed to his decision. However, he pointed to the clearest sign: looking at the whiteboard in his office where the vision remained, still incomplete: to create a financial infrastructure enabling any business to operate globally as though it were local.

This choice is becoming increasingly prophetic. Airwallex now reports over $1.3 billion in annualized revenue and is expanding at an 85% year-over-year rate. It handles nearly $300 billion in annualized transaction volume, according to him. Achieving this success hasn’t come without challenges — and Zhang argues that these struggles are fundamental to the journey.

His conviction extends well beyond mere business tactics. Zhang was raised in Qingdao, a northern Chinese port city, relocating to Melbourne at 15, alone and with limited English, staying with a host family. Following a financial crisis in his family, he undertook four jobs to complete a computer science degree at the University of Melbourne, as reported by the Australian Financial Review — bartending, dishwashing, working overnight shifts at a gas station, and fruit-picking during school breaks, which he has described as the toughest job he ever held. Later, he spent years developing trading algorithms at an Australian investment bank, a well-paid position that never felt entirely “fulfilling.”

Prior to Airwallex, he initiated around 10 business ventures: a magazine at 14, a real estate firm, import-export activities transferring wine and olive oil from Australia to Asia, textiles in reverse, and a burger franchise.

While managing a coffee shop in Melbourne, the concept for Airwallex emerged. As he tried to pay coffee bean suppliers in Brazil, Indonesia, and Guatemala, his co-founder Max Li repeatedly saw payments vanish within correspondent banking systems — flagged and halted by American banks enforcing OFAC sanctions, sometimes returning weeks after being sent. “This compelled me to delve into how correspondent banking operates,” Zhang stated, “the functioning of SWIFT, and how we could develop our own global money movement network.”

The core concept remains but has expanded significantly. Airwallex now possesses nearly 90 financial licenses across 50 markets. Zhang estimates Stripe holds perhaps half as many. Acquiring those licenses has been highly labor-intensive — the process in Japan alone took seven years. In some emerging markets, it was necessary to purchase shell companies with licenses no longer issued by central banks and completely rebuild the underlying technology.

“You can’t simply vibe-code an integration with Mexico’s central bank,” Zhang explained. “We require a secure facility — access to the central bank integration necessitates a biometric scan just to enter.”

The rationale for obtaining these licenses goes beyond superficial regulatory compliance. In Japan, for example, Stripe and Square can process payments, but they must instantaneously transfer funds to the merchant’s account. Airwallex, thanks to its fund transfer operator license, is able to retain those funds within its ecosystem. This allows a customer to generate bank accounts, issue cards, and spend money without leaving the platform.

The foreign exchange economics alone are substantial: a U.S. seller settling transactions in Australian dollars avoids the 2% to 3% conversion fee typically charged by processors like Stripe when moving money back to U.S. dollars — and can utilize those local balances to pay local suppliers, manage payroll, and cover digital marketing costs, all at interbank rates.

“You no longer operate merely as a U.S. entity,” Zhang noted. “You function like a company with global entities, but without the necessity of physically establishing those entities.”

This deliberate slow growth was intentional, and Zhang frequently returns to a principle he refers to as the “path of maximum resistance.” Each license, each bank integration, every local payment rail that Airwallex painstakingly assembled has erected a barrier against competition. “It took us six and a half years to hit $100 million in annual recurring revenue,” Zhang stated. “Then, it took just over three years to reach a billion.”

The competitive reasoning, in his perspective, hinges on a fundamental understanding of owning infrastructure versus depending on someone else’s. If you lack control over the full payment process and something goes awry, you can’t access the underlying data to clarify it for your customer. You cannot seamlessly introduce new products on top of another’s system. “Building on existing infrastructure,” he remarked, “is simply not scalable.”

Historically, Airwallex and Stripe have operated in largely distinct regions, targeting different customers. That is evolving. As Stripe ventures deeper into international territories, and as Airwallex makes substantive advancements into the U.S. market, the overlap is increasing.

Previously, the main purchasers of Airwallex were the CFO offices in Australia and Southeast Asia, where it has gained a solid foothold — finance directors, treasury teams — distinguishing it from Stripe, whose customer acquisition has largely stemmed from U.S. developers selecting a default option for a new business. Over 90% of Airwallex’s clients first engage with a business account product, leading to utilizing payments and spend management subsequently. More than half are using multiple products, says Zhang.

Nonetheless, there are obstacles Zhang does not underestimate. The most significant might be Stripe’s status as Silicon Valley’s darling, its privately held shares creating millionaires throughout the tech scene. Additionally, a brand gap persists. Airwallex must embed itself within the mindset of engineers and developers — not merely finance teams — so that founders instinctively turn to it. “Our brand simply isn’t there yet,” he remarked. “It’s a tougher competition to win.”

This competitive landscape is closely monitored from various perspectives. Sequoia supported Airwallex early on — although the deal was brokered through Sequoia Capital China, which has since rebranded as Hongshan — and continues to be one of the company’s largest investors. The firm Greenoaks Capital also has stakes in both companies.

Zhang dismissed any notion of discomfort regarding the overlapping investors. According to him, the investors are wagering on a vast market.

However, this raises the valuation discussion. Stripe was valued at $159 billion in a February tender offer — marking a 74% increase from the previous year — after processing $1.9 trillion in total payment volume in 2025. Meanwhile, Airwallex received an $8 billion valuation in December, making it worth approximately a twentieth of that. But Zhang contends that Stripe’s payment volume is only about six times Airwallex’s, not 20 times. With 85% annual growth and expected revenue of $2 billion within the next year, Airwallex is rapidly narrowing the revenue gap more quickly than the valuation gap may indicate.

Whether the market will eventually take notice remains uncertain — a possibility an IPO, which Zhang suggests is at least three to five years away, would bring to light.

For now, Zhang asserts his focus is on long-term objectives: one million customers by 2030, $20 billion in annual revenue, and increasing average revenue per customer from about $12,000 to around $20,000. A range of AI-driven autonomous finance products — systems that not only present data but also execute transactions — are currently being launched. The premise is that a decade’s worth of financial data across the entire corporate finance spectrum, from revenue collection to treasury management to vendor payments and expenses, has provided a training dataset that cannot be easily replicated by competitors overnight, he suggests.

Now it is a question of whether all this hard work is sufficient to penetrate Stripe’s market share. At present, it seems the competition remains at a distance. Zhang and Collison were never close, but they maintained a friendly rapport during past merger discussions. Last year, they were both present at Greenoaks Capital’s annual event. They did not interact.

At a fashionable location close to the San Francisco pier, Sam Altman’s verification initiative World marked its latest phase and swift growth of its goals.  And it’s commencing with Tinder.

Tools for Humanity (TFH), the organization behind the World initiative, revealed on Friday its intentions to incorporate its verification technology into dating applications, event and concert ticketing platforms, corporate entities, email, and various sectors of public life.

“We are nearing a point of highly advanced AI, which is bringing about many positive changes,” Altman stated before a crowded audience at The Midway. “We are also approaching an era where AI will produce more content than humans,” he continued. “I’m sure many of you have experienced moments where you wonder, ‘Am I interacting with an AI or a human, or how much of each, and how can I tell?”

World (previously known as Worldcoin) sets itself apart from many of its identification verification counterparts by enabling the validation of a genuine, living individual using a digital service while maintaining that person’s privacy. There’s some intricate cryptographic science behind this (referred to as “zero-knowledge proof-based authentication”). The key point: The firm is developing what it refers to as “proof of human” tools, which are systems that can confirm human actions in a landscape saturated with AI agents and bots.

Its primary verification tool is a spherical digital reader labeled the Orb, which scans a user’s eyes and transforms their iris into a unique and anonymous cryptographic identifier (identified as a verified World ID). This ID can then be employed to access World’s offerings, although users may also utilize World’s application without one.

Altman kept his comments succinct on Friday (TFH’s co-founder and CEO, Alex Blania, was unavailable due to a sudden hand surgery, as per Altman). He then handed much of the presentation over to World’s chief product officer, Tiago Sada, and his associates.

Sada detailed that World was unveiling the latest iteration of its application (the previous version was launched during an event in December), alongside an array of new integrations for its technology.

World has been gearing up for some time to launch a verification service for dating apps — most notably, Tinder. Last year, Tinder initiated a World ID pilot project in Japan. That pilot seemingly succeeded, as World announced that Tinder would be introducing its verification integration in international markets — including the U.S. The initiative incorporates a World ID badge into the profiles of users who have undergone its verification procedures, thereby confirming them as a real person.

World is also engaging the entertainment sector by launching a new feature named Concert Kit, which allows musical performers to reserve a specific number of concert tickets for World ID-verified individuals. This is intended to protect fans from scalpers who frequently use automated ticket-buying bots to seize seats. Concert Kit is compatible with major ticketing platforms, such as Ticketmaster and Eventbrite, and the company is promoting it through partnerships with 30 Seconds to Mars and Bruno Mars — both of whom intend to utilize it for their forthcoming tours.

The event included numerous other announcements, including some targeting businesses. A Zoom/World ID verification integration aims to combat a potential deepfake threat during business calls, and a Docusign collaboration is designed to verify that signatures originate from genuine users.

The firm is also developing several features in anticipation of the chaotic landscape of the agentic web, including one termed “agent delegation,” wherein an individual can assign their World ID to an agent for online tasks on their behalf. A collaboration with the authentication firm Okta has also established a system (currently in beta) that confirms an agent is acting on behalf of a human. This system is structured so that a World ID can be linked to a specific agent, and when the agent navigates the web to perform tasks for that person, websites will understand that a verified individual is behind the actions, explained Okta’s chief product officer, Gareth Davies, at the event.

Thus far, World has faced challenges in scaling, primarily due to the verification process itself. For much of the company’s past, obtaining its gold standard required visiting one of its facilities and going through an eyeball scan by an Orb — a rather inconvenient (not to mention peculiar) experience.

Nonetheless, World has consistently made efforts to enhance the convenience and incentive framework for verification. In the past, it provided its cryptocurrency, Worldcoin, to select users who registered and has distributed its Orbs to major retail outlets to enable users to verify themselves while out shopping or enjoying a coffee. Now the company is announcing a significant expansion of its Orb presence in New York, Los Angeles, and San Francisco. The company has also promoted a service where interested users could arrange for World to deliver an Orb to their location for remote verification.

In an interview with TechCrunch, Sada also mentioned that World has sought to address the scaling issue by developing various levels of verification. The top tier is Orb verification, but below that, World has previously introduced a mid-tier option, which uses an anonymized scan of an official government ID through the card’s NFC chip.

The firm also presented a low-tier option, or what Sada described as “low friction”— indicating low effort, and also “low security” — which simply involves taking a selfie.

Selfie Check, which Sada’s team showcased during the event, is structured to uphold user privacy.

“Selfie is designed to be private,” stated Daniel Shorr, one of TFH’s executives, during the presentation. “This means that we optimize the local processing done on your device, on your phone, ensuring that your images remain yours.”

Selfie verification, of course, is not a new concept, and fraudsters have long found ways to deceive it. “Naturally, we exert our utmost effort, and it’s among the best systems you’ll find for this. But it has limitations,” Sada told TechCrunch. Developers wanting to integrate World’s services can select from the three distinct verification levels based on the degree of security they prioritize, he noted.

OpenAI is parting ways with two key figures behind its most ambitious projects. Kevin Weil, who spearheaded the company’s scientific research efforts, and Bill Peebles, the mind behind the AI video tool Sora, both confirmed their exits on Friday. These departures come as OpenAI focuses on enterprise AI and its upcoming “superapp.”

The exits follow OpenAI’s choice to scale back on “side ventures,” including customer-oriented initiatives like Sora and OpenAI for Science. Last month, Sora, which was costing an estimated $1 million daily in computing expenses, was discontinued.

OpenAI for Science was the internal research collective that developed Prism, an AI-enabled platform designed to hasten scientific breakthroughs. It is being integrated into “other research teams,” as stated in Weil’s social media announcement regarding the changes.

“It’s been an enlightening two years, from serving as Chief Product Officer to joining the research team and initiating OpenAI for Science,” Weil remarked. “Propelling science forward will be one of the most remarkably positive effects of our quest for AGI.”

The team faced a brief and tumultuous journey following its official launch in October 2025. Weil retracted a tweet asserting that GPT-5 had resolved 10 previously unsolved Erdős mathematical challenges, but that assertion quickly unraveled when the mathematician managing erdosproblems.com challenged it.

Weil’s exit followed closely behind the release of GPT-Rosalind, a new model aimed at enhancing life sciences research and drug development.

In a social media post announcing his exit, Peebles praised Sora for sparking a “significant wave of investment in video throughout the industry” and contended that the type of research that led to the video tool necessitates autonomy from the main corporate agenda.

“Fostering chaos is the sole path for a research lab to flourish in the long run,” he asserted.

OpenAI is also bidding farewell to Srinivas Narayanan, its chief technology officer for enterprise applications, as reported by Wired. Narayanan allegedly communicated internally that he was leaving to devote more time to his family.

This article was revised to reflect the departure of Srinivas Narayanan.

Nicholas Moore, who admitted to infiltrating the U.S. Supreme Court’s electronic document filing system multiple times over a period of months, received a sentence of a year of probation on Friday.

Moore also breached the network of AmeriCorps, a federal agency that administers volunteer programs with stipends, and the systems of the Department of Veterans Affairs, responsible for healthcare and benefits for military veterans. 

The individual boasted about his activities on an Instagram account titled @ihackedthegovernment, where he shared the personal details of those he had compromised. Moore utilized one of his victim’s credentials to gain access to the U.S. Supreme Court’s electronic document filing system, AmeriCorps, and the Department of Veteran Affairs.

He was confronted with the possibility of a year in prison and a $100,000 penalty for damages. Later, prosecutors requested only probation. 

“I made a mistake,” Moore expressed during the sentencing session on Friday, per The Hill. “I am really sorry. I respect laws, and I aspire to be a good citizen.” 

The AI coding firm Cursor is approaching a new funding round in which the four-year-old organization aims to secure at least $2 billion in additional capital, as reported by four individuals familiar with the situation. Returning investors Thrive and Andreessen Horowitz are anticipated to spearhead the funding at a valuation of $50 billion, before the influx of new capital, according to the sources.

Battery Ventures, a new investor, may also take part in the funding, as noted by two sources. Strategic investor Nvidia is likewise expected to contribute financially, one individual stated.

Although the funding round is already oversubscribed, the terms of the deal are not finalized and could still be adjusted.

If the funding goes through, it would nearly double Cursor’s previous $29.3 billion post-money valuation that was assigned to the company during its last fundraising effort six months prior.

In spite of intense competition from other AI-coding solutions, such as Anthropic’s Claude Code and OpenAI’s upgraded Codex, Cursor’s revenue continues to rise swiftly.

Cursor predicts concluding 2026 with an annualized revenue run rate exceeding $6 billion, according to two individuals. This growth trajectory suggests that the company expects to at least triple its annualized revenue in the coming 10 months. In February, Cursor reported reaching $2 billion in annualized revenue, calculated by extrapolating its most recent monthly sales over the course of a year, as reported by Bloomberg.

Like many AI-coding startups that depend on third-party models, Cursor experienced negative gross margins until recently, indicating that the costs of operating the product exceeded the revenues it could generate. The launch of a proprietary Composer model last November, along with access to cheaper models like China’s Kimi, has enabled the company to achieve marginal gross margin profitability, the sources noted.

On a more detailed level, the organization has achieved positive gross margins on sales to large enterprises, but continues to incur losses on accounts for individual developers, as indicated by one individual.

By reducing reliance on external providers, Cursor aims to prevent being supplanted by its own suppliers, particularly Anthropic, whose Claude Code has become the startup’s primary competitor.

Cursor and Battery Ventures opted not to comment. Thrive, a16z, and Nvidia did not respond to requests for comments.

Cursor, formerly known as Anysphere, was co-founded in 2022 by Michael Truell, Sualeh Asif, Arvid Lunnemark, and Aman Sanger while they were studying at MIT.