DeepSeek’s recent decision to drastically cut pricing on its V4-Pro model by 75% should have been unequivocally good news for enterprise AI vendors and developers. Instead, many are discovering that cheaper models don’t automatically translate into healthier margins.
The reason is simple: While inference costs plummet, agent systems are voraciously consuming tokens faster than prices are declining. For the last 2 decades, software economics was dictated by the same rule. Infra became cheaper every year whereas applications became more capable. AI was initially hypothesized to follow the same pattern. As frontier models improved and token prices dropped, many assumed inference would become a negligible operating expense.That assumption has begun crumbling exponentially.
A chatbot usually turns one user question into one model call. An agent turns it into a chain of planning, retrieval, tool use, verification, summarization, and follow-up decisions. The user sees one answer. The vendor pays for the loop. That is the 100x problem: The same user-visible request can cost a lot more to serve as an agentic workflow than as a chatbot or retrieval-augmented generation (RAG) response. In longer-running workflows, the multiplier is higher. Falling model prices help, but they do not fix a product architecture that turns one prompt into dozens of billable operations.
The scale of what is now at stake is clear in how model providers themselves are pricing developer relationships. OpenAI’s proposed program to give every Y Combinator startup $2 million in API credits — a number that would have funded an entire seed round in any prior tech cycle, and when the same cohort got by on a few thousand dollars of AWS credits — is less a recruiting perk than an admission of what it now costs to run an AI-native company through its first year of product. For established enterprises retrofitting agents into existing product lines, the absolute numbers are larger still.
In a single-turn chatbot, one user message produces roughly one model call. Input-to-billed ratio is about 1:5.
In a multi-step agent rolled out across customer support, sales operations, finance, legal review, and engineering, that ratio routinely lands at 1:700 or higher. Every loop iteration carries forward the cumulative conversation, tool outputs, and reasoning traces. Each step appends; nothing is dropped.
A “simple” agent query like “What did our top customer ask about last week?” typically touches seven priced operations before returning an answer:
User prompt (~50 tokens)
System prompt and tool definitions (~3,000 tokens, repeated on every call)
Retrieval (~5,000 tokens of context)
Model call #1 — tool selection (8,000 in / 200 out)
Tool execution (~4,000 tokens returned)
Model call #2 — summarization (12,000 in / 400 out)
Model call #3 — follow-up decision (12,400 in / 100 out)
One sentence in, roughly 35,000 input tokens billed. Somewhere between $0.10 and $0.40 per query on a frontier model. Multiply that by a million queries a month — the table-stakes volume for any enterprise B2B feature — and the line item is six figures.
The dominant pricing story for enterprise AI has been seat-based SaaS: Pay per-user per-month, deliver agent capability, capture margin. That model assumes a reasonably bounded cost-per-user.
Token amplification breaks the assumption. A power user running 50 agent invocations a day on a $40/seat plan can cost more in inference than the plan charges. Token amplification shatters the traditional SaaS pricing model. When a power user’s daily agent activity costs more in inference than their monthly subscription fee, vendor gross margins turn negative, a paradox that compounds as customers deepen their agent adoption, the very usage curve vendors are selling to their boards. Several vendors are now privately reporting negative gross margins on heavy users, mirroring recent cloud expenditure reports from the Bessemer ‘Supernova’ cohort, where the correlation between AI-agent adoption and gross margin contraction has moved from a theoretical risk to a primary P&L headwind.
The visible symptoms have started leaking into public coverage. Bloomberg this week documented a widening gap between Salesforce’s Agentforce marketing demos and the capabilities actually shipping to customers. This is the kind of gap that opens predictably when promised functionality is technically possible but uneconomical to serve at the price the seat plan implies. Salesforce is the most-watched case, not a unique one.
“For my team, the cost of compute is far beyond the costs of the employees.” — Bryan Catanzaro, VP of Applied Deep Learning, Nvidia
The strategic implication is not “AI is expensive.” It is that the dominant business model assumed by most AI-native company plans does not survive contact with agentic workloads.
Consider an enterprise software vendor charging $40 per-user per-month for an AI-enabled support assistant. A traditional chatbot might cost only a few cents per user per day in inference, leaving healthy gross margins.
Now replace that chatbot with a fully agentic workflow capable of investigating tickets, querying internal systems, drafting responses, validating outputs, and escalating exceptions. If a heavy user executes 50 to 100 agent requests per day, inference consumption can increase by an order of magnitude. What was once a negligible infrastructure cost becomes a material operating expense.
This creates an unusual dynamic: The customers receiving the most value from the product are often the customers generating the highest inference costs. In extreme cases, vendors can find themselves with their most engaged users contributing the least profit. The result is a growing realization across enterprise software that agent adoption and margin expansion are no longer automatically aligned.
The technical responses are known and converging. They are not novel, but they are critical for survival
Cost-aware routing: This technique involves a small classifier model that decides which tier (Haiku, Sonnet, Opus equivalents) handles each query. Well-tuned routers cut inference bills by around 60% without any degradation in quality
Prompt caching: Anthropic, OpenAI, and Google now offer 75 to 90% discounts on cached prefixes.
Context discipline: You can truncate tool outputs, prune reasoning traces, and cap tool depth to prevent your agent from going down a rabbit hole
Speculative decoding: for self-hosted deployments, this technique guarantees 2 to 3X effective throughput on the same GPUs.
“Organizations using orchestration-led governance report stronger productivity gains — a holistic orchestration layer is associated with six times greater productivity impact than compliance‑only approaches” — IBM
The companies building this layer well are starting to look less like microservice operators and more like financial trading systems: Every routing decision priced, every path with its own P&L, every tenant on a metered budget.
Four moves separate the companies that will still have margin in 24 months from the ones that won’t:
Make inference cost a first-class metric. Track it per-feature, per-tenant, per-query class the same way cloud cost was tracked starting in the mid-2010s.
Budget like a media buyer. Set cost-per-thousand-queries ceilings per feature. Cap them. Alert on overruns. Engineering will not enforce this on its own.
Treat the router as core infrastructure, not an optimization. It is the new load balancer.
Audit prompts quarterly. A 4,000-token system prompt that grew organically over six months is a six-figure bill in slow motion. Most teams have never read their own production prompts end to end.
Negotiate volume commits early. Frontier-model vendors now offer reserved-instance-style prepaid commits at substantial discounts. List price is the worst price any enterprise will ever pay.
The structural shift underneath agentic AI is not that it is expensive. As DeepSeek’s price cut today underscores, frontier inference unit costs are dropping roughly 3X per year, and the curve is not slowing.
The shift is that amplification is outrunning the price cuts. Cutting per-token costs 75% does not help a company whose agents are doing 700X more tokens per user query than its pricing model assumed. For the first time since the cloud era began, architecture decisions are again financial decisions in real time. A prompt redesign is a margin event. A poorly bound agent loop is an outage with a credit card attached.
The companies that survive the next 24 months of AI infrastructure pricing will not be the ones running the cheapest model. They will be the ones whose agents are smart and know what they cost to think.
That is the 100X problem. And it is arriving faster than the price cuts can hide it.
Maitreyi Chatterjee is a senior software engineer at a big tech company.
Devansh Agarwal works as an ML engineer at a leading tech company.
Welcome to the VentureBeat community!
Our guest posting program is where technical experts share insights and provide neutral, non-vested deep dives on AI, data infrastructure, cybersecurity and other cutting-edge technologies shaping the future of enterprise.
Read more from our guest post program — and check out our guidelines if you’re interested in contributing an article of your own!
California drew more than $335 billion in venture capital funding this year, reports the Los Angeles Times, citing data released Thursday by PitchBook on private market funding:
Its next biggest competitor, New York, raised less than a tenth of California’s total. Texas raised 1/40th of the amount… Although a campaign for a new tax on billionaires has convinced some ultra-rich residents to shift to other states and businesses often complain that high property and energy costs and an anti-business regulatory regime make it too tough to make money in the state, the inability of the top talent, companies and investors in AI to set up elsewhere shows California’s enduring attraction.
The state’s economy grew 5% last year to a record $4.25 trillion, making it larger than every country other than the U.S., China and Germany. It is home to nearly 400 billion-dollar startups — more than any other state, according to CB Insights… Among metropolitan regions, Los Angeles ranked behind only Silicon Valley and New York, which attracted $98 billion and $11.5 billion in venture investment, respectively… Investors poured in nearly $8 billion across 207 deals in the Los Angeles, Long Beach, and Santa Ana metro areas, up 28% from a year earlier, according to PitchBook…
Nearly 90% of invested dollars [in California] went to AI firms, up from last year, when around 65% of new funds were allocated to AI. “If you’re a tech company and you’re not an AI company, you have a very, very difficult opportunity ahead of you to raise capital,” Stanford said.
Highly anticipated: CUDA has become so embedded in high-performance computing that most developers treat it as inseparable from Nvidia hardware. A small London startup is trying to change that by making CUDA code run across different chips without forcing developers to start over. Spectral Compute has built a compiler called SCALE that serves as a drop-in replacement for Nvidia’s NVCC, letting developers run existing CUDA code on other hardware, including AMD GPUs, without rewriting it.
Spectral Compute was founded in 2018 by four engineers with about 60 years of combined HPC optimization experience. The founders say the effort grew out of frustration: while working at an AI firm, they grew tired of the cost of Nvidia GPUs and the poor performance of alternative compilers, which pushed them to build their own solution using LLVM and Clang.
Unlike tools that translate CUDA into another language or operate on already compiled binaries, SCALE works as a compiler in its own right, recompiling CUDA directly for the target hardware. The model follows the way CPU compilers work, where code can run on different architectures and performance differences mostly come from the hardware, not the compiler.
Spectral is working from the assumption that CUDA is here to stay, noting that it accounts for about 80% of HPC code in use today.
“We take the approach that’s industry-standard for CPUs, but apply it to GPUs,” Giulio Malitesta, head of growth at Spectral, told HPCwire. He added that it’s “the same approach that enables C++ to run, for example, on AMD and ARM CPUs, where nobody expects a performance gap that isn’t directly caused by differences in the underlying hardware.”
Spectral is working from the assumption that CUDA is here to stay, noting that it accounts for about 80% of HPC code in use today. “CUDA is basically the de-facto standard of HPC,” Malitesta said. “We need to accept that as a fact and just do the work as compiler engineers to make it available on different platforms that are not necessarily Nvidia, but also improve on Nvidia GPUs.”
Several other tools also aim to make CUDA portable, but each has notable limitations. AMD’s HIPIFY converts CUDA code into C++ for its ROCm platform, but it doesn’t fully leverage low-level features such as PTX. Intel’s SYCLomatic migrates about 90% of the code, leaving the remaining 10% for manual cleanup. Tools like ZLUDA work at the binary level, which can hurt performance.
Spectral argues its method avoids those tradeoffs. By recompiling from source and checking results against NVCC outputs, the company says it can preserve accuracy while improving performance. Benchmarks published by Spectral show that SCALE can significantly outperform HIPIFY-based approaches on AMD GPUs, with gains in some cases approaching six times.
So far, the company has focused on AMD hardware but is working toward supporting other AI accelerators, though it hasn’t named them. It also continues to support Nvidia GPUs, where it believes there is still room to improve performance through better compilation.
The broader CUDA ecosystem adds another layer of complexity. There are hundreds of specialized libraries, including cuDNN, cuTENSOR and cuDF, that many applications depend on. Spectral is working to expand support for those, and it plans to roll out PyTorch compatibility to better integrate with common AI workflows.
Even as it works to make CUDA more portable, Spectral says it is not trying to compete directly with Nvidia. The company joined Nvidia’s Inception program in June and says it is working across the industry. “We’re on the good side of Nvidia and we’re on a good side with AMD,” said Ruben van Dongen, head of academic solutions and business development. “Of course, we want to be friends with the entire industry. We are neutral, truly neutral.”
SCALE has been shipping for only about two years, so Spectral does not yet have a long track record. Spectral has around 30 employees and is expanding. It sells the compiler to commercial users while offering it free to academic and nonprofit groups. The software has already been tested on large systems, including the Frontier supercomputer at Oak Ridge National Laboratory.
For the users Spectral is targeting, the appeal is mostly practical. Rewriting large CUDA codebases for different hardware is time-consuming and resource-intensive. Spectral is pitching a simpler path. “Especially in the field of research, the researchers lack time,” van Dongen said. “Instead of having to rewrite the entire code base or port away from their current existing code base, they can just recompile with our solution and even increase performance benefits.”
Spectral is stepping into a market dominated by Nvidia at a time when demand for GPU and AI infrastructure is rising quickly. Spectral’s approach hinges on a simple idea: keep CUDA as the standard, but break its dependence on a single vendor’s hardware.
Attitudes towards AI differ by country, gender, profession, age, and political affiliation. A few of those gaps are startling. This article is chock-full of stats. Read it for the surprises, or glance at the bar graph below for a quick overview.
Let’s start with geography, the widest split of all. Ask people in China whether they trust AI and, Edelman finds, nearly nine in 10 say yes; ask Americans and barely a third do. The same chasm shows up, in the Stanford AI Index, on the larger question of whether AI’s benefits outweigh its drawbacks, where most Chinese say it’s good stuff and most Americans have their doubts.
Here’s a possible explanation. Where economies are young and growing fast, AI reads as a ladder up; where they are mature, it reads as a threat to jobs and more. Trust in AI seems to track two things, confidence in institutions and the expectation of personal gain, and both run higher in many Asian countries than in a wary West.
In the U.S., men are about twice as likely as women to expect AI to be good for society, Pew finds, and the gap is wider still among the researchers who build it. The tempting explanation, that women use the tools less, no longer holds: over the past two years women have drawn even with men in using chatbots, yet they trust them less. Women are also likelier to say AI is moving too fast.
Adults under 50 reach for ChatGPT at twice the rate of their elders, Pew reports, yet it is the under-30s who are most convinced it will be bad for society. Here, familiarity breeds unease, and for a concrete reason: the young are not only the heaviest users but the most exposed. AI may be coming first for the entry-level jobs they are trying to land, and they sense it, with Gen Z likelier than any older group to expect it to cut into their job prospects, per the Harris Poll.
Among the AI researchers surveyed, most expect the technology to help the country over the next two decades, Pew’s survey shows; among the public, fewer than one in five do. Some of that is knowledge, since the experts grasp what the systems can and cannot do and fear the lurid scenarios less.
Of course, the people who design AI have their careers and fortunes riding on its success, while the people who answer phones or drive trucks see mainly the threat to their own. The same pattern runs across industries, from technology workers who welcome AI on the job to transportation workers who oppose it. As per Miles’ Law, where you stand depends on where you sit.
The last divide is one that’s moved in recent years, and it’s moved fast. Two years ago Republicans were the AI skeptics; Democrats have since caught up and passed them. Today, just over half of Republicans now trust Washington to regulate AI; barely a third of Democrats do, Pew finds.
AI companies are now more admired on the right than the left, a Harris Poll shows. Democrats are cooling on companies they once cheered, and Republicans are warming to a boom their side now champions. That said, in both parties more people worry that regulation will do too little than too much; what they split on is whom they trust to do the reining.
Despite some loud voices, there is no single verdict on AI. Optimism comes from those with the most to gain, in the rising economies and inside the labs; doubts rise from those with the most to lose or the most to fear. Whatever AI turns out to be, it is being built by the people most enthusiastic about it, for a public that is not.
More than three years after ChatGPT’s launch brought generative AI into the mainstream, OpenAI is broadening its focus beyond individual users to families.
OpenAI is hiring a dedicated product manager in San Francisco to build experiences for families, caregivers, and older adults across its products. The role calls for experience building products for parents and families, and other trust-sensitive consumer experiences, according to the job posting.
The hiring comes as ChatGPT’s audience continues to broaden beyond younger users. According to Sensor Tower estimates shared exclusively with TechCrunch, the share of ChatGPT users aged 35 and older globally rose to 31% in Q2 from 26% a year earlier, while the share of users aged 18 to 24 fell to 29% from 34%. In the U.S., nearly one in four smartphone users who are parents used ChatGPT during the quarter, up from 16% a year earlier, the firm estimates.
OpenAI did not respond to requests for comment about the job posting.
A dedicated product role focused on families signals that OpenAI is beginning to think about its products less as tools for individual productivity and more as technology designed for households, said Ben Bajarin, chief executive of technology consultancy Creative Strategies.
“This is similar to the path Google, Apple, and Meta eventually followed as their platforms became embedded in everyday life, but AI raises the stakes because the assistant is not just mediating content or devices,” he told TechCrunch.
That shift also brings new trust and safety challenges. Stephen Balkam, chief executive of the Family Online Safety Institute, said the hiring reflects both the maturation of OpenAI and a growing recognition that AI products used by children and teenagers require different safeguards than those designed for adults.
“I see this as safety by redesign,” Balkam told TechCrunch. “You take the initial product or service that was released… not really with kids in mind… so this is a much-needed reaction and response.”
The comments come as new research published this week by the Family Online Safety Institute found that parents are underestimating how often their children use generative AI. While 27% of U.S. parents said their child had used generative AI in the past week, 38% of children reported doing so themselves, according to the survey of more than 4,000 families in the United States and Australia.
Balkam told TechCrunch that AI companies should build products differently for younger users, with stronger content controls, age-appropriate experiences, parental oversight, and reminders to inform users that they are interacting with an AI — and not a human.
The hiring also comes amid growing scrutiny of how AI companies protect younger users. OpenAI has faced multiple lawsuits from parents alleging that ChatGPT contributed to harm suffered by their children, including in cases involving suicide.
In response to some of those concerns, OpenAI has introduced a series of safety measures over the past year, including parental controls for teen accounts, routing sensitive conversations to reasoning models designed to better handle signs of distress, and, more recently, an optional “Trusted Contact” feature that can alert a family member or caregiver in cases of potential self-harm.
AI companies, Balkam said, have an opportunity to avoid the mistakes made by social media platforms, which for years treated children much like adults before adding stronger safeguards amid mounting public pressure and regulatory scrutiny.
The hiring also aligns with OpenAI’s broader efforts around families. In a recent workshop organized with the San Antonio Spurs Community Impact organization and the Positive Coaching Alliance, the company said it aimed to explore AI’s role in learning, coaching, and youth engagement.
That said, the demographic shift is not unique to ChatGPT, though OpenAI’s audience is changing in some distinct ways.
Sensor Tower estimates that users aged 25 to 34 account for 40% of the global app audiences for Anthropic’s Claude and Google’s Gemini, matching ChatGPT, compared with 33% for Microsoft’s Copilot. Copilot, however, skews older, with 20% of its users aged 45 and above, compared with 14% for Claude, 12% for Gemini, and 11% for ChatGPT.
While ChatGPT remains relatively underpenetrated among older users, it is adding them faster than its rivals. The share of users aged 45 and above rose three percentage points year-over-year in the second quarter, compared with a two-point increase for Copilot and declines for Claude and Gemini, according to Sensor Tower.
Among U.S. smartphone users who are parents, Gemini had the widest reach at 32% in Q2, followed by ChatGPT at 24%, Claude at 4%, and Copilot at 2%.
For Bajarin, OpenAI’s decision to hire a product manager focused on families signals where consumer AI is headed. As AI becomes a technology shared across generations, he expects companies to roll out family plans, child and teen profiles, caregiver tools, shared household memory, AI tutoring, and stronger safety controls.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Looking for the most recent Strands answer? Click here for our daily Strands hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle, Connections and Connections: Sports Edition puzzles.
Today’s NYT Strands puzzle was pretty tricky. The answers all sound like they could be in a Pepsi ad. Some of the answers are difficult to unscramble, so if you need hints and answers, read on.
I go into depth about the rules for Strands in this story.
If you’re looking for today’s Wordle, Connections and Mini Crossword answers, you can visit CNET’s NYT puzzle hints page.
Read more: NYT Connections Turns 1: These Are the 5 Toughest Puzzles So Far
Today’s Strands theme is: Just what I needed
If that doesn’t help you, here’s a clue: Ahhh!
Your goal is to find hidden words that fit the puzzle’s theme. If you’re stuck, find any words you can. Every time you find three words of four letters or more, Strands will reveal one of the theme words. These are the words I used to get those hints but any words of four or more letters that you find will work:
These are the answers that tie into the theme. The goal of the puzzle is to find them all, including the spangram, a theme word that reaches from one side of the puzzle to the other. When you have all of them (I originally thought there were always eight but learned that the number can vary), every letter on the board will be used. Here are the nonspangram answers:
The completed NYT Strands puzzle for July 13, 2026.
Today’s Strands spangram is HITSTHESPOT. To find it, start with the H that is five letters down on the far-left row, and wind up and over.
Netflix executives have reportedly discussed adding always-on, genre-based live channels and folding rival subscriptions such as Peacock into Netflix as billed add-on tiles, per a Wall Street Journal report. Neither is confirmed; both are internal discussions with no launch date or pricing. The logic is engagement and unskippable ad inventory, at a moment when free ad-supported rivals are capturing casual viewing and Netflix is defending an addictive-design lawsuit it disputes.
TL;DR
Netflix executives have reportedly discussed adding always-on live channels to the service. The channels would run genre-based programming around the clock, all comedies or all action films, according to a Wall Street Journal report relayed by The Verge.
They have also discussed folding rival streaming subscriptions into Netflix itself. Peacock was named specifically, appearing as a tile on the Netflix home page and billed through Netflix.
Neither is a product. Both are internal conversations, with no launch date, pricing, or confirmation from the company.
The comparison writes itself. A grid of always-on genre channels bundled with other people’s services, billed on one invoice, is a fairly precise description of the thing Netflix spent two decades dismantling.
The bundling half is not novel. Amazon Prime Video and Apple TV+ have sold third-party subscriptions as add-ons for years, and it is a reliable way to take a cut of someone else’s revenue.
The channels half is more interesting, because it addresses a problem Netflix built itself.
Netflix made browsing exhausting. The library grew so large that picking something became the hardest part of the evening, and the company has already turned to generative AI to fix the content overload it helped create.
A channel solves that instantly. It removes the decision entirely, which is precisely why linear television survived as long as it did.
It is also a defensive move. Free ad-supported services like Pluto TV and Tubi have captured exactly this casual, low-effort viewing, and Paramount is consolidating Pluto TV onto a unified stack to press the advantage.
Follow the inventory. A channel that never stops is a stream of unskippable ad slots, which a browse-and-pick catalogue is not.
That matters more than it used to. Netflix has been chasing attention wherever it lives, striking short-form video deals with publishers to compete with YouTube and TikTok.
The financial context is not comfortable either. The company authorised a $25bn share buyback after its stock fell 10%, which is what a business does when it wants to reassure people.
Always-on channels are, by design, engagement machinery. They exist to keep the screen on after the thing you chose has finished, and to make stopping require an act of will.
Netflix is currently defending a lawsuit from the Texas attorney general alleging addictive design and improper data collection, claims the company disputes and which remain unproven. Shipping a feature explicitly built to reduce the friction of not stopping is, at minimum, awkward timing.
None of this makes the idea bad. Removing decision fatigue is a real service, and plenty of people want the television equivalent of a radio station.
But it is worth naming the trade. Netflix disrupted cable by giving viewers control, and it may be about to discover that control was never what most viewers wanted, only what they said they wanted.
Off a CRT television, the footage captures Mario mid-jump in Bob-omb Battlefield. The camera follows him across familiar terrain. No emulator sits between the console and the screen. This moment marks real hardware execution after roughly 3000 separate builds and months of relentless iteration by one dedicated developer. The project began as a fork of earlier work by malucard, who first adapted the open-source Super Mario 64 decompilation for PlayStation 1 targets.
Early versions showed some promise, but they were rough around the edges, with trees floating around, animations stalling or breaking, and the camera refusing to act properly. In several instances, performance was sluggish. Still, these early builds demonstrated that the concept could be made to work, even if it wasn’t remotely playable on a console; however, that’s where Eyepatch Entertainment stepped in, and they decided to take on the challenge of getting it stable and running smoothly on a PlayStation’s far-from-generous 2MB of RAM.
Super Mario 64, of course, was initially intended for the Nintendo 64, which had more memory and rasterization processors. In contrast, the PlayStation ran a different CPU, had a simpler graphics system, and used a disc rather than a cartridge. Every image, procedure, and byte of data had to be rethought to function inside the PlayStation’s limited memory space to prevent the game from crashing all the time.
The first round of work was focused on getting the game to boot properly and not run out of memory. That’s a significant issue, and levels like Bob-omb Battlefield were practically impossible to work on owing to allocation concerns. The game occasionally crashed, and when it did, it was limited to a few frames per second. You have to fix the CD image handling to remedy the sector alignment and controller mode issues that were causing texture corruption. It was difficult to debug at first, simply comparing screenshots to what we expected to happen, but after he got GDB and backtracing using the PCSX-Redux emulator, things improved significantly.
Then, in a big breakthrough, he successfully got the game to run on an actual PlayStation. Now, I understand that sounds simple, but getting it to do anything close to what we saw in the emulator was a huge feat. Of course, it wasn’t all smooth sailing; a few additional concerns surfaced. For example, because of the way the framebuffer was flipped, Mario’s model was allowing in light from behind, causing the game to drop frames and freeze. So he fixed that and a few other bugs, and the game began to take shape. He changed the topography to just build the necessary level detail and simplified elements of the 3D models, such as Mario’s legs. This made a big impact in memory and performance.
Audio, oh boy, was a whole other ballgame. On an actual PlayStation, he would occasionally hear stillness or a lot of crackling. The music was encoded as 8-bit ADPCM, which worked flawlessly on the emulator, however the real device required 4-bit XA-ADPCM. We had to re-encode the entire audio. We changed the sector interleaving so that the PlayStation could read it properly, and we even included additional automated checks to ensure that the same issue did not occur again.
The good news is that the most recent builds are beginning to resemble what he envisioned. The skybox, with its flowing clouds and ocean backdrop, is operating perfectly. The Peach sequence is playing out exactly how he had imagined. After very rigorous tuning, Bob-omb Battlefield now runs at a consistent 20 to 28 frames per second. The Chain Chomp is loading properly, animating correctly, and even demolishing its gate on time. Even the door and painting transitions function without crashing the game, and the castle interior runs nicely once we set the frame rate to 30 FPS to avoid timing difficulties.
Welcome back to TechCrunch Mobility, your hub for the future of transportation and now, more than ever, how AI is playing a part. To get this in your inbox, sign up here for free — just click TechCrunch Mobility!
I am back from vacation. What did I miss? Turns out, quite a lot — including the end of the Uber-Waymo partnership in Phoenix. Uber and Waymo still have robotaxi service partnerships in Atlanta and Austin. The question is not if, but when will these agreements end? But that isn’t the most intriguing question, in my opinion. I am far more intrigued by how these two companies will behave once the remaining partnerships end.
There is already tension with Uber executives taking not-so-subtle shots at Waymo. I expect that once the partnerships end, these thinly veiled barbs will be replaced with more direct action. One battleground will be policy, specifically markets where robotaxi companies are angling to get access.
This week, we saw another interesting development in the autonomous vehicle industry on the federal stage. National Highway Traffic Safety Administration administrator Jonathan Morrison issued a directive to autonomous vehicle developers, stating that it is unacceptable for their vehicles to interfere with first responders or law enforcement.
The money quote: “Let me be clear: the inability to detect and appropriately respond to such situations represents a functional insufficiency. Emergency scenes are not rare or extreme ‘edge cases.’ As such, NHTSA is today issuing a call to action for AV developers and operators to immediately focus their resources on fixing this issue.”
Morrison’s letter never calls out any one robotaxi company and it was sent to every AV developer listed in the Department of Transportation’s Standing General Order. But it sure seems like Morrison is directing the agency’s ire at Waymo.
A previous TechCrunch investigation found that Waymo — which operates the largest robotaxi fleet in the United States, with vehicles in cities such as Los Angeles, Phoenix, and San Francisco — has had repeated run-ins with first responders. And just this week, San Francisco supervisor Bilal Mahmood said he plans to submit a letter of inquiry to examine how autonomous vehicles affected public transit services and emergency responders following a July 4 fireworks show that resulted in massive gridlock. Local news outlets reported that numerous Waymo robotaxis had to be towed after running out of power during the lengthy traffic jam.
Morrison’s letter has gravitas. But will there be substantive consequences for AV developers? It’s hard to tell at this point. For now, the NHTSA has demanded companies present the agency with “solutions” by the end of the month.
One more news item from the feds. Take a look at the new 2026 Regulatory Plan and Unified Agenda, which was updated last week. It contains a long list of proposed changes to Federal Motor Vehicle Safety Standards (FMVSS) requirements, which govern vehicle design and equipment requirements. These proposed changes could help autonomous vehicle companies like Tesla and Zoox, which are developing vehicles without steering wheels, pedals, or other features required on human-driven cars.
Got a tip for us? Email Kirsten Korosec at kirsten.korosec@techcrunch.com or my Signal at kkorosec.07, or email Sean O’Kane at sean.okane@techcrunch.com.
We usually focus on venture deals, but this week I wanted to highlight Rivian and the sale of 86.25 million Class A common shares priced at $15.50 each (that includes an added 11.25 million in additional shares that underwriters opted to buy).
In all, Rivian said it expects to raise $1.32 billion in new capital. The raise comes at a notable time for the EV maker. The company started delivering its new R2 SUV last month and recently raised its sales forecast for 2026. The company said it now expects to deliver between 65,000 and 70,000 vehicles after outperforming its own expectations in the second quarter due to robust growth quarter-over-quarter in EDV and R1, coupled with the introduction of R2 deliveries.
The company didn’t explain the reason for the raise. But as a reminder, Rivian is not yet profitable and scaling up production of the R2 — or any vehicle for that matter — isn’t cheap!
Other deals that got my attention …
Bidbus, a Los Angeles-based startup that built a digital marketplace where multiple dealers can bid on a car, raised $15 million in a Series A funding round led by Ibex Investors. Mucker Capital, FJ Labs, Motley Fool Ventures, Data Point Capital, Walter Ventures, and the Car Dealership Guy’s Yossi Levi also participated.
Lyft said it plans to acquire Serveo’s bike-share business in Spain. Terms weren’t disclosed, but the ride-hailing company said it is expected to close this year.
TaiSan, a U.K. battery startup, raised £4.65 million in a seed funding round co-led by Eos Advisory and the Midlands Engine Investment Fund II. InnoEnergy, AFI Ventures, EverQuest Capital Partners, Exergon, Heartfelt Ventures, Adeline Arts & Science, Techmind, angel investor François Badelon, and matched funding from Innovate UK also participated.
AssuranceAmerica, a U.S. insurance provider, confirmed a data breach that affected the personal information and driver’s license numbers of 6.9 million people, making it the largest known spill of Americans’ driver’s license information this year.
Beta Technologies, the electric vehicle takeoff and landing developer, completed operational flights conducted under the U.S. Department of Transportation and Federal Aviation Administration’s new eVTOL Integration Pilot Program. The flights covered about 275 nautical miles covering Virginia and Maryland.
Longtime followers of Tesla will remember the heady days when Elon Musk battled various short sellers of the company’s stock. Musk is more polarizing than ever, and one exchange-traded fund creator has found a way to tap into that negative sentiment with two new anti-Elon exchange-traded funds.
GM brand Chevrolet built an all-American EV truck. Senior reporter Tim De Chant asks, Why is nobody buying it?
Manna Aero, the Ireland-based autonomous drone delivery startup, is scaling up in the United States with a factory and operations center in Tulsa, Oklahoma, that it says will employ 1,000 in the next few years.
Slate Auto teamed up with Crayola to offer its EV truck and SUV customers vehicle wraps in five crayon colors. (Reminder: The basic Slate EV vehicle isn’t painted. Instead, it comes in a gray composite material that can be customized with a vehicle wrap. The company has hundreds of options to choose from.)
TechCrunch podcast Build Mode just launched its third season, and it’s a banger. Build Mode is hosted by Isabelle Johannessen, who heads TechCrunch’s Startup Battlefield program. Unlike Equity — the TC podcast I co-host along with Anthony Ha and Sean O’Kane — Build Mode is designed to help early-stage founders.
The new season kicks off with Precursor Ventures founder and managing partner Charles Hudson, who talks about what early-stage founders need to know before raising their first institutional round.
Check it out: The new rules of early-stage fundraising with Charles Hudson.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
For years, Phone Link has felt like that one app everyone knows exists but rarely remembers to open. Microsoft apparently wants to change that. According to a report from Windows Central, the company is working on a major overhaul of how smartphones integrate with Windows 11, making phones feel like a native part of the operating system instead of something users access through a separate app.
One of the biggest changes reportedly involves the Phone Companion panel in the Start menu. Instead of simply showing basic device information, Microsoft is said to be expanding it to display recent phone activity that users can scroll through without opening Phone Link. Hovering over these activities could even reveal additional details, such as an entire message or photo preview.
Microsoft is also testing a brand-new smartphone flyout in the Windows 11 system tray. Whenever a connected phone is nearby, a dedicated phone icon will appear next to the Wi-Fi and battery indicators. Clicking it would open quick controls for features such as Do Not Disturb, vibrate mode, and find phone settings, while also showing battery level and connection status. Perhaps the neatest addition is support for dragging files directly onto the phone icon, instantly transferring them to the connected device.
Microsoft isn’t stopping there. The company is also exploring clipboard history syncing between Windows 11 and smartphones using the native Windows Clipboard feature. While clipboard sync already exists today, it only remembers the last copied item. The new approach would reportedly synchronize an entire clipboard history, allowing users to access a synced list of previously copied text and content across both devices.
Another interesting addition is a dedicated Messages app for Windows 11. Rather than living inside Phone Link, SMS conversations would get their own standalone application that can be pinned to and launched from the Start menu, making texting from a PC feel much more like using a native Windows experience.
According to the report, all of these features are currently being explored and prototyped internally, meaning there’s no guarantee they’ll all ship as described. Microsoft is expected to gather feedback from Windows Insiders before committing to shipping anything concrete into future Windows 11 updates.
If all of this sounds familiar, that’s because Microsoft has been moving in this direction for a while. Windows 11 already lets users browse their phone’s storage directly from File Explorer and even use supported smartphones as wireless webcams. The reported changes build on that foundation by making smartphone features feel less like an add-on and more like they’re baked directly into the Windows UX shell.
The funny thing is, Microsoft spent years trying to convince people to buy Windows Phones. That obviously didn’t work out. Now, instead of fighting Android and the iPhone, it’s embracing them, and honestly, that might be the smarter strategy. If these features arrive as described, Windows 11 could finally make the jump between PC and phone feel almost invisible.
Although we’d rather bring you news of clever modifications and repairs down on the farm, more often than not, the name “John Deere” has appeared on the pages of Hackaday because of their opposition to farmers actually being able to work on the machines their livelihoods depend on. But thanks to a settlement reached between the company and the Federal Trade Commission this week, farmers seem to have been handed a much-needed win in the Right to Repair battle.
When a lawsuit against a company ends in a settlement, it usually means spending money they would rather pay than go to court. Indeed, earlier cases against John Deere have resulted in plenty of checks being written. But this time around, the FTC agreement requires Deere to make its diagnostic and repair software available to owners and independent shops. It also has a clause that prevents them from retaliating against owners who want to handle their own repairs rather than going through the company’s official service channels — hard to believe that’s something that actually needs to be specified, but it does give you a hint as to just how bad the situation has been. We’ll definitely be keeping an eye on this story.
Sounds like the Feds were busy this week, as the Federal Communications Commission also gave the green light to Reflect Orbital to launch a prototype satellite for their controversial “sunlight as a service” concept. The company plans to put the spacecraft into a roughly 600 km orbit around the planet, where it will deploy its 324-square-meter reflector and angle itself to illuminate a spot on the ground. It might sound like something a Bond villain would come up with, but Reflect Orbital says the capability will be used to beam sunlight directly onto solar panels at night and to provide light for search-and-rescue operations.
As you might expect, providing such a service on a global scale would require many such reflectors, which is where the concern really comes in. Critics note that a sky full of literal mirrors can cause all sorts of issues, ranging from the scientific to the scenic. The American Astronomical Society points out that each satellite in the constellation could appear to be four times as bright as the full Moon, and that it’s possible an amateur sky watcher could get an eyeball full of redirected sunlight should one of them unexpectedly zip past the aperture of their backyard telescope.
Moving from 600 km above to 400 meters below the surface of the ocean, the Royal Canadian Geographical Society and Woods Hole Oceanographic Institution have provided our first look at the wreck of Ernest Shackleton’s final ship, Quest. The schooner-rigged steamship was launched in 1917 and had a storied career that included not only a number of scientific expeditions but service during the Second World War. The ship ultimately met its fate in 1962 when it was damaged by ice and sank off the north coast of Labrador. The exact location of the wreck was unknown until its discovery in June of 2024.
Now, before you start questioning your knowledge of history, we should probably clarify that Shackleton was not exploring the Labrador Sea in 1962. He did, however, die aboard Quest in 1922 at the age of 47 as he was preparing to depart on another expedition to the Antarctic.
This next one isn’t new, but it’s the first time we’ve come across this gallery of gorgeous Soviet-era control rooms. Hackaday isn’t the place to dive into the political and socioeconomic aspects of the USSR. All we know is that they were putting out some damn fine-looking control panels back then. Half of them look like they wouldn’t be out of place on a Moon base, and the white lab coats with the little hats really complete the retrofuturism vibe. Now we have to go watch Chernobyl again.
In software news, FreeCAD has received a new tool that we know many in the community will be excited about: Banana For Scale. Forget the confusion between Metric and Imperial measurements. Placing a 3D banana in the scene alongside your rendered part provides a globally recognized size reference. While the free and open-source CAD package has often been criticized for being behind its commercial counterparts in terms of user interface and overall feature set, we think this addition should go a long way toward evening the scales — no pun intended.
Finally, Phoronix reports that Linux 7.2-rc3 includes several vital updates to device drivers for the Sega Dreamcast. Users running Linux on the ill-fated PlayStation 2 competitor will benefit from improvements made to the keyboard, mouse, and joystick interfaces. These fixes join the improved code for the console’s GD-ROM optical drive that emerged back in April. The “Year of the Linux Desktop” continues to be elusive, but it certainly looks like 2026 may finally be the Year of Linux on the Dreamcast.
See something interesting that you think would be a good fit for our weekly Links column? Drop us a line; we’d love to hear about it.
Open Thread: What Great Books Have You Read Recently?
Whats Hidden Inside This Cash Register? #treasure #reselling #money
Loro Piana Fall 2026 Enters Houston’s Art Scene
Weekend Open Thread: Nutriplenish Leave-In Conditioner
Anthropic’s new “J-lens” reveals a silent workspace inside Claude that mirrors a leading theory of consciousness
$1,000 Credit Alert! BlockDAG X Exchange Pre-Registration Now Officially Open, Polkadot Dips & Zcash Rebounds
AXT Shares Jump Nearly 14% as Semiconductor Materials Maker Rebounds on AI-Linked Indium Phosphide Demand
Joshua Pacio ‘more complete’ ahead of ONE rematch vs Malachiev
2026 Genesis Scottish Open Thursday TV coverage: Round 1
Best Time to Enter Small Caps Right Now? Another Bull Run? | Financially Free
Anthropic brings Claude Cowork to mobile and web as usage data shows most users aren’t coding
SK hynix (000660.KS) Stock Dips as $28B Nasdaq ADR Offering Drives AI Memory Expansion
Avoid entering in FOMO #bitcoin #cryptocurrency #trading #scalping
Super Eagles star Moses Simon opens up on Liverpool transfer regret
We have punished the disrespect
Binance lists Strategy’s STRC stock as company expands Bitcoin funding
Character.AI enters the microdrama arena with its own productions, but there’s a twist
9 Best Keyboards (2025), Tested and Reviewed
Enbridge: AI Tailwind Priced In (Rating Downgrade)
“What’s going on?!” Carl Froch discusses Floyd Mayweather Jr financial issues
You must be logged in to post a comment Login