BUSINESS

TheStreet aims to feature only the best products and services. If you buy something via one of our links, we may earn a commission.Why we love this dealIf your backyard or front porch is empty, you’re wasting valuable square footage. Transforming your patio space into an outdoor oasis starts with the right furniture. There are countless styles to consider, like bistro sets or a classic Adirondack chair, but we love a matching set that’s perfect for stretching out and taking cat naps in the sun. The MF Studio 3-Piece Rattan Patio Conversation Set fits the bill with a cozy love seat and chaise lounge that can be configured multiple ways, and it’s currently 50% off with a Walmart deal. Normally, you’d have to pay $510 to score this gorgeous patio set that’s glamorous enough to sit on the balcony of a high-end boutique resort, but the limited-time deal brings the price down to just $255. A 3-piece furniture set with a sectional couch and a generously sized coffee table like this one is rarely available at the price point, especially when the design looks so polished. With this deep discount, the patio set is an even more popular pick, with over 100 sets sold in the past 24 hours alone, so you won’t want to snag it for yourself.MF Studio 3-Piece Rattan Patio Conversation Set, $255 (was $510) at Walmart

Courtesy of Walmart

Shop at WalmartWhy do shoppers love it?Featuring a loveseat and chaise lounge topped with cozy cushions, this patio set has room to comfortably seat up to four guests. The third piece is a matching coffee table, which has a handy two-in-one design, allowing you to use it as a convenient surface to place your snacks and beverages or as an ottoman for kicking up your feet. The versatility of this set doesn’t end there. With options to set them up in over seven different configurations, the seat and table can adapt to a variety of layouts and floor plans.”I love the look of this patio set and the different configurations you can make,” one shopper raved. They also appreciated the plushness of the cushions, writing that they’re “very comfortable and thicker than other sets I looked at.” As a bonus, these cushions come with a water-repellent fabric that’s to keep them fresh and clean after splashes or storms.Related: Walmart’s bestselling 4-piece patio set with soft cushions is on sale for $109Designed to withstand all types of weather, this patio set is made from a durable synthetic rattan. It has the same earthy tan hue of natural rattan, but offers more resilience for outdoor use. The appearance of the set is further elevated with solid acacia wood on the feet, armrest, and table of the patio furniture, giving it an extra touch of luxury. Overall, the build of this patio furniture is solid and sturdy, which is proven by the couch’s weight capacity that exceeds 700 pounds.Pros and cons of the $255 MF Studio patio setPros:It’s a gorgeous design. Natural materials like rattan and acacia wood have a timeless look, but it’s also a popular choice for current interior decor trends this season.It comes with cushions: Many times more affordable patio furniture sets will come without cushions, but you don’t have to buy them separately with this deal.It allows for multiple configurations: If you have a small balcony or porch, this patio furniture can be arranged multiple ways to best fit your setup.Cons:Assembly is required: If you’re not handy, you’ll want to ask someone for help with the assembly process.It only comes in one color: We love the beige seats and light-hued rattan, but it’s also the only color available, limiting your options.Shop more patio furniture dealsAlpha Joy 3-Piece Wicker Patio Set, $300 (was $460) at WalmartLausaint Home 3-Piece Patio Outdoor Conversation Set, $185 (was $330) at WalmartLazzo 3-Piece Patio Outdoor Conversation Set, $166 (was $200) at WalmartUpgrade your outdoor space with the MF Studio 3-Piece Rattan Patio Conversation Set while it’s on sale for $255 at Walmart. This limited-time deal won’t last long, and it’s going fast, so we suggest adding it to your cart ASAP.

The Magnificent Seven had their worst day since April 2025 after Alphabet and Tesla spooked investors on AI spending. Bitcoin fell less than 1%, and dogecoin led the majors lower.

Trust isn’t damaged by the news itself — it’s damaged by how it’s delivered; here’s the four-part framework that keeps teams engaged even when the message is hard.

Most marketing delays trace back not to weak strategy, but to vague requests, invisible work and inconsistent follow-up.

Along with the recent collapse of several low-cost airlines, travel agencies and other companies selling various travel services have also had a tough first half of 2026.The string of recent travel company bankruptcies include British firms Great Little Escapes, Salamander Voyages, and Jetline, French cruise operator Expedis Exploration, Boston-based GoPlay Sports Tours and Australian tour-booking giant AVG Travels.While not selling any trips or tours, travel consulting firm Govassist LLC just became the latest to file for Chapter 11 protection in the U.S. Bankruptcy Court for the District of Puerto Rico on July 22.Travel company Govassist files for Chapter 11 bankruptcy in Puerto RicoThe company was founded out of Guaynabo in northern Puerto Rico in 2010 and provided consulting and assistance with applying for non-immigrant tourist visas to the U.S. to travelers who require them. The report from Bondoro shows that the company reported $1.1 million in assets and $15.9 million in liabilities as well as no ability to cover these debts to its creditors.Juan C. Bigas of Juan C. Bigas Lawis representing the company in the bankruptcy proceedings.Related: Another airline will be dissolved, all flights canceledOn its website, Govassist positions itself as a “personal visa consultant” offering help filling out forms for applications like the Electronic System for Travel Authorization (ESTA) required of travelers with visa-free agreements with the U.S. and a full visa for citizens of countries without it.Particularly for the ESTA visa, any “consulting” help that a private company can provide is very limited in scope given that travelers can simply go on the CBP website or mobile platform and enter one’s personal details and passport information for $40.27 USD for a two-year period if the electronic authorization is approved (if it is not, the money is refunded minus a $10.27 processing fee).

Govassist advertised services helping U.S.-bound travelers apply for electronic travel authorization and visas.Shutterstock

These are the kinds of visa consultation services sold by GovassistGovassist charges $129, comprehensive of the application fee, to fill this out for the traveler while stating that it powers part of this process through AI. The company could not be immediately reached for comment on the bankruptcy filing so it is also not immediately clear whether it will restructure or close down.More Travel News:Airline to launch unusual new flight to Cayman Islands from the U.S.There is a very cool Irish version of swimming pigs in the BahamasUnexpected country is most luxurious travel destination for 2026Low-cost airline launches easier way to get to Sri LankaSome recent travel agency bankruptcies in 2026:AVG Travels: The Melbourne-based travel agency selling cheap vacation packages to travelers in Australia and New Zealand sent more than 200 travelers an email saying that the trips were canceled before entering bankruptcy in May 2026.GoPlay Sports: In April 2026, the men’s basketball team of the University of Dallas was left without a planned trip to compete in the United Kingdom after Boston-based GoPlay Sports Tours LLC accepted two payments of $30,000 and then went unreachable.Havantur: Havantur was forced to shut down its main European office in Franceat the start of 2026 after tourist numbers to the Caribbean country plummeted due to U.S. military actions in Venezuela and threats against the country.Vegas Vacations and North America Destinations: Two travel agencies in the Canadian province of British Columbia, Vegas Vacations and North America Destinations, were shut down by regulators within a few days of each other in January 2026 after multiple travelers complained of buying trips with invalid plane tickets and hotel bookings.Related: Another airline cancels 3 flights to U.S., offers refunds

Anthony Joshua faces Kristian Prenga on July 25 at the Jeddah Superdome. Here’s the full card, location, DAZN start time and how to watch “The Comeback.”

The fixed-income ETF boom is often described as a response to yield, but the more revealing shift may be in how investors want to use bonds. Rather than treating a bond fund as a broad income sleeve, households and advisers are looking for vehicles that put future spending—retirement income gaps, tuition, renovations or travel—on a calendar. That changes the question from what a fund yields to whether its cash flows arrive when a bill does.Danielle Retski, an ETF capital market specialist at Northern Trust, described that demand as a search for “goal-driven solutions.” In the first half of the year, fixed-income ETFs took in 29% of all ETF flows despite representing 16% of ETF assets under management, she said. The mismatch suggests that investors are seeking a more deliberate role for bonds alongside risk-taking elsewhere in their portfolios.A distributing bond ladder is built around that scheduling problem. The underlying bonds mature in staggered calendar years, or rungs, and Northern Trust’s ladder ETFs return monthly interest while distributing principal when bonds mature rather than automatically reinvesting it. The design is notably different from a perpetual bond fund, where proceeds are generally rolled forward into new holdings.The Spending Date Is Becoming Part of the Investment ChoiceThe practical attraction is not limited to retirees. Retski cited college tuition, philanthropy, home improvement, travel and private-school payment plans as situations in which investors may want predictable outlays. In each case, the issue is not merely generating portfolio income; it is avoiding the need to decide, year after year, which investment to sell when an expected expense arrives.Any time that you would want consistent cash flows, our ladder ETFs take that federally tax-exempt income and put it in a way where in practice, investors are getting monthly interest income and also annual principal return to them so they can manage their spending needs with their income.That framing makes bond ladders a household-planning tool as much as an interest-rate instrument. For a retiree delaying full Social Security benefits, Retski offered a five-year ladder as an example of a bridge for expected cash-flow needs. For a family setting aside money for education or a renovation, the same structure can tie a future payment to a maturing rung rather than to an uncertain sale of a longer-lived fund.Returning Principal Solves One Problem but Narrows the Use CaseThe annual return of principal is also the feature that makes a ladder unsuitable as a catchall bond allocation. Retski said the products are designed for investors using goal-based investing, cash-flow management or budgeting tools. Someone whose primary objective is ongoing exposure to bonds, rather than a defined stream of future cash, is confronting a different portfolio question.The trade-off is especially relevant when rate expectations are changing. Retski said holding bonds to maturity and returning principal each year can minimize interest-rate risk and give investors duration control. But that benefit follows from a time-defined structure: cash is being paid out rather than simply remaining invested in a perpetual strategy.Tax and Inflation Concerns Are Being Folded Into the Same PlanThe product menu also shows how investors are trying to address several planning concerns in one decision. Northern Trust offers municipal bond ladder ETFs, MUNA–MUND, intended to provide federally tax-exempt income, and TIPA–TIPD TIPS Ladder ETFs, which use Treasury Inflation-Protected Securities. Retski said municipal bonds can be useful where tax-exempt cash flow is a priority, while TIPS are meant to help with inflation that exceeds what markets have embedded in the breakeven rate.TIPS illustrate why a spending plan still needs an inflation lens. Their principal rises with inflation and falls with deflation, Retski said, whereas a nominal Treasury’s yield includes a fixed market expectation for inflation. She pointed to shocks such as the war in Iran and a global pandemic as examples of events that can produce short-term inflation pressure not fully reflected in that expectation.The growing use of fixed-income ETFs, then, is not simply a referendum on yields or the Federal Reserve. It reflects a preference for making portfolio cash flows legible against real household obligations. A bond ladder cannot remove the need to decide whether a particular fund, tax feature or maturity schedule fits an investor’s circumstances. Its narrower promise is more concrete: for money earmarked for a known purpose, the timing of income and principal can be designed to matter as much as the return.

Kevin Harvick believes his 2003 Brickyard 400 win at the Indianapolis Motor Speedway was the “tentpole” victory that launched him on a NASCAR Hall of Fame career.

Every generation gets handed a financial script early, and the script tends to outlive the conditions that wrote it.Millennials got theirs somewhere between the 2008 layoffs and the third round of student loan paperwork.You graduated into a broken labor market. You rented for years longer than your parents did. You watched housing costs detach from wages and stay detached.That script hardened into something closer to a diagnosis, and it followed the generation into its 40s. It also shapes how you read your own account statement.So the annual benchmark reports land in a predictable way. You open one, hunt for the average balance for your age group, find yourself somewhere underneath it, and close the tab feeling the same as you did before, only with better documentation.That reflex is worth interrupting this particular year, and the reason sits in an annual report that most people never read past the first page.The 25th edition of How America Saves, the retirement study Vanguard has published since 2001, landed in June, carrying age-bracket detail that cuts hard against the script in a way the summary coverage skipped past almost entirely.Why the average 401(k) balance keeps making you feel behindThe number that travels is the average, and the average is close to useless for this particular job.Average participant balance hit $167,970 at the end of 2025, while the median, the midpoint where half of savers sit above and half below, was $44,115, according to Vanguard.More Personal Finance:Fidelity discloses wealth move that triggers hidden IRS taxMarket pivot point is here – how Investors should get readyDave Ramsey shares strong warning on 401(k)s, IRAsThat spread is not a rounding error. It is the whole reason benchmark stories make people feel worse than the underlying data warrant.Average balances are “more representative of the results experienced by longer-tenured, more affluent, or older participants,” the firm wrote. Vanguard puts its own average at roughly the 75th percentile, meaning three out of four participants hold less than that figure.One in four participants had less than $10,000 saved, while 35% held more than $100,000, and 18% held at least $250,000.None of it is wrong. It is just the wrong comparison for anyone trying to judge their own account.

Vanguard’s 2026 data show Millennial 401(k) medians up 15%, putting seven figures within reach.EF Volart / Getty Images

What Vanguard’s new Millennial 401(k) numbers actually showMillennials now span the 25-to-34 and 35-to-44 brackets, and both of them moved hard last year.The younger bracket’s median balance reached $18,732, up from $16,255 a year earlier. The older bracket’s median hit $46,919, up from $39,958. Both gains outran the 16% move in the all-participant median, Vanguard reported.Markets did most of that work rather than virtue. The average one-year participant return was 19.3% in 2025.Related: Vanguard warns of Social Security traps costing retireesTwo behavioral readings matter more for what happens next. Participants under age 45 held roughly 90% of plan assets in equities at the median, the heaviest allocation of any age group. And when the first quarter of 2026 turned choppy, only 5% of Millennials touched their allocation while 18.4% raised their savings rate, “in large part due to auto increases,” according to Fidelity.Positioned correctly and not trading. That pairing is rarer than it sounds.The tax positioning tracks, too. Roth adoption ran at 20% for the 25-to-34 group and 19% for the 35-to-44 group, the two highest rates of any age band, Vanguard found.Paying tax now on a balance with three decades of compounding ahead of it is the right trade when you are early.The generation is not maxing out, to be clear. Only 10% of the younger bracket and 15% of the older one hit the statutory limit last year.Here is the Millennial ledger in one place:Median balance for ages 25 to 34 reached $18,732, up from $16,255, based on Vanguard’s 2026 and 2025 editions.Median balance for ages 35 to 44 reached $46,919, up from $39,958, according to the same two reports.The average one-year participant return came in at 19.3% for 2025, Vanguard noted.The total 401(k) savings rate hit a record 14.4% in the first quarter of 2026, according to Fidelity.The employee deferral ceiling “increased to $24,500, up from $23,500 for 2025,” the IRS confirmed. Running the millionaire math on a median Millennial saverI ran the projections myself rather than trusting the round numbers that circulate every summer.Start with a 30-year-old sitting exactly at the median, $18,732, contributing at the 11.3% combined employee and employer rate Vanguard reports for that bracket, applied to the $90,000 median participant income. Thirty-five years at a 7% annual return produces about $1.6 million. Drop the assumption to a grim 6%, and it still clears $1.27 million.The 40-year-old is the harder case. Starting at $46,919 with an 11.8% combined rate, 25 years at 7% lands near $926,000.Short of the milestone, and that is exactly where the pessimistic version of this story usually stops.What my analysis turned up is the size of the shortfall. Closing it takes a 13.1% total contribution rate instead of 11.8%. On a $90,000 income, the difference is $97 a month.Vanguard already recommends a 12% to 15% total contribution rate. The median 40-year-old saver is not short by a lifestyle or a windfall. They are short by 1.3 percentage points, sitting inside a band the firm publishes every single year.I checked the top of that band, too. At 15%, the same 40-year-old lands near $1.11 million. Push to this year’s $24,500 ceiling, and it is roughly $1.8 million.The distance between the median outcome and the good one is measured in single-digit percentage points of pay.What a seven-figure 401(k) balance will actually buy in 2051Now for the part that gets left out of the cheerful version.Compounding runs both directions. At 2.5% annual inflation, a million dollars in 2051 buys roughly $539,000 in today’s money. The 30-year-old’s $1.6 million in 2061 works out to about $677,000.The milestone, in other words, is the wrong finish line. Millionaire is a headline. Your replacement income is the actual question.The more durable read on this year’s data is that the system is carrying weight the individual saver used to carry alone, through automatic enrollment, automatic escalation and target-date defaults. That machinery does not care how the generation feels about its own script.It also has a leak. Hardship withdrawals reached 6% of participants in 2025, triple the 2021 rate, at a median of $1,900.A $1,900 withdrawal is not a retirement problem. It is an emergency fund problem showing up in the wrong account, and it is the most fixable item here.Expect next year’s snapshot to look worse on balances alone. Total 401(k) assets slipped to $9.9 trillion by March 31 from $10.1 trillion at year-end 2025, according to the Investment Company Institute.Watch the deferral rate instead of the balance. One is weather. The other is the only variable on the page you actually control, and the 2026 numbers say Millennials are already moving it.Related: Vanguard sounds alarm on growing housing market problem

Microsoft AI released two new in-house models into public preview on Wednesday — MAI-Image-2.5-Pro, its highest-fidelity image generator to date, and MAI-Voice-2-Flash, a speech model built for high-volume enterprise workloads — while publishing production data that amounts to the company’s most aggressive argument yet that it can power its own products without leaning on OpenAI’s frontier models.The announcement, made by Microsoft AI’s Superintelligence team, lands roughly a year after the company committed to building purpose-built models internally, and it arrives with an unusual level of specificity about where those models now run: Bing, PowerPoint, OneDrive, Dynamics 365, Excel, GitHub Copilot, and Azure. The message to enterprise buyers — and, implicitly, to OpenAI — is that Microsoft’s homegrown models are no longer research projects. They are production infrastructure serving millions of users.”Each of these enhancements is a step toward the same goal: Microsoft products, powered by Microsoft models,” the company wrote in its announcement blog.How MAI-Image-2.5-Pro and MAI-Voice-2-Flash stake out opposite ends of the AI cost curveThe two new releases occupy opposite ends of what Microsoft calls the quality-speed-cost curve, and the positioning is deliberate. MAI-Image-2.5-Pro targets the premium tier: hero imagery, detailed editing, and precise in-image text rendering — the last of which has long been a notorious weak spot for image generation models. Microsoft priced the model at $5 per million text input tokens, $8 per million image input tokens, and $106 per million image output tokens. The base MAI-Image-2.5 model recently launched at No. 2 for image editing on Arena, the community leaderboard that has become a de facto scoreboard for generative media.The creative industry appears to be taking notice. Rob Reilly, global chief creative officer at advertising giant WPP, called the Pro model “a strong leap forward for GenMedia tools” in a statement included in Microsoft’s announcement, adding that “Microsoft has firmly established itself among the leaders in generative AI.”MAI-Voice-2-Flash goes the other direction. First previewed at Microsoft’s Build conference, Flash runs twice as fast as MAI-Voice-2 and costs 32% less, priced at $15 per million characters. It is designed for the unglamorous but enormous market of high-volume voice — call centers, voice agents, and real-time speech applications where latency and cost-per-call matter more than marginal gains in expressiveness. Together, the two models reflect a strategy of building families of models rather than a single flagship, because, as the company put it, a creative studio chasing maximum fidelity has very different needs from a customer service operation handling millions of calls a day.Microsoft’s production metrics show in-house models cutting GPU costs by up to 89%The model launches are arguably less newsworthy than the deployment metrics Microsoft attached to them — numbers that read like a systematic case for swapping out third-party frontier models across its product portfolio. Bing Image Creator now runs entirely on MAI-Image-2.5, end to end, marking the first time the consumer image tool is fully in-house. In PowerPoint, Microsoft says MAI-Image-2.5 reduces GPU costs by up to 84% compared with GPT-Image-2, OpenAI’s image model. In OneDrive, where MAI-Image-2.5 is now the default for key image-editing scenarios, the company reports a 26% increase in save rates, roughly 25% lower P95 latency, and 2.5 times greater efficiency under medium-utilization production workloads.On the voice side, MAI-Voice-2-Flash now powers Dynamics 365 Contact Center — the platform used by customers including T-Mobile and EasyJet — where Microsoft claims GPU cost reductions of up to 89%. The model is also integrated into Azure Voice Live for developers building speech-to-speech agents.Perhaps the most consequential deployment sits in healthcare. Microsoft’s Dragon Copilot, used by 170,000 medical providers and responsible for processing 28 million patient encounters last quarter, now runs on MAI-Transcribe-1.5 for its multilingual workflow across 58 languages. Microsoft says internal evaluations show a 50% relative reduction in both transcription and language-identification error rates across most languages — a meaningful claim in a domain where transcription errors can propagate directly into clinical notes.Inside the ‘hill-climbing’ strategy that lets small models beat GPT-5.6 in ExcelIn a companion post published the same day, Microsoft detailed the methodology behind these results — what it calls its “hill-climbing machine,” an integrated flywheel of data, models, and the product “harness” that surrounds them.The clearest example is MAI-Code-1-Flash, the lightweight coding model launched in GitHub Copilot in June. Microsoft says the model achieves an approximately 10% higher code accept rate than GPT-5.4 Mini and Claude Haiku 4.5 in VS Code, while using 10% fewer median tokens. Developer retention tells a similar story: users were 6% more likely to return across multiple days than with GPT-5.4 Mini, and 11% more likely than with Claude Haiku 4.5.Then Microsoft did something more interesting. It took the MAI-Code-1-Flash checkpoint and further trained it inside an Excel reinforcement learning environment, teaching a coding model the tools and workflows of spreadsheet knowledge work. The result, according to production user feedback, is a model on par with GPT-5.6 for the most common Excel tasks — while being small enough to run on Nvidia’s older H100 and even A100 GPUs rather than requiring the latest-generation accelerators.That hardware detail deserves emphasis. Every major AI company is fighting for allocation of cutting-edge chips, and a model that delivers frontier-adjacent quality on two-generation-old silicon fundamentally changes the deployment economics. It also frees the newest hardware — including Microsoft’s now-operational GB200 cluster — for training rather than serving.Satya Nadella’s ‘frontier diffusion’ manifesto redraws the OpenAI relationshipMicrosoft CEO Satya Nadella framed the announcements in a lengthy post on X titled “Frontier Diffusion & Control,” which functions as something close to a strategic manifesto. “We can now take saturated frontier capabilities and deliver them at scale and at lower cost through models optimized for high-usage products, while continuing to use frontier models for frontier needs,” Nadella wrote, adding that Microsoft is “beginning to route traffic across our first-party surfaces to MAI whenever our models match or outperform frontier alternatives.”Translated from executive prose: capabilities that were state-of-the-art a year ago are now table stakes, and Microsoft believes it can replicate them cheaply for the specific, repetitive tasks that dominate real product usage. Why pay frontier prices for a frontier model when a user just wants to reformat a spreadsheet column?Nadella was careful to note that “frontier models from OpenAI and Anthropic are part of the orchestration system alongside MAI” — but he also articulated a pointed principle of model independence, arguing that a company’s evaluations “should continue to hill climb even when any given model has been removed.” “Keeping the harness, memory, context, and skills outside the model, he argued, is what gives Microsoft control. The subtext is hard to miss. Reuters reported in April that Microsoft’s exclusive license to OpenAI’s technology had been revised into a non-exclusive arrangement, and The Information reported last September that Microsoft had begun incorporating Anthropic models into some products. Wednesday’s announcement completes the triangle: Microsoft as orchestrator, with its partners’ frontier models as interchangeable components and its own models absorbing an ever-larger share of routine traffic.”Developers cheer cheaper task-specific models while skeptics question Microsoft’s track recordThe response online captured both the appeal and the skepticism surrounding the strategy. “I love when people use small models for niche tasks,” wrote one X user, @mavihsk, responding to Nadella’s post. “Why do I have to use the all-knowing model just to change my field in Excel?” Another user, @nabu_lines, distilled the pitch neatly: “cost and performance both improve when you stop overusing the biggest model.”Others were less charitable about Microsoft’s execution track record. “Microsoft is the worst when it comes to listening to user feedback,” wrote designer @designedbyabin, arguing the company “will lose the AI race because they repeatedly failed to understand user needs.” And one user, @tokenoverflow, offered a drier critique of the model-independence pitch: “i want it keep hill climbing after removing microsoft.”The skeptics raise a fair point. Microsoft’s self-reported metrics — accept rates, save rates, GPU savings — come from its own internal evaluations, not independent benchmarks, and the company chooses which comparisons to publish.But the strategy’s logic does not depend on any single number. Nadella’s framing that software now has “real marginal cost for the first time” explains why Microsoft is obsessive about tokens, GPUs, and serving costs: when AI features run on every keystroke across a billion-user product portfolio, an 84% GPU cost reduction is not an optimization. It is the difference between a viable business and a money pit.Why Microsoft is turning its internal AI playbook into an Azure productThe final piece of the strategy is that Microsoft is selling the playbook, not just the models. Nadella explicitly positioned the hill-climbing approach as “a template for every other AI native, SaaS, or Enterprise company,” and Microsoft is packaging the toolchain through Foundry and what it calls Frontier Tuning — letting enterprises train specialized models against their own proprietary evaluations and reinforcement learning environments. That turns Microsoft’s internal cost-cutting exercise into an Azure product, and it gives enterprise customers a reason to run their AI workloads on Microsoft’s cloud even if the models themselves come from elsewhere.The company’s emphasis on models trained “on clean, traceable, enterprise-grade data, without distillation from third-party models” serves the same commercial end. In an industry facing mounting scrutiny over training data provenance, Microsoft is betting that enterprise buyers — and courts — will care where model capabilities come from. Microsoft says it is now extending the hill-climbing approach to Copilot Chat, Outlook, and PowerPoint, and both new models are available in public preview through Microsoft Foundry and the MAI Playground. “None of this is an endpoint,” the company wrote. “We’re just getting started.”Seven years ago, Microsoft bet more than $13 billion that OpenAI would build the future of AI. Wednesday’s announcement suggests the company has since learned a cheaper lesson: the future of AI may belong to whoever builds the frontier, but the profits belong to whoever makes it ordinary.