YT YouTube2Text

YouTube2Text → Transcripts → Microsoft Just SHOCKED The Entire Tech Industry With Latest Announcement!

Microsoft Just SHOCKED The Entire Tech Industry With Latest Announcement! — Transcript

by States 2050 · 4,205 words · 638 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01Sometimes a company makes a bet so
0:03enormous that the entire financial world
0:05stops and asks if they have lost their
0:08mind. In the middle of 2026, Microsoft
0:11did exactly that. For 3 years, they had
0:15been quietly spending money at a pace
0:17that made investors physically ill. They
0:20were buying land, pouring concrete,
0:23ordering chips by the thousands, and
0:25building data centers across the planet
0:27at a scale that nobody in the industry
0:30could match. Then Microsoft took the
0:32stage in San Francisco and revealed what
0:35all of that spending was for. Four
0:38things announced in a single keynote. an
0:41artificial intelligence agent called
0:43Scout that lives inside your computer
0:46and does your work before you even open
0:48your laptop. A quantum chip that can
0:52supposedly hold its state a thousand
0:54times longer than anything they built
0:57before.
0:58Seven brand new AI models built from
1:01scratch by Microsoft's own team with no
1:04outside help and a complete reinvention
1:06of Windows that turns it from a passive
1:09desktop into an active brain for
1:12autonomous software. The stock market
1:14had already been smashing Microsoft for
1:17months. Their shares had dropped almost
1:2024% in the first half of the year alone,
1:23the worst stretch the company had
1:25suffered since the DOT crash more than
1:27two decades earlier. Investors were
1:30openly calling the spending reckless.
1:32And the response from Microsoft was to
1:35announce even more. This was not a
1:37product update. It was a $190 billion
1:41wager that the future of computing
1:44belongs to whoever builds the
1:45infrastructure for it first. To
1:48understand whether this is genius or
1:50madness, you need to look at what
1:52Microsoft was actually building during
1:54those three quiet years and why the
1:57market's reaction might end up being the
1:59wrongest call since someone said the
2:01internet was a fad.
2:04the $190 billion bet and a Wall Street
2:07panic.
2:09What Microsoft was doing during those
2:11three quiet years was not mysterious. It
2:14was physical. They were building
2:16buildings, massive windowless
2:20warehouse-sized structures filled with
2:22servers and cooling systems and enough
2:25electrical wiring to light up a small
2:27city. Every one of those data centers
2:30was designed to do one thing. run
2:33artificial intelligence at a scale
2:35nobody else could match. While Google
2:38and Amazon and Meta were also investing
2:40in AI, Microsoft was outspending all of
2:43them on the infrastructure layer. They
2:46were buying land next to power plants
2:48because AI data centers need so much
2:50electricity that being close to the
2:53source matters. They were ordering
2:55custom chips from suppliers and
2:57designing their own. They were signing
2:59deals with utility companies to secure
3:02energy capacity for the next decade.
3:05This was not a software project. It was
3:08a construction project the size of a
3:10national highway system. The logic
3:12behind this spending was simple, even if
3:15the price tag was terrifying. Microsoft
3:18believed that artificial intelligence
3:20was not a product you could just bolt
3:22onto existing software.
3:25It was a new kind of computing that
3:27required a new kind of foundation. You
3:30could not run a billion AI agents on
3:32servers built for web traffic. You
3:35needed machines designed specifically
3:37for the math that AI models do. You
3:40needed networks that could move data
3:42between those machines at speeds that
3:44regular internet infrastructure could
3:46not handle. And you needed so much of it
3:49that whoever controlled the most
3:51capacity would control the market.
3:54Microsoft was building that capacity
3:56before anyone else. And they were
3:58building it big enough that catching up
4:00would take competitors years. Wall
4:03Street looked at this and saw a company
4:06throwing money into a furnace. The
4:08analysts who downgraded the stock were
4:10not stupid. They were looking at real
4:13numbers. Microsoft's capital
4:15expenditures had jumped to levels that
4:17made no sense compared to current
4:20revenue from AI. The infrastructure was
4:23sitting there humming, burning
4:25electricity, and most of it was not
4:28being used yet. It was like building a
4:31stadium before the team existed. The gap
4:34between what Microsoft was spending and
4:36what it was earning from AI was the
4:39widest in the tech industry. Investors
4:41who had ridden the stock up for years
4:44were suddenly looking at a company that
4:46seemed to be racing toward a cliff. But
4:48Satcha Nadella has been here before. In
4:512015, when he pushed Microsoft into
4:55cloud computing, the same voices said
4:57the company was late and the market was
5:00crowded. Azure is now one of the most
5:03profitable businesses on Earth. When he
5:05pushed Microsoft toward artificial
5:07intelligence before most people could
5:10spell it, the same voices said it was a
5:12fad. That partnership with Open AI
5:15turned out to be the most consequential
5:18deal in the industry. Nadella's strategy
5:20has always been the same. Build the
5:23infrastructure before people realize
5:25they need it and then be the only one
5:27who has it when demand arrives. The
5:30difference this time is the scale. This
5:33is not a software bet. This is a bet
5:36that the entire economy is about to
5:39change how it processes information and
5:41that the company with the most computing
5:43power wins everything. A lot of you
5:47found this channel through a video
5:48exactly like this one. If you want to
5:51keep finding stories like it, a like and
5:53a subscribe are what make that happen.
5:56The most disruptive thing Microsoft
5:58announced on that stage in San Francisco
6:01has a simple name and it might be the
6:03thing that finally makes all this
6:05spending start to pay off.
6:08Scout, the virtual co-worker that never
6:10sleeps. The most disruptive thing
6:13Microsoft announced has a very simple
6:16name. It is called Scout. But do not let
6:19the simple name fool you. Scout is not a
6:22chatbot. It is not just another version
6:25of C-Pilot with a fresh coat of paint.
6:28Microsoft calls it an autopilot agent.
6:31It is an alwayson artificial
6:33intelligence that lives inside your
6:35Microsoft 365 setup. The biggest
6:39difference is that it does not wait for
6:41you to ask it anything. It just starts
6:44working. Imagine you have a co-orker who
6:47sits next to you and never sleeps. This
6:50co-worker reads all of your team's
6:52messages. They read your Outlook email
6:55threads. They look through your
6:56SharePoint files. They do all of this
6:59while you are getting your morning
7:00coffee. Before you even open your
7:03laptop, Scout has already caught a
7:05calendar conflict that you would have
7:07missed. It has drafted a status report
7:10that you forgot was due today. It has
7:12flagged an email chain that was about to
7:15slip through the cracks. It does this
7:17while the problem is still happening,
7:19not after it is too late. Scout runs
7:22with its own digital identity inside
7:25your organization. It has its own
7:28account. It has its own permissions. It
7:30has its own audit trail. In Microsoft's
7:33own words, it is a virtual co-worker. It
7:37is not a feature you switch on and off.
7:40It is a new member of the team. This is
7:42a huge shift in how we think about
7:44software. For decades, software has been
7:48a tool. You pick up a hammer, you use
7:51it, and you put it down. Scout is not a
7:54hammer. Scout is a carpenter. You give
7:56it a job and it goes and does it. Scout
7:59is built on something called OpenClaw.
8:02This is an open-source agent framework
8:05that spread across the artificial
8:07intelligence world in early 2026.
8:11The creator of OpenClaw got hired away
8:13by OpenAI, but the framework itself was
8:16already out in the open. Microsoft took
8:19this framework and wrapped it in the
8:21kind of governance that big enterprises
8:23actually require. Governance is a big
8:26word for rules and locks. When one Open
8:29Claw agent reportedly went rogue inside
8:32a researcher's inbox earlier in the
8:34year, the industry got a scary preview
8:37of what an unsupervised artificial
8:39intelligence agent could do. Scout is
8:42Microsoft's answer to that problem. It
8:45is autonomous artificial intelligence
8:47with an accountability structure bolted
8:50on top of it. Microsoft expects Scout to
8:53help manage something massive. They
8:55project there will be 1.3 billion
8:58artificial intelligence agents
9:00automating office work by the year 2028.
9:04That is not a typo. It is 1.3 billion
9:08agents. These agents are not just simple
9:11scripts. They are smart systems that can
9:14understand context and make decisions.
9:17Each one of these agents will be taking
9:19a slice of the scheduling, drafting, and
9:22coordinating that currently fills the
9:24workday for hundreds of millions of
9:26people. Think about how much time you
9:28spend just sending emails back and forth
9:31to set up a meeting. Scout is designed
9:34to make all of that invisible. It just
9:36happens in the background. The question
9:39every IT manager and CEO is quietly
9:42asking is not whether this is
9:44impressive. It clearly is. The question
9:47is whether they are ready to trust it.
9:49Can you trust a virtual worker to read
9:52your emails and draft your reports? Can
9:55you trust it to not make a mistake that
9:57costs the company millions? These are
10:00the questions that will decide if Scout
10:02becomes the new normal or just another
10:04failed experiment. But while Scout was
10:07the most disruptive announcement, it was
10:10not the most confusing one. Microsoft
10:13also dropped a bombshell in the world of
10:15physics that had scientists arguing in
10:18real time.
10:20Majorana 2, a quantum leap or a physics
10:24fight.
10:26The quantum announcement was the one
10:28that made people forget about the money
10:30for a minute. Microsoft stood on stage
10:33and told a room full of developers and
10:36journalists that they had solved one of
10:38the hardest problems in physics. Their
10:40new chip Myerana 2 uses cubits that
10:44survive for 20 full seconds. To
10:47understand why that number matters, you
10:50need to know that in quantum computing,
10:52a cubit is like a soap bubble. the
10:55slightest vibration, the tiniest
10:57temperature change, any microscopic
11:00disturbance, and it pops. When a cubit
11:03pops, the computation collapses, and you
11:06start over. Every quantum computer ever
11:09built has struggled with this. The best
11:12machines from Google and IBM keep their
11:14cubits alive for micros seconds at a
11:17time. Microsconds are so short that you
11:20can barely measure them. Microsoft was
11:23standing on stage claiming 20 full
11:25seconds and some of their cubits lasted
11:28close to a minute. That number is the
11:31difference between a quantum computer
11:33that works in a lab and one that works
11:35in a business. At 20 seconds, you can
11:38actually run calculations that mean
11:40something. You can model a molecule. You
11:44can solve an optimization problem. You
11:46can do the kind of math that would take
11:48a classical supercomput millions of
11:50years. and you can do it in an
11:52afternoon. Microsoft says this puts them
11:55on track for a commercially viable
11:57quantum computer by the year 2029. Their
12:01original timeline was much longer. They
12:03just cut it in half. If that timeline
12:06holds, the consequences are almost
12:09impossible to imagine. Quantum computers
12:12are not just faster versions of the
12:14laptops on our desks. They operate on
12:17completely different rules of physics. A
12:20classical computer, no matter how
12:22powerful, processes information one
12:24piece at a time. A quantum computer can
12:28process many possibilities at once. That
12:31means certain problems that are
12:32practically unsolvable today become
12:35trivial. Modeling complex molecules for
12:38drug companies so we can discover
12:40medicines in weeks instead of decades.
12:43Cracking the optimization math that
12:46controls global supply chains and
12:48financial markets. And the one that
12:50keeps security experts awake at night.
12:53Breaking the encryption that protects
12:55every bank account, every email, and
12:58every password on the internet. Whoever
13:01owns a working quantum computer at scale
13:04will have computational power that no
13:07classical machine can match or defend
13:09against. But the physics community was
13:12not clapping. On the exact day of the
13:15announcement, a major scientific
13:17publication reported that independent
13:20physicists were challenging the entire
13:22foundation of Microsoft's approach. The
13:25topological cubit, which is the core
13:28idea behind Myerana 2, has been
13:31controversial since the beginning.
13:33Outside experts called the evidence
13:35extraordinary and thin. That word
13:39extraordinary is doing a lot of work in
13:41that sentence. It means the claims are
13:44so big that the proof needs to be
13:46undeniable and the experts were saying
13:49the proof was not there yet. Microsoft
13:52has been here before with the original
13:54Majerana 1 chip in 2025. They unveiled
13:58it with fanfare and the scientific
14:01community spent months asking questions
14:03that were never fully answered. The
14:06doubts never went away. Now, Microsoft
14:09was asking the field to trust them again
14:11on a timeline that was twice as
14:13aggressive with stakes that were twice
14:15as high. The difference between 2029 and
14:19a longer timeline is not just a number.
14:23It is the difference between building
14:24something that changes the world and
14:27building something that changes the
14:28world while your competitors are still
14:30in the planning phase. The argument will
14:34not be settled by a keynote speech. It
14:37will be settled in peer-reviewed
14:38journals where scientists publish their
14:41data and let other scientists try to
14:43tear it apart. That process is slow and
14:46it is brutal and it does not care about
14:49stock prices or marketing campaigns.
14:52Microsoft is betting that their data
14:54will survive that process. If it does,
14:58they will have something that no other
15:00company on Earth has. If it does not,
15:03they will have spent years and billions
15:05of dollars chasing a physics problem
15:07they could not solve. But while the
15:10physicists were arguing, Microsoft
15:12quietly announced something else that
15:14got less attention, but might matter
15:17more in the short term.
15:20Seven new AI models to escape Open AI's
15:23shadow.
15:25Under the two huge headline
15:27announcements, Microsoft slipped in
15:29something the industry has been waiting
15:31for. With a mix of anticipation and
15:33dread, they announced seven new AI
15:36models built entirely in-house by
15:38Microsoft AI. These were not licensed
15:41from OpenAI. They were not co-developed
15:44with a partner. They were their own.
15:47This is a massive deal for the company.
15:50And to understand why, you need to
15:52understand the tricky situation
15:54Microsoft has been in. Right now, 45% of
15:58Microsoft's $625 billion contracted
16:02backlog is tied to one customer. That
16:06customer is Open AI. If you do not know,
16:09Open AAI is the company that makes Chat
16:11GPT. Microsoft has a very close
16:14relationship with them. But that
16:16concentration of money has spooked Wall
16:19Street in a way that no amount of
16:21reassurance has fixed. Investors are
16:24worried. What if OpenAI goes away? What
16:27if they decide to use a different cloud
16:29provider? It is a huge risk to have
16:32nearly half a trillion dollars of future
16:34revenue riding on a single partner. and
16:38that partner is out signing deals with
16:40Amazon and Oracle at the same time. It
16:43is like having all your money invested
16:45in one stock. It is great when the stock
16:48goes up, but it is terrifying if it
16:51crashes. Microsoft has been trying to
16:54tell investors that everything is fine,
16:56but Wall Street is not listening. They
16:59see the risk every quarter. They see the
17:01reliance on a partner that is growing
17:03more independent every day. They see a
17:06future where Microsoft might just be a
17:08landlord for a tenant who eventually
17:11buys their own house. Microsoft's
17:14solution to this problem is the MAI
17:16series of models. The lineup covers
17:19reasoning, coding, image generation,
17:22transcription, and voice. MAI thinking 1
17:26is Microsoft's first dedicated reasoning
17:29model. This means it can think through a
17:32problem step by step. MAI code one flash
17:35is an agentic coding model built for
17:38GitHub copilot. This means it can write
17:41software for you. Then there is MAI
17:44image 2.5, MAI transcribe 1.5 and MAI
17:49voice 2. They shipped seven models at
17:52once and they announced them next to
17:55Scout and Majerana 2 almost like a
17:57footnote. But they are not a footnote.
18:00They are a safety net. They are a shield
18:03against the risk of losing open AI.
18:06These MAI models are not frontier
18:09models. A frontier model is the biggest,
18:12smartest, most expensive model in the
18:14world like GPT5.
18:17The MAI models are not that. But they do
18:20not have to be. They just have to be
18:22good enough for the everyday jobs that
18:25Microsoft products do. They need to be
18:27good enough to transcribe a meeting.
18:30They need to be good enough to write an
18:32email. They need to be good enough to
18:34generate an image for a presentation.
18:36And they need to do it at a price that
18:38does not leave Microsoft beholden to
18:41OpenAI. Every single time a Microsoft
18:44model is good enough, fast enough, and
18:47cheap enough to run inside a Microsoft
18:50product without calling OpenAI servers,
18:53Microsoft becomes a little less
18:54dependent. Every time an email is
18:57transcribed using MAI transcribe 1.5,
19:01that is money that stays in Microsoft's
19:03pocket. This is about control. This is
19:07about making sure that the future of
19:08artificial intelligence at Microsoft is
19:11not controlled by someone else. They
19:14want to own the engine, not just rent
19:16it. And that brings us to the final
19:19piece of the puzzle. All of these models
19:21and agents need a place to live. And
19:24Microsoft just announced that they are
19:26changing the very foundation of where
19:28they live. They are rebuilding Windows
19:31from the ground up.
19:33Windows rebuilt for AI agents.
19:36For 30 years, Windows has been the most
19:39popular operating system in the world.
19:42But Microsoft's build 2026 message for
19:45Windows was not about a new look or a
19:48faster file explorer. It was about a
19:51change at the architecture level. They
19:54are tearing down the old walls and
19:56putting up new ones. Windows is becoming
19:59a platform for AI agents. This means new
20:03runtimes, new developer APIs, and new
20:06ondevice AI frameworks. It is built not
20:10for people clicking menus, but for
20:12software acting on their behalf while
20:14they do something else. This is the
20:17biggest change to Windows since it was
20:19first created. Think about how you use a
20:22computer today. You open a program, you
20:25click a button, you type some words, you
20:28move your mouse, the computer waits for
20:30you to do something. It is a passive
20:33tool. It just sits there until you tell
20:36it what to do. You are the driver and
20:39the computer is the car. But the new
20:41Windows is not going to be like that.
20:44The new Windows is going to be active.
20:47It is going to have a small local model
20:49on the device. This model will handle
20:52work that needs to happen instantly like
20:55figuring out what you are trying to do
20:57or sorting basic files. The heavier
21:00requests, the ones that need a lot of
21:02brain power, will go to Azure. Azure is
21:05Microsoft's giant cloud system. The
21:08device and the cloud will work as one
21:11continuous surface. The AI agents will
21:14live in both places at the same time.
21:17That is what Microsoft means when it
21:19says Windows is no longer just a desktop
21:22operating system. It is an AI operating
21:25system. The next release of Windows 11
21:28is expected in the fall of 2026. It will
21:31ship with an ondevice AI runtime code
21:34named Copilot Engine. After that comes
21:38Windows 12. Microsoft has signaled that
21:40in Windows 12 AI agents will be first
21:43class citizens. They will get native
21:46hardware acceleration.
21:48This means the computer chips inside
21:50your laptop will be built specifically
21:52to run these AI agents fast and
21:55efficiently. The Windows most of us grew
21:58up with is being retired. The one that
22:00politely waited for you to doubleclick
22:02an icon is going away. Whatever replaces
22:06it is not going to wait around like
22:08that. It is going to anticipate what you
22:11need. It is going to act before you even
22:13ask. This is a huge change for the
22:16billions of people who use Windows every
22:19day. It is a huge change for the
22:21companies that build software for
22:23Windows. Developers now have to build
22:25for agents, not just for humans.
22:29Imagine a world where your computer
22:31knows you have a meeting in 10 minutes.
22:33It knows the traffic is bad. It knows
22:36you need to send a file to your boss
22:38before you leave. The new Windows will
22:41do all of that for you. It will send the
22:44file. It will warn your boss. It will
22:46pull up the meeting notes. You will not
22:49have to do anything. The computer will
22:51just be ready for you. This is the world
22:54Microsoft is building. It is a world
22:57where the operating system is no longer
22:59just a desktop. It is a partner. But
23:02there is one big problem hiding
23:04underneath all of this amazing
23:06technology. The humans are not ready for
23:09it. The adoption problem and the future
23:12of the bet.
23:14Here is the uncomfortable part
23:16underneath all of this. Microsoft's AI
23:19segment has hit a $ 37 billion
23:22annualized run rate. It is growing 123%
23:27year-over-year.
23:29Azure's AI revenue is accelerating by
23:32almost every technical measure. The
23:34business is working. The numbers are
23:37going up. The money is coming in. But
23:40there is a hidden problem. Of
23:42Microsoft's 450 million paid Microsoft
23:45365 seats, only a tiny fraction of
23:49people actually use the AI features.
23:52There are 450 million workplaces that
23:55already pay for co-pilot access, but
23:58only somewhere between 3% and 4.4% of
24:02users touch it daily. Think about that.
24:06Almost everyone who has access to this
24:08AI is ignoring it. The infrastructure is
24:11built, the data centers are running, the
24:14models work, and the humans in front of
24:17the screens still are not convinced.
24:20They are not using it. It is like
24:22building a massive highway and nobody is
24:25driving on it. It is like buying a gym
24:28membership and never going. The
24:30technology is there but the habit is
24:33not. People are creatures of habit. They
24:36like to do things the way they have
24:38always done them. Getting them to change
24:41is hard. That is the problem Scout is
24:44meant to solve. And Scout does not solve
24:47it by waiting for people to change their
24:49habits. It removes the need to form a
24:52habit at all. You do not have to
24:54remember to use Scout because it is
24:56already there when you open your laptop.
24:59It is reading and working before you
25:01have asked it for anything. You do not
25:04have to learn a new tool. You do not
25:06have to change your routine. The AI just
25:09blends into the background. It becomes
25:12part of the woodwork. You do not use it.
25:15It just uses itself for you. Whether
25:18that sounds like freedom or surveillance
25:20depends on who you ask. The executives
25:23who sign the contracts see a
25:25productivity jump. They see work getting
25:27done faster. They see money being saved.
25:31The employees whose email scout has
25:33already read before they finish their
25:35coffee are asking a very different
25:37question. They are asking if they want a
25:39robot reading their messages. They are
25:42asking if they are being watched. It is
25:45a question of trust. It is a question of
25:48privacy. And it is a question that
25:50Microsoft does not have a perfect answer
25:52for yet. Microsoft has placed a $190
25:56billion bet that the fix for slow AI
26:00adoption is not better marketing. It is
26:03better technology shipped deep enough
26:05into the operating environment that
26:07opting out takes more effort than opting
26:10in. They are betting that people will
26:12just accept it because it is easier than
26:15fighting it. That is a bet with huge
26:18upside and real risk. Nadella made this
26:21bet in San Francisco in June of 2026. He
26:25made it with the confidence of someone
26:27who has been right about this industry
26:29before. He knows that being right early
26:32can look for an uncomfortably long
26:34stretch, almost exactly like being
26:36wrong. The companies that underestimated
26:39Microsoft in 2012 lost a decade. The
26:43ones that dismissed the Azure bet in
26:452015 missed the cloud era. Whether 2026
26:50gets remembered as the year Microsoft
26:52laid the foundation for the next 20
26:54years of computing or the year it
26:57overextended in the dark will not be
26:59settled on stage. It will be settled
27:02slowly in quarterly earnings. It will be
27:05settled in boardroom AI adoption
27:07reports. It will be settled in whether
27:09Scout becomes something people cannot
27:11work without or something they never
27:14notice. It will be settled in whether
27:16Myerana 2 survives peer review. And it
27:19will be settled in whether $190 billion
27:22earns back a return that justifies the
27:25faith. The tech market just got blown
27:28up. Whether what grows back belongs to
27:30Microsoft is still being written. If you
27:33enjoyed this breakdown of the biggest
27:35tech announcement of the year and want
27:37to see more like it, please leave a
27:39like, subscribe, and turn on all
27:41notifications so you never miss a future
27:44video. There are related videos on the
27:47screen right now if you want to keep
27:49watching. So, here is a question for
27:51you. Do you think you would trust an AI
27:54agent like Scout to read your emails and
27:57do your work for you, or would you turn
28:00it off the second you could? Let me know
28:02in the comments.

About this transcript

This page contains the full transcript of Microsoft Just SHOCKED The Entire Tech Industry With Latest Announcement! by States 2050, generated from the public captions YouTube serves with the video. The transcript has 4,205 words across 638 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.