Let me tell you about the biggest mistake investors are making right now.
They’re still playing AI Phase A.
Phase A was the rollout. It was OpenAI launching ChatGPT and watching it become the fastest-growing application in human history. It was Google scrambling to release Gemini. It was Microsoft betting $13 billion on a company most people had never heard of. It was Nvidia’s stock exploding as everyone realized you needed their chips to train these models.
Phase A made a lot of people very wealthy. And if you owned the right stocks, congratulations.
But Phase A is over.
We’re now entering AI Phase B: Infrastructure.
And I’m going to tell you exactly what that means for your money.
NEW: FREE Buy Alert: 3 Stocks to Own in 2026
Why Phase B Changes Everything
Here’s what happened during Phase A: the world’s largest technology companies proved that artificial intelligence works. They built models that can write code, analyze legal documents, generate images, and hold conversations that pass the Turing test without breaking a sweat.
The technology is real. The demand is overwhelming. And now comes the hard part.
You see, these AI models don’t run on dreams and venture capital. They require physical infrastructure on a scale that most people cannot comprehend. Data centers consuming enough electricity to power small cities. Billions of precisely labeled data points to train and refine the models. Global connectivity networks to deliver AI services to every corner of the planet.
Phase A answered the question: “Can we build AI that actually works?”
Phase B answers a different question: “Can we build enough infrastructure to scale it?”
And right now, the answer is no.
The total addressable market for AI infrastructure is projected to explode from $38 billion in 2025 to $94 billion by 2029, according to corporate analysis from Equinix. That’s a 147% increase in four years. Data center vacancy rates have collapsed to historic lows. Power grids are straining under demand that’s growing faster than anyone predicted.
Microsoft has committed over $80 billion to AI infrastructure expansion. Meta is building what they call the largest AI research cluster in the world. Google has publicly stated they will spend “whatever it takes” to maintain their position.
These companies proved the technology works. Now they’re in a desperate race to build the infrastructure to deploy it at scale.
And that, my friend, is where the real money will be made in 2026.
Editor’s Note: Massive transfer of government-owned land, minerals, and energy is set to hit the market… Here’s how to profit starting with one $10 company… [Full Story]
The winners of Phase A were the model builders and chip makers. The winners of Phase B will be the companies that own the physical infrastructure these tech giants need. The data centers. The training data pipelines. The global connectivity networks.
In this report, I’m going to show you three investments positioned to dominate AI Phase B. One is a data center giant quietly becoming the landlord of artificial intelligence. Another is a small company that five of the “Magnificent Seven” tech giants literally cannot function without. And the third represents what may be the largest IPO in American history, confirmed just weeks ago by the most prolific industrialist of our generation.
But first, you need to understand why Phase B is different from anything we’ve seen before.
The Infrastructure Crisis Nobody’s Talking About
Here’s a number that should concern you: 100 kW per rack.
A few years ago, a high-density data center rack consumed about 10 kW of power. Today, AI workloads are pushing that number to 100 kW per rack. That’s a tenfold increase in power requirements, and the infrastructure simply doesn’t exist to support it at scale.
This is the Phase B bottleneck.
During Phase A, the constraint was algorithmic. Could we build models smart enough to be useful? The answer turned out to be yes.
During Phase B, the constraint is physical. Can we build enough data centers, train enough models with quality data, and deploy enough connectivity to meet demand?
Right now, the honest answer is: we’re trying.
Equinix, the largest data center provider in the world, currently has 53 major expansion projects underway across 40 metros in 24 countries. They’ve secured $15 billion in funding for their hyperscale joint venture. They’re building facilities capable of handling those 100 kW racks that AI workloads demand.
And they still can’t build fast enough.
The hyperscalers, Amazon, Microsoft, Google, Meta, are all competing for the same limited resources: land, power, cooling systems, and construction capacity. They’re signing long-term leases at premium rates. They’re locking in power purchase agreements years in advance.
This is what a Phase B scramble looks like.
And the companies that own the picks and shovels of this infrastructure buildout are positioned to benefit regardless of which AI models ultimately win.
That’s the key insight. During Phase A, you had to bet on which model or which chip would dominate. During Phase B, you can own the infrastructure that all of them need.
Investment #1: The Landlord of Phase B
There’s a company that operates 260 data centers in 71 markets worldwide. They control more interconnected data center space than any other company on Earth. And right now, they’re trading at a forward P/AFFO of just 18.6, well below their 10-year average.
The company is Equinix (NASDAQ: EQIX).
If Phase B is about building the physical infrastructure to scale AI, Equinix is the landlord collecting rent from everyone doing the building.
In Q3 2025, Equinix delivered 5.2% revenue growth and 8.6% AFFO per share growth with what the company called “robust demand and record bookings.” Their CEO, Adaire Fox-Martin, stated plainly that the company’s strong performance is “a clear signal of accelerating momentum, for Q4 and into 2026.”
But the numbers that matter for Phase B investors are in the expansion pipeline.
Equinix currently has 53 major projects underway, including 11 xScale builds delivering over 90 megawatts of new capacity. They’ve leased 400 megawatts of hyperscale capacity through their xScale joint venture, which has secured a staggering $15 billion in funding.
And here’s the Phase B detail that tells you everything: Equinix is now building data center facilities capable of supporting those high-density computing environments approaching 100 kW per rack. They’re not building for Phase A workloads. They’re building specifically for the infrastructure demands of Phase B.
Why does this matter to you as an investor?
Because Equinix operates what might be the most powerful network effect in the data center industry. With over 10,000 customers, including 2,100 network providers across cloud, financial services, and content providers, Equinix has built what they call their “interconnection platform.” This isn’t just real estate; it’s a network of networks.
Their Equinix Fabric virtual connections now exceed 50,000, with Google Cloud pioneering 50 Gbps bandwidth provisioning through their platform. When hyperscalers need low-latency connections between AI compute resources and the broader digital ecosystem, they go to Equinix.
The financials tell the Phase B story. Equinix delivered revenue of $2.02 billion in Q2 2025, up 14% year-over-year. They’ve raised their 2025 guidance upward, now projecting revenue of $9.233 to $9.333 billion and AFFO of $3.703 to $3.783 billion. The company maintains an EBITDA of $3.67 billion and a dividend yield of approximately 2.46%.
Their balance sheet is built for the Phase B buildout: over $3.1 billion in cash with a net debt to EBITDA ratio of 3.4x, providing substantial flexibility for aggressive expansion.
But what really excites me is the dividend story. Equinix has delivered approximately 10% dividend growth over the past twelve months while simultaneously investing billions in capacity expansion. For income-focused investors who also want growth exposure to AI Phase B, this is precisely the profile you want to own.
With shares trading around $770 and a market cap of approximately $75 billion, Equinix represents what I consider one of the most asymmetric risk-reward opportunities in the infrastructure space. Analysts’ price targets range from $804 to $1,218, with 18 out of 25 recent ratings coming in as “Buy.”
The risks are real. Cloud providers might build their own data centers. Construction delays and NIMBY opposition could slow expansion. Foreign exchange fluctuations have already impacted guidance.
But here’s my Phase B thesis: in a world where AI infrastructure demand is doubling every few years, the company that owns more global data center real estate than any competitor is positioned to benefit regardless of which AI models or applications ultimately win.
Equinix isn’t betting on whether OpenAI beats Google or whether Microsoft triumphs over Meta. They’re collecting rent from all of them as they scramble to build out Phase B infrastructure.
Investment #2: The Data Layer of Phase B
The first investment I showed you is a $75 billion industry titan. The next one is completely different.
It’s a small company, with a market cap around $1.65 billion. It doesn’t get mentioned on CNBC. Most analysts have never heard of it. And yet, this company works directly with five of the “Magnificent Seven” tech giants and three additional Big Tech companies.
I’m talking about Innodata Inc. (NASDAQ: INOD).
And to understand why this company matters for Phase B, you need to understand something about how AI actually works.
Here’s the dirty secret of artificial intelligence: AI is only as good as the data it learns from.
During Phase A, the model builders proved they could create intelligent systems. But those systems require massive quantities of precisely labeled, meticulously curated training data. We’re talking billions of data points, each one requiring human judgment to classify, annotate, and verify.
As Phase B scales these models to enterprise and government deployment, the data requirements don’t shrink. They explode.
The big AI labs call this “the data problem.” And it’s getting exponentially larger as models scale toward trillions of parameters. The quality of training data is now the critical differentiator. Bad data produces bad AI. And the companies building frontier models need data partners they can trust with their most sensitive intellectual property.
Enter Innodata.
This company has quietly positioned itself as the “picks and shovels” provider to the largest AI model builders on Earth. Their CEO, Jack Abuhoff, calls their approach “smart data” rather than “scale data,” meaning they analyze model deficiencies and prescribe precise datasets to improve factuality, safety, and reasoning.
The Phase B financial results have been nothing short of explosive.
In Q3 2025, Innodata delivered record revenue of $62.6 million, representing 20% year-over-year organic growth. Adjusted EBITDA hit $16.2 million, representing a healthy 26% margin. And the company reaffirmed guidance for 45% or more organic revenue growth in 2025.
But those numbers are just Phase B getting started.
Management has publicly stated they expect “transformative” growth in 2026, driven by several major catalysts that align perfectly with the Phase B infrastructure buildout.
First, pre-training data contracts. Earlier in 2025, Innodata recognized that model quality is increasingly tied to the quality of pre-training data. They invested $1.3 million to build new capabilities, and that investment has paid off spectacularly. The company has now disclosed $42 million in signed pre-training contracts, with an additional $26 million expected to close soon, totaling $68 million in potential revenues largely flowing into 2026.
This is pure Phase B demand. As AI companies scale their models, they need exponentially more high-quality training data. Innodata is positioned to supply it.
Second, the launch of Innodata Federal. In November 2025, the company announced a dedicated government-focused business unit designed to serve U.S. defense, intelligence, and civilian agencies. They’ve already secured an initial high-profile project expected to generate approximately $25 million in revenue, mostly in 2026.
The U.S. government is entering its own Phase B, scaling AI capabilities across defense and civilian applications. And they’ve tapped Innodata to help build the data infrastructure. General Richard D. Clarke, a retired four-star Army general and former Commander of U.S. Special Operations Command, has joined the Innodata board to guide this expansion.
Third, sovereign AI. Governments around the world are racing to build independent AI capabilities. Innodata is in active discussions with several sovereign AI programs across the Middle East and Asia, which they describe as “one of the most significant structural shifts in the global AI ecosystem.”
This is Phase B going global. Every nation that wants AI capability needs data infrastructure. Innodata is positioning itself as a critical supplier.
And here’s what caught my attention about their customer base: management disclosed that six of their eight major Big Tech customers are forecast to grow meaningfully in 2026, with several expanding substantially. They’ve also landed or expect to finalize five new Big Tech customers shortly, including two they describe as “global leaders in commerce, cloud and AI.”
When the biggest technology companies in the world are simultaneously expanding their relationships with a single data provider, that tells you something about Phase B demand.
The stock has already appreciated significantly, up over 30% in the past year and over 1,600% over three years. Some might say you’ve missed the boat.
I disagree.
The generative AI market is projected to grow from $37 billion in 2022 to $1.36 trillion by 2032, a compound annual growth rate of 45%. The data annotation and labeling market alone is expected to hit $3.63 billion in 2025 and grow at 26.5% annually through 2035.
Phase B is just beginning. And Innodata is building the data infrastructure that makes it possible.
Related: [BREAKING] Elon is taking SpaceX public in 2026. $100 Pre-IPO Opportunity-of-a-lifetime Open!
The company trades at a forward P/E of around 38 to 44, elevated compared to the broader industry but justified for a company delivering 45%+ revenue growth with expanding margins. They have $73.9 million in cash, no debt, and an untapped $30 million credit facility with Wells Fargo.
The risk is customer concentration. One large customer has historically represented a significant portion of revenue. And competition from larger integrators is always a concern.
But here’s my perspective: when five of the seven most valuable technology companies in the world are paying you for critical Phase B data infrastructure, you’re doing something right.
Innodata may be small today. I don’t think it stays that way as Phase B accelerates.
Investment #3: Phase B Goes Orbital
Editor’s Note: Have you seen this?! SpaceX is going public in 2026.
And here’s a way to get in for just $50… [Full Story]
Now we come to the investment opportunity that I believe could define 2026. And I need to be completely direct with you about what this is and what it isn’t.
On December 10, 2025, Elon Musk confirmed via a single tweet what space industry insiders had been speculating about for months: SpaceX is planning an IPO in 2026.
SpaceX, the most valuable private company in the world, is going public.
According to multiple reports from Bloomberg, Reuters, and other major financial outlets, SpaceX is targeting a valuation of approximately $1.5 trillion and aims to raise more than $30 billion. If achieved, this would make it the largest IPO in American history.
Now, you might be wondering: what does a rocket company have to do with AI Phase B?
Everything.
Here’s what most investors don’t realize: SpaceX isn’t primarily a rocket company anymore. Approximately 70% of their revenue now comes from Starlink, their satellite internet constellation. And their plans for the IPO proceeds include building data centers in space.
This is Phase B infrastructure taken to its logical extreme.
The vision is to create an orbital network of satellites using solar energy to process, store, and transmit data. Essentially taking the function of terrestrial data centers and putting them into space. Space-based data centers would have essentially unlimited access to solar power and would be free from the power grid constraints that currently limit terrestrial data center expansion.
Remember that Phase B bottleneck I mentioned, the 100 kW per rack power demands that are straining electrical grids worldwide? Space-based data centers could bypass that constraint entirely.
But before we get to the orbital vision, let’s talk about what SpaceX has already accomplished, because the execution is staggering.
SpaceX launched more rockets in 2025 than the rest of the world combined. On December 8, just two days before Musk confirmed the IPO, SpaceX launched its 159th Falcon 9 rocket of the year, putting 29 new Starlink satellites into orbit. One of those satellites was the 3,000th Starlink satellite launched in 2025 alone.
Read that number again: 3,000 satellites launched in a single year. SpaceX is operating at a scale and pace that no other entity in human history has achieved in space.
At last report, Starlink has surpassed 8 million subscribers and is growing rapidly. Revenue for 2025 is estimated at approximately $15 billion, with analysts projecting growth to $22-24 billion in 2026, representing 50% or greater annual growth.
According to internal SpaceX documents accessed by The Wall Street Journal, the company projected it could generate upwards of $36 billion in annual revenue from Starlink alone, with 60% operating profit margins, meaning $22 billion in annual operating profit.
For Phase B investors, Starlink represents something crucial: global connectivity infrastructure that can deliver AI services anywhere on Earth. As AI scales from tech company experiments to global deployment, you need connectivity everywhere. Starlink provides it.
The company also holds billions in NASA contracts, including a $255 million deal to launch the Nancy Grace Roman Space Telescope in 2027. And the Wall Street Journal reported SpaceX is set to secure a $2 billion contract to work on the Golden Dome missile defense project.
Now, here’s the investment challenge: how do you actually gain exposure?
Before the IPO, direct investment is essentially impossible for retail investors. SpaceX shares trade in private markets characterized by low transparency and high capital requirements.
However, there are ways to position yourself. Alphabet (Google) and Fidelity were early investors in Starlink and hold private stakes in SpaceX. EchoStar has exposure through strategic satellite industry positioning. Several specialized ETFs provide indirect exposure to the space economy.
But the real opportunity will come when the IPO actually launches.
Here’s my prediction: despite the eye-popping valuation, this IPO will be massively oversubscribed.
Traditional space companies like Lockheed Martin and Boeing earn less than 10% operating profit margins. Space startups like Rocket Lab and Redwire struggle to earn any profit at all. Meanwhile, terrestrial ISPs like Comcast earn operating margins of 38% on residential internet and 57% on business customers.
If SpaceX is really becoming a space-based ISP with satellite infrastructure and plans to add orbital data centers, the margin profile could be extraordinary. And with 70% of revenue already coming from Starlink, that’s exactly what’s happening.
The risks are substantial. Starship development has experienced explosive failures. Musk’s politics and business practices are controversial. The $1.5 trillion valuation at approximately 60-100 times revenue leaves little room for error.
But here’s my Phase B thesis: SpaceX is building infrastructure that extends beyond Earth’s surface. They’re not competing for the same limited power grids and land resources as terrestrial data centers. They’re building the next layer of global infrastructure.
When the SpaceX IPO opens to public investors in 2026, I believe it will represent the boldest Phase B infrastructure bet available.
Editor’s Note: Have you seen this?
Elon Musk just announced a 2026 $1-trillion IPO…
And here’s a way to get in for just $50… [Full Story]
The Phase B Opportunity
Let me bring this together for you.
AI Phase A proved the technology works. Chatbots can pass the bar exam. Image generators can create photorealistic art. Language models can write code that actually runs.
Phase A is over. The technology debate is settled.
AI Phase B is about building enough infrastructure to deploy these capabilities at global scale. It’s about data centers that can handle 100 kW racks. It’s about training data pipelines that can feed trillion-parameter models. It’s about connectivity networks that can deliver AI services to every corner of the planet.
The winners of Phase A were the model builders and chip makers.
The winners of Phase B will be the companies that own the physical infrastructure.
Equinix gives you exposure to the physical backbone of AI Phase B, a diversified REIT with global reach, growing dividends, and a management team aggressively expanding into high-density AI computing. They’re the landlord of Phase B, collecting rent from everyone scrambling to build capacity.
Innodata gives you exposure to the data layer that AI models cannot function without, a small company punching far above its weight, serving five of the Magnificent Seven, and positioned for what management calls “transformative” growth in 2026. They’re building the data infrastructure that makes Phase B possible.
And SpaceX, when it goes public, will give you exposure to Phase B infrastructure that transcends terrestrial constraints, global connectivity today, orbital data centers tomorrow.
These aren’t speculative technology plays. They’re infrastructure investments in the physical buildout that Phase B demands.
The model builders proved AI works. Now someone has to build the infrastructure to scale it.
That’s Phase B. And it starts now.
Tom Anderson Wall Street Watchdogs
Disclaimer: Wall Street Watchdogs is committed to uncovering the truth about financial markets and helping individual investors prepare for systemic risks that mainstream media won’t discuss. We receive no compensation from the companies or assets we analyze. This article is for educational purposes only and should not be construed as investment advice.
