Next TMT Talks
What’s the driving force behind Friday’s $1.75 trillion SpaceX IPO? Well, it’s probably the same D2D technology that’ll soon be driving your smart phone. In this audio chat, we catch up with Ookla analyst Mike Dano … who actually flipped over one of those Starlink birds while summiting Washington’s Mount Rainier last week.

What’s got two legs, a torso, a 5G connection and works in the automobile factory 24 hours a day?

Humanoid robots are coming, with bipedal machines designed after the human form planned to arrive in pilot programs for industrial uses over the next years, around 2028 these anthropoid devices will start to be deployed in public spaces like airports, hospitals and retail establishments.

By early next decade, between 2032-2035, these walking biped computers — the literal embodiments of “embodied AI” — are expected to hit mainstream consumer and industrial acceptance.

Unsurprisingly, humanoid robots were already being hyped up at events like the Consumer Electronics Show in Las Vegas this January and in publications including Forbes. I was told that the dancing robots I saw displayed at several vendor stalls at Mobile World Congress (MWC) in Barcelona this year were likely being controlled by human operators offstage, although I have no way to confirm or deny that. It does seem entirely possible given the stage of development the industry is at with humanoid robots!

Barclays Investment Bank predicts that the market for AI-enabled humanoid robots will reach a base value of $40 billion by 2035, with upside predictions of up to $200 billion. The bank said that the market is currently valued at $2 billion - $3 billion.

The current rise of humanoid robots is closely linked to the increased deployment of private 5G networks in Asia, Europe and North America, as such networks can provide the uncongested, low-latency connectivity that the biped robots will need to operate on factory floors and more.

“We're closely following humanoid robotics from a connectivity perspective,” SNS Telecom & IT’s 5G research director, Asad Khan, told Next TMT Networks.

A good example of the current state of the market is Chinese automotive engine parts supplier Fulin Precision's recent adoption of 100 semi-humanoid robots that are coordinated by a private 5G-Advanced network. These robots have freed factory workers from repetitive box-moving tasks, the analyst claimed.

“Basically, these are wheeled robots with a bionic dual-arm design manufactured by AgiBot,” Khan said.

“Tesla could potentially emerge as a prime mover in Western markets in the future,” the analyst noted. This shouldn’t surprise anyone who has heard Elon Musk’s pronouncements on how Tesla’s future will be shaped by its Optimus robot.

Approximately "80% of Tesla's value will be Optimus," Musk wrote on X in 2025.

The SNS analyst also noted, however, that Tesla’s current shipment volumes are fewer than 200 units annually right now, compared to more than 5,000 units from Chinese suppliers such as AgiBot. Musk said on Tesla’s Q4 2025 earnings call earlier this year that the company is retooling its Fremont, Calif. factory to produce 1 million Optimus robots a year.

AI maven Nvidia is also trying to break into the humanoid robot market. The firm has just introduced its Isaac GR00T reference design for humanoid robots for academic research. It is not known how the design might be used in the future, but Nvidia has broken into 5G and 6G research through similar academic reference designs.

“The reference design is well suited for industrial uses,” Khan noted. “We expect widespread adoption in industrial settings, although this is likely to take several years before we see fleets of truly humanoid robots supported by private 5G networks in factories and other facilities.”

“Some of the other companies that we've been tracking in this space include Unitree, UBTech, Leju Robotics, EngineAI, Fourier Intelligence, Agility Robotics, Figure AI, Tesla, Boston Dynamics and NEURA Robotics,” Khan concluded.

Usage-based and hybrid pricing models are reshaping B2B revenue—but for finance teams, they're also introducing serious complexity around revenue recognition, forecasting, and operations.

Join Rebecca Schwartz, Co-founder of Tabs, and Amit Dhir, Partner at PwC, for a live session on June 10th from 1–2PM EDT. They'll break down how modern pricing decisions ripple through your financial workflows—and how to scale without adding manual overhead.

You'll leave with practical frameworks, real-world examples, and a clear path forward.

Register now to secure your spot—and if you can't make it live, a recording will be available.

Can't wait? Explore more Tabs webinars to catch up on past sessions.

Save My Spot

AI RAN is heading into the field trial stage later this year, with actual commercial deployments expected to start in 2027.

AI-RAN (Artificial Intelligence - Radio Access Network) is a telecom technology integrating AI directly into mobile network infrastructure to optimize cellular performance and host AI applications at the network edge.

“The alignment is around the different long-term visions that the telcos have,” Roy Chua, founder and principal analyst at AvidThink told Next TMT Networks. “Machine learning techniques have been used for capacity and spectrum optimization, and AI will continue to be used in similar ways, and in assisting with the management of the RAN.”

Chua also noted that T-Mobile is cautiously optimistic about AI RAN, while SoftBank is “pushing hard” on the technology. Chua highlighted SoftBank’s development of AITRAS orchestration to use spare RAN compute for separate AI and RAN workloads.

Indosat, he stated, is the first South East Asian operator to deliver an AI RAN Layer 3 call, while NTT DoCoMo and Elisa are “enthusiastic.” 

Earl Lum, president at EJL Wireless, noted that China Mobile, China Unicom and China Telecom are also going after AI RAN using Huawei and ZTE gear. He said that the Chinese AI is “going in a different direction” versus Western vendors like Nokia and Ericsson are.

Huawei is using its Ascend processors to run both AI and telecom workloads. Unlike some Western operators, the Chinese operators are focused on level 4 autonomous networks, rather than level 5. This leads to an autonomous network that can self-heal and make decisions, but isn’t fully level 5 autonomous. 

That doesn’t mean that all mobile operators are gung-ho about AI RAN, Chua said. “Verizon prefers GPUs in centralized facilities and hasn't seen a viable [total cost of ownership] yet,” he noted. “AT&T doubts AI compute at the far edge today, and demoed AI-in-RAN with Ericsson without GPUs.”  

Liverpool-based New Wave band Flock of Seagulls predicted a ubiquitous presence for AI RAN in mobile networks as far back as 1982, when the technology was so far away.

“Orange, Vodafone and Telus are in the cautious-PoC middle,” the AvidThink analyst stated. “They are exploring while evaluating the economics of GPUs at the far edge.”

Western vendors including Ericsson, Nokia and Samsung have partnered with a slew of operators. “T-Mobile is working with Nokia, NVIDIA and Ericsson on different aspects of AI-RAN, Chua said. At MWC this year, Nokia announced completed functional tests of GPU-accelerated AI-RAN at T-Mobile's AI-RAN Innovation Center.

That investor was wrong. Gamma is now worth $2B, with 50M users and more than half their growth driven by word of mouth.

They're one of 6 AI-native startups in HubSpot for Startups' free Bold Bets Playbook. Replit grew revenue 50x after half the team pushed back on the strategy. Ramp generated 100M+ views from a single stunt. Clay's co-founder wouldn't hang up a sales call until the prospect DMed him in Slack.

Each one took a GTM risk most founders would never greenlight. Each one paid off.

Get the free Bold Bets playbook

“The real milestone is still not here yet,” the analyst said.

Nokia committed to field-trial-ready AI-RAN in Q4 2026, with commercial 5G products set to be deployed in late 2027.

“So far, most of the early results are in functional proof-of-concepts, and we haven't seen economic data” Chua said. “The question of whether we will see more GPU-type architectures replace today's specialized ASICs and CPU deployments is still unanswered.” 

Ericsson and Intel announced an AI-native 6G collaboration, and Ericsson demoed RAN AI features running on Granite Rapids - Xeon 6 with an integrated virtual RAN Boost acceleration. Intel says CPUs "easily" handle 9 billion-parameter models inline, with no GPU required.

Samsung is in the same camp, validating vRAN on Granite Rapids with Vodafone (they've also shared that AMD CPUs can power vRAN even for massive MIMO situations).

Ericsson also demoed Cloud RAN portability on Nvidia's Aerial/CUDA platform with T-Mobile at MWC26. Taking a multi-pronged approach, Ericsson is investing in its own custom silicon for AI-RAN.

d“So far, we've seen that money can encourage vendor alignment, but not operator deployment yet. The [total cost of ownership] and business case aren't clear, and many telcos like Verizon or Orange aren't convinced, at least not with today's GPU platform, Chua stated.

“Nevertheless, what we anticipate is that architectures will morph over time -- NVIDIA, or other vendors, will ship higher efficiency GPUs/XPUs that can support ML models providing value in 5G or 6G deployments.

Likewise, Intel and AMD can add more acceleration functions — including vector/matmul units — to better handle [machine learning] functions. There isn't really a pure CPU or GPU today, whether desktops, mobile devices, or servers -- it's a blend of architectures that address a mix of workloads,dd the analyst said.  

The business case of running other AI workloads leveraging GPUs at the far edge is still being evaluated. “There are limited use cases today and no signs of monetization for that as yet,” Chua concluded.

Kingscrowd Capital's senior analyst reviewed RISE Robotics and projected potential growth to a $500 million valuation. The community round is open now on Wefunder. You don't have to be an institutional investor to get in at today's price.

See the Opportunity

Given that SpaceX is about to hold the largest Initial Public Offering (IPO) in history on Friday, targeting a $1.75 trillion valuation, Next TMT Networks took a look at its largest competitor in putting up satellite swarms for communications.

Amazon Leo is the biggest name rival to Elon Musk’s Starlink Mobile in the low Earth orbit (LEO) direct-to-device (D2D) satellite communications race.

Jeff Bezos’ Amazon Leo division is currently hiring for 578 open positions for the space unit, largely to deal with the heavy hardware, software and technical requirements needed for the new Leo arm.

As TMF Associates analyst Tim Farrar told TMT Next Networks that prospects for Amazon Leo catching up to the company that holds the pole position in fixed satellite broadband and streaking ahead in the D2D market will be a heavy lift.

Amazon Leo currently has 331 satellites deployed and will require 578 satellites,— orbiting 350 - 450 miles above the earth — to run its initial broadband commercial service, which will launch in 2028. Amazon plans to eventually launch 3,236 satellites to form its full LEO constellation by July 2029.  

Farrar noted that Amazon’s recent Blue Origin rocket explosion on the launch pad at Cape Carneveral was a setback to, but not fatal for, Amazon’s space plans. “It must be frustrating for them though,” he stated.

The firm recently asked the FCC to waive its July 30 deadline for Amazon to launch half its LEO satellite swarm. The FCC granted Amazon’s request last Friday, despite complaints from SpaceX.

“They don't want to cut them off,” Farrar noted.

The agency wrote that Amazon would be “a new competitor-at-scale in low Earth orbit that offers high-speed broadband to customers in unserved and underserved areas,” to the dominant Starlink and SpaceX paradigm.

The FCC is still requiring that Amazon Leo launches half of its satellite fleet by March 30, 2028. The July 2029 deadline for Amazon to get all its first generation satellites in orbit still stands, too.

aSo despite some small respite, Amazon will need to keep designing, building and launching satellites to meet the deadline, which perhaps explains the company’s push to hire more satellite engineers and systems designers. 

“The bigger issue here is how much progress Starlink has made in terms of getting all the economies of scale in place,” Farrar suggested. He noted that Starlink now has around 10 million customers globally, who are all largely using the fixed broadband service, and is producing around 200,000 terminals a week. The analyst said he thought it would  be difficult for Amazon Leo to catch up.

Starlink already has over 9,200 operational satellites up, enabling it to provide fixed broadband satellite services across the world. There are also reportedly around 7,180 Starlink Gen2 satellites operational — the 2nd generation satellites can support D2D satellite to cellphone communications, as well as facilitating the 3GPP Non-Terrestrial Networks (NTN) specification, so that 5G phones can get a D2D connection. 

“It’s a long road for Amazon to try and catch up,” Farrar commented.

Amazon Leo has signed some important airline deals that will deliver in-flight internet connectivity to customers on planes. JetBlue became the first airline to sign on for the Amazon satellite service in September 2025.

JetBlue expects to begin rolling out the system in 2027 with a phased rollout to follow, the airline said in a statement.

At the end of March, Delta signed a similar deal with Amazon Leo, where the airline said it would install the equipment on an initial 500 aircraft in its fleet in 2028. This comes after Delta rejected an airline Internet deal using Starlink satellite connectivity, choosing Amazon instead.

Still, Starlink is continuing to sign on its own airline customers, signing on American Airlines to use its system in late May 2026. American said it will install the Starlink satellite equipment “on more than 500 narrow-body aircraft beginning in Q1 2027.”

Amazon hasn’t lost the satellite race … yet.

“Amazon is putting enormous resources into this and the significant investment to buy Globalstar that adds to that scale,” the analyst noted. Amazon Leo announced it would acquire Globalstar for around $12 billion on April 14. 

The deal will enable Amazon to add to its LEO satellite network. Amazon also signed a deal with Apple to take over the Globalstar’s D2D services for Apple iPhones and watches.