Sixty Deals, Then a Seat

Viasat's board did not simply add two directors on May 6, 2026. It put both of them straight onto the committee running the company's strategic review, and disclosed in the same filing a cooperation agreement with Carronade Capital Management carrying customary standstill, voting and other provisions. One of the two was Shekar Ayyar, chairman and chief executive of Arrcus, Inc., a private networking software company with almost nothing to do with satellites.

The board went to ten directors, eight of them independent. Ayyar took a Class II seat with an initial term running to the 2028 annual meeting.

What a board is buying when it seats someone like this is not satellite engineering. It is transaction experience, and his is unusually dense.

More Than Sixty Transactions

Before Arrcus, Ayyar ran strategy and corporate development at VMware across enterprise software and communications, and later ran the company's Telco and Edge Cloud business as executive vice president and general manager. Arrcus, announcing his hire, put a number on the first of those jobs: more than 60 M&A transactions and investments, among them Nicira, AirWatch and VeloCloud - three of the deals that defined what VMware became in networking and mobility.

The second credential is the one Viasat's filing named outright. Ayyar sat on the board of Altair Engineering and stayed through its sale. Siemens completed that acquisition on March 26, 2025, at an enterprise value of approximately $10 billion.

A satellite operator running a strategic review, a shareholder that had been in dialogue with the board for a year and has now signed a standstill, and a new director who has been inside a ten-billion-dollar sale. The logic is not subtle.

The Thesis He Arrived With

Arrcus announced Ayyar as chief executive and chairman on September 15, 2021, alongside fresh capital from Liberty Global, SoftBank Corp and Samsung Next, who joined Clear Ventures, General Catalyst Partners and Lightspeed Venture Partners on the register. His statement that day was precise about which market he had come for.

"I am delighted to be joining Arrcus as CEO at this inflection point in the industry when customers are seeking edge-native, large-scale distributed and disaggregated networking to support their 5G and edge deployments," he said.

5G. Edge. That was the wager in 2021, and it was the consensus wager: the telecom operators were about to rebuild their networks, and somebody was going to sell them the software.

The Thesis That Replaced It

Every Arrcus announcement of 2026 leads with a different word. The word is inference.

On March 3, Arrcus announced a collaboration with Fujitsu built around its MONAKA processor. "AI inference is increasingly happening closer to where data is generated, in factories, hospitals, warehouses, and regional networks, and that fundamentally changes infrastructure requirements," Ayyar said. The same day brought a partnership with Lightstorm across Asia-Pacific, which he said "enables AI operators across Asia-Pacific to optimize network performance and improve infrastructure efficiency."

Thirteen days later, at NVIDIA's GTC, Arrcus announced that its Inference Network Fabric would integrate with NVIDIA's Dynamo framework, BlueField-3 DPUs and Spectrum-X Ethernet. "AI is entering its inference era, where networking becomes the control plane for performance and economics," Ayyar said.

Then TELUS, on June 19: "The era of centralized AI is giving way to distributed intelligence, and the network is becoming the control plane for performance, sovereignty and economics." Then UfiSpace, on July 24, tying the ArcOS operating system to Broadcom Tomahawk switch platforms for AMD-powered clusters: "AI infrastructure needs to be complemented by smart, programmable networking."

Five releases in five months, one argument repeated without variation: the network is where the economics of AI get decided.

The Pivot Has No Numbers

The complication sits in the comparison. The build-out Ayyar was hired in 2021 to catch is not the build-out paying the bills in 2026. Arrcus still sells the same core product - ArcOS, a disaggregated network operating system - but the buyer, and the reason given for buying, have changed completely. In 2021 the pitch was total cost of ownership for operators modernising to 5G. In 2026 it is GPU utilisation and where inference traffic is permitted to sit.

That is a pivot rather than a plan executed, and there is no public way to score it. Arrcus is privately held and publishes no revenue. The only external evidence of how the second thesis is landing is the partner list: NVIDIA, Fujitsu and 1Finity, AMD platforms through UfiSpace, TELUS in Canada, Lightstorm across Asia. Announcements, not numbers.

Sovereignty Is The Selling Point

Three of those five deals turn on the same idea, and it is a political one before it is a technical one: that a network should be able to enforce where AI inference is allowed to happen. TELUS's stated interest is public safety and government work inside Canada. Fujitsu's is sovereign infrastructure in Japan. Lightstorm's is data-sovereignty compliance across Asia-Pacific.

Ayyar's TELUS statement makes the frame explicit. Sovereignty appears in his list beside performance and economics - not as a compliance cost but as a thing the fabric sells. It is a defensible reading of where the market is heading. It is also, conveniently, an argument for buying programmable networking software instead of a proprietary fabric.

The Company He Now Reviews

Viasat has no time to spare. In its first quarter of fiscal 2027, reported on August 4, it posted a net loss of $52 million - an improvement on the $56 million a year earlier, driven mostly by lower interest expense - on revenue down 1 percent and adjusted EBITDA down 7 percent. What is growing is the defence side of the house: total backlog up 19 percent, Defense and Advanced Technology backlog up 32 percent, awards in that segment up 22 percent.

A shrinking top line, a swelling defence book, a strategic review already under way, and a director seated in May who has done this kind of arithmetic before.

Ayyar's argument at Arrcus is that the network becomes the control plane for economics. At Viasat he is now on the board that has to work out what the underlying businesses are worth, separately or together. His term runs to 2028. The review has no published deadline.