Chapter 4: The Enterprise Decision

May through June 1995 — Novell headquarters, Provo; Compaq headquarters, Houston; joint engineering workshop

The boardroom had two futures in it before anyone said a word. The UnixWare transfer folder sat at Gordon Maddix’s elbow, closed, representing a transfer that had been stayed for the last month. Beside it, Sarah set down a thinner packet — the source audit, the license, the governance charter, the staged engineering plan, the conditional statement from Karen Whitlock and David Rosen, Avery’s letter, carrying his engineering director’s signature rather than merely Avery’s private assurances.

The room wasn’t arranged for a presentation. It was arranged for a vote. Every director’s copy of her packet already had tabs in it, and handwritten questions in margins she hadn’t written herself. Sarah had spent four months learning to read a room like this one, and what she read now was a board that had already done its homework and arrived wanting confirmation, not persuasion — which was its own kind of danger, since confirmation could run in either direction just as easily.

Helen Vance chaired the strategy committee, and she opened by taking away the only argument Sarah hadn’t had to make in weeks. “We’re not deciding whether Microsoft’s position is dangerous, Ms. Caldwell. Everyone in this room already believes that. We’re deciding which risk Novell is willing to own.” She laid out the three real choices without editorializing any of them: complete the UnixWare transfer and recover something now; keep the product proprietary and keep funding it at a level that hadn’t changed its trajectory in two years; or publish a legally cleared foundation, keep the enterprise services commercial, and fund the work required to make the combination real.

A director from finance took the third option apart first. Publish the foundation and Novell destroyed an asset before it had proven a single dollar of replacement revenue. Once it was out, there was no selling exclusivity back.

“He’s right about the mechanics,” Denise said, before Sarah could soften it for her. “The initial grant can’t be withdrawn. A future management team can stop contributing to it. It cannot reclaim what’s already released, and it cannot stop a competitor from using the same code.”

Sarah didn’t try to argue the premise. “The asset is already losing value while it’s closed — Mr. Maddix has seen the same purchase histories I have. This isn’t a plan to monetize the operating system. It’s a plan to use the operating system to expand what we sell around it. NDS, NCP, GroupWise, management, clustering, certification, support — every one of those grows if this foundation exists in more places than Novell can reach alone.”

Maddix let her finish without stepping in front of her. It was the first time in four months nobody in the room was protecting her from a question. She’d noticed, walking in, that he’d taken a seat at the side of the table rather than at its head — not distance, exactly, but a small, deliberate refusal to let anyone read his face before they’d read her answer.

The questions came fast after that, and each one wanted an operational answer, not a conviction. What stopped a competitor from taking the published foundation and simply outselling Novell with it? Nothing in the license — Novell would have to win that on integration, trusted releases, certification, and support, the same way it would have had to win it eventually against Microsoft. What happened if the BSD contacts walked away entirely? Their participation was useful, not load-bearing; the compatibility work stood on its own. What happened if Compaq walked away? The program stayed nonexclusive by design — Compaq accelerated things, it didn’t gate them, and it would get no veto for showing up first. Did any of this abandon the NetWare customers already running Novell’s network? No — NCP and NDS stayed intact, mixed-environment administration stayed a requirement, and nobody’s working servers got replaced on Novell’s schedule instead of the customer’s own.

A director near the end of the table, who hadn’t spoken yet, asked the question Sarah suspected mattered most to him personally: what happened to the people currently earning a living selling and supporting UnixWare as it existed today, once the strategy shipped. She told him the truth rather than the reassurance — some of that work moved onto the new foundation’s support and certification side, some of it didn’t survive the transition intact, and she wasn’t going to promise him a headcount number she couldn’t back up in this room.

“How long,” another director asked, “before there’s something real to ship?”

Eric answered that one himself. A narrow, bootable reference source first — cleared, honest about what it didn’t yet include. Restricted utilities and drivers replaced in waves after that. Novell’s own services ported in parallel, not bolted on afterward as an afterthought.

“And how will this board know, twelve months from now, whether the experiment is working or simply expensive?” Vance asked.